Self‐management interventions for people with chronic obstructive pulmonary disease

Abstract

Background

Self‐management interventions help people with chronic obstructive pulmonary disease (COPD) to acquire and practise the skills they need to carry out disease‐specific medical regimens, guide changes in health behaviour and provide emotional support to enable them to control their disease. Since the 2014 update of this review, several studies have been published.

Objectives

Primary objectives

To evaluate the effectiveness of COPD self‐management interventions compared to usual care in terms of health‐related quality of life (HRQoL) and respiratory‐related hospital admissions.
To evaluate the safety of COPD self‐management interventions compared to usual care in terms of respiratory‐related mortality and all‐cause mortality.

Secondary objectives

To evaluate the effectiveness of COPD self‐management interventions compared to usual care in terms of other health outcomes and healthcare utilisation.
To evaluate effective characteristics of COPD self‐management interventions.

Search methods

We searched the Cochrane Airways Trials Register, CENTRAL, MEDLINE, EMBASE, trials registries and the reference lists of included studies up until January 2020.

Selection criteria

Randomised controlled trials (RCTs) and cluster‐randomised trials (CRTs) published since 1995. To be eligible for inclusion, self‐management interventions had to include at least two intervention components and include an iterative process between participant and healthcare provider(s) in which goals were formulated and feedback was given on self‐management actions by the participant.

Data collection and analysis

Two review authors independently selected studies for inclusion, assessed trial quality and extracted data. We resolved disagreements by reaching consensus or by involving a third review author. We contacted study authors to obtain additional information and missing outcome data where possible. Primary outcomes were health‐related quality of life (HRQoL), number of respiratory‐related hospital admissions, respiratory‐related mortality, and all‐cause mortality. When appropriate, we pooled study results using random‐effects modelling meta‐analyses.

Main results

We included 27 studies involving 6008 participants with COPD. The follow‐up time ranged from two‐and‐a‐half to 24 months and the content of the interventions was diverse. Participants' mean age ranged from 57 to 74 years, and the proportion of male participants ranged from 33% to 98%. The post‐bronchodilator forced expiratory volume in one second (FEV1) to forced vital capacity (FVC) ratio of participants ranged from 33.6% to 57.0%. The FEV1/FVC ratio is a measure used to diagnose COPD and to determine the severity of the disease. Studies were conducted on four different continents (Europe (n = 15), North America (n = 8), Asia (n = 1), and Oceania (n = 4); with one study conducted in both Europe and Oceania).

Self‐management interventions likely improve HRQoL, as measured by the St. George’s Respiratory Questionnaire (SGRQ) total score (lower score represents better HRQoL) with a mean difference (MD) from usual care of ‐2.86 points (95% confidence interval (CI) ‐4.87 to ‐0.85; 14 studies, 2778 participants; low‐quality evidence). The pooled MD of ‐2.86 did not reach the SGRQ minimal clinically important difference (MCID) of four points. Self‐management intervention participants were also at a slightly lower risk for at least one respiratory‐related hospital admission (odds ratio (OR) 0.75, 95% CI 0.57 to 0.98; 15 studies, 3263 participants; very low‐quality evidence). The number needed to treat to prevent one respiratory‐related hospital admission over a mean of 9.75 months' follow‐up was 15 (95% CI 8 to 399) for participants with high baseline risk and 26 (95% CI 15 to 677) for participants with low baseline risk. No differences were observed in respiratory‐related mortality (risk difference (RD) 0.01, 95% CI ‐0.02 to 0.04; 8 studies, 1572 participants ; low‐quality evidence) and all‐cause mortality (RD ‐0.01, 95% CI ‐0.03 to 0.01; 24 studies, 5719 participants; low‐quality evidence).

We graded the evidence to be of ‘moderate’ to ‘very low’ quality according to GRADE. All studies had a substantial risk of bias, because of lack of blinding of participants and personnel to the interventions, which is inherently impossible in a self‐management intervention. In addition, risk of bias was noticeably increased because of insufficient information regarding a) non‐protocol interventions, and b) analyses to estimate the effect of adhering to interventions. Consequently, the highest GRADE evidence score that could be obtained by studies was ‘moderate’.

Authors' conclusions

Self‐management interventions for people with COPD are associated with improvements in HRQoL, as measured with the SGRQ, and a lower probability of respiratory‐related hospital admissions. No excess respiratory‐related and all‐cause mortality risks were observed, which strengthens the view that COPD self‐management interventions are unlikely to cause harm. By using stricter inclusion criteria, we decreased heterogeneity in studies, but also reduced the number of included studies and therefore our capacity to conduct subgroup analyses. Data were therefore still insufficient to reach clear conclusions about effective (intervention) characteristics of COPD self‐management interventions. As tailoring of COPD self‐management interventions to individuals is desirable, heterogeneity is and will likely remain present in self‐management interventions.

For future studies, we would urge using only COPD self‐management interventions that include iterative interactions between participants and healthcare professionals who are competent using behavioural change techniques (BCTs) to elicit participants' motivation, confidence and competence to positively adapt their health behaviour(s) and develop skills to better manage their disease. In addition, to inform further subgroup and meta‐regression analyses and to provide stronger conclusions regarding effective COPD self‐management interventions, there is a need for more homogeneity in outcome measures. More attention should be paid to behavioural outcome measures and to providing more detailed, uniform and transparently reported data on self‐management intervention components and BCTs. Assessment of outcomes over the long term is also recommended to capture changes in people's behaviour. Finally, information regarding non‐protocol interventions as well as analyses to estimate the effect of adhering to interventions should be included to increase the quality of evidence.

Plain language summary

Self‐management for people with chronic obstructive pulmonary disease

Review question
We looked at the current evidence on the effects of self‐management interventions for people with chronic obstructive pulmonary disease (COPD). In particular, we assessed their effectiveness on health‐related quality of life (HRQoL) and hospital admissions related to COPD. We also wanted to assess whether self‐management interventions are safe by evaluating the number of deaths.

Background
COPD is a common and long‐term lung condition that slowly worsens over the years, and causes symptoms such as breathlessness, coughing, wheezing and increased sputum (mucus) production. This leads to loss of well‐being (also known as reduction in HRQoL) in people with COPD. Self‐management interventions encourage people to develop the skills and behaviours they need to successfully manage their disease, and the emotional and practical issues that may go along with it. In this update, we reviewed the current evidence on the effects of self‐management on HRQoL, hospital admissions related to COPD, deaths from any cause and related to COPD, as well as other health outcomes.

Search date
We searched for studies up until January 2020.

Study characteristics
We included 27 studies, involving 6008 participants, that evaluated the effectiveness and safety of COPD self‐management interventions. The average age of the participants ranged between 57 and 74 years. Between 33% to 98% of the participants in the studies were male. Studies were conducted on four different continents (15 in Europe, eight in North America, one in Asia, and four in Oceania; with one study conducted in both Europe and Oceania). All studies had control groups of participants who received usual care – that is, care typical for people with COPD. The studies lasted between two‐and‐a‐half to 24 months.

Key results
Self‐management interventions improved HRQoL in people with COPD compared to usual care, but this did not reach a clinically meaningful improvement. The number of participants with at least one hospital admission related to COPD was reduced amongst those who participated in a self‐management intervention. We found no difference in number of deaths between self‐management and usual care groups, which strengthens the view that COPD self‐management interventions are unlikely to cause harm. We have been strict about only including studies that met our definition of a COPD self‐management intervention. Despite this, the studies were still quite different from one another in terms of the intervention components used, duration of the self‐management intervention and the study populations. It should be noted, that heterogeneity in future interventions will be inevitable as individual tailoring of self‐management interventions is desirable; it will never be a 'one size fits all' intervention.

Quality of the evidence
Our confidence in the evidence for the main findings in this review ranged from ‘very low’ to ‘moderate’, due to the nature of the COPD self‐management intervention – none of the studies prevented participants and personnel from knowing what treatment the participants were getting. Additionally, none of the studies provided detailed information about the extent to which participants adhere to the self‐management intervention or whether any further treatments were given during the course of the study. Consequently, study evidence could not be graded higher than ‘moderate’ in any of the studies.

Summary of findings

Summary of findings 1. Self‐management interventions compared to usual care for people with chronic obstructive pulmonary disease.

Self‐management interventions compared to usual care for people with chronic obstructive pulmonary disease
Patients or population: people with chronic obstructive pulmonary disease (COPD) Settings: hospital, outpatient clinic, primary care, home‐based Intervention: COPD self‐management interventions Comparison: usual care
Outcomes	Anticipated absolute effects* (95% CI)			Relative effect (95% CI)	Number of participants (studies)	Quality of the evidence (GRADE)	Comments
	Risk with usual care	Risk with self‐management interventions	Difference
HRQoL Assessed with: St. George’s Respiratory Questionnaire adjusted total score Scale from: 0 to 100 Note: lower scores indicate better HRQoL Follow‐up range: 3 to 12 months	The mean HRQoL ranged from 30.9 to 71.1 points	‐	2.86 points lower (4.87 lower to 0.85 lower)	‐	2778 (15 comparisons of studies)	⊕⊕⊝⊝ Lowa	‐
Respiratory‐related hospital admissions Assessed with: number of participants with at least one respiratory‐related hospital admission Follow‐up range: 3 to 12 months	317 per 1000	258 per 1000 (209 to 312)	‐	OR 0.75 (0.57 to 0.98)	3263 (16 comparisons of studies)	⊕⊝⊝⊝ Very lowb	‐
Respiratory‐related mortality Assessed with: number of respiratory‐related deaths Follow‐up range: 3 to 24 months	4.2%#	7.0% (3.8 to 12.5)#	2.7% more participants (0.4 fewer to 8.3 more)#	OR 1.70 (0.89 to 3.26)#	1572 (8 comparisons of studies)	⊕⊕⊝⊝ Lowc	Pooled risk difference of 0.01 (95% CI ‐0.02 to 0.04)#
All‐cause mortality Assessed with: number of all‐cause deaths Follow‐up range: 3 to 24 months	8.4%#	7.3% (5.1 to 10.4)#	1.1% fewer participants (3.3 fewer to 2.0 more)#	OR 0.86 (0.59 to 1.26)#	5719 (25 comparisons of studies)	⊕⊕⊝⊝ Lowd	Pooled risk difference of ‐0.01 (95% CI ‐0.03 to 0.01)#
All‐cause hospital admissions Assessed with: number of participants with at least one all‐cause hospital admission Follow‐up range: 3 to 12 months	397 per 1000	367 per 1000 (318 to 415)	‐	OR 0.88 (0.71 to 1.08)	2633 (11 comparisons of studies)	⊕⊕⊕⊝ Moderatee	‐
Health status ‐ Dyspnoea Assessed with mMRC Dyspnoea Scale total score Scale from: 0 to 4 Note: lower scores indicate less dyspnoea Follow‐up range: 3 to 12 months	The mean dyspnoea score ranged from 2.1 to 3.1	‐	0.31 lower (1.23 lower to 0.6 higher)	‐	356 (3 comparisons of studies)	⊕⊕⊝⊝ Lowf	‐
ED visits Assessed with: mean number of visits Follow‐up range: 12 to 24 months	The mean number of ED visits ranged from 0.7 to 3.1	‐	0.52 lower (0.89 lower to 0.15 lower)	‐	1939 (6 comparisons of studies)	⊕⊕⊝⊝ Lowg	‐
Health status ‐ Anxiety and depression Assessed with HADS total score Scale from: 0 to 21 Note: higher scores indicate more active symptoms of anxiety and depression Follow‐up range: 3 to 24 months	The mean anxiety score ranged from 4.7 to 10.2 The mean depression score ranged from 3.8 to 9.1	‐	Anxiety: 0.57 lower (1.01 lower to 0.13 lower) Depression: 0.45 lower (0.80 lower to 0.10 lower)	‐	Anxiety: 1647 Depression: 1653 (9 comparisons of studies)	⊕⊕⊕⊝ Moderateh	‐
COPD exacerbations Assessed with: number of COPD exacerbations per participant (regardless of definition) Follow‐up range: 12 to 24 months	The mean number of COPD exacerbations ranged from 1.2 to 2.8	‐	0.06 lower (0.26 lower to 0.15 higher)	‐	1401 (7 comparisons of studies)	⊕⊕⊕⊝ Moderatei	‐
*The risk in the intervention group (and its 95% CI) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). #The absolute and relative effects do not include comparisons of studies that reported zero events (respiratory‐related mortality: two studies with no deaths and thus excluded; all‐cause mortality: three studies with no deaths and thus excluded). The reported effects are in this case overestimated and should be interpreted with caution. As a result, the pooled risk difference that includes all study data is more accurate. CI: confidence interval; COPD: chronic obstructive pulmonary disease; ED: emergency department; GRADE: Grading of Recommendation, Assessment, Development and Evaluation; HADS: Hospital Anxiety and Depression Scale; HRQoL: health‐related quality of life; mMRC: modified Medical Research Council; OR: odds ratio.
GRADE Working Group grades of evidence High quality: we are very confident that the true effect lies close to that of the estimate of the effect. Moderate quality: we are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low quality: our confidence in the effect is limited: the true effect may be substantially different from the estimate of the effect. Very low quality: we have very little confidence in the effect estimate: the true effect is likely to be substantially different from the estimate of the effect.

^aThe majority of the studies had high risk of bias. Heterogeneity was moderate (I² = 60%) (risk of bias ‐1, inconsistency ‐1).
^bThe majority of the studies had high risk of bias. Heterogeneity was moderate (I² = 49%). The 95% CI was wide (risk of bias ‐1, inconsistency ‐1, imprecision ‐1).
^cThe majority of the studies had high risk of bias. Heterogeneity was substantial (I² = 63%) (risk of bias ‐1, inconsistency ‐1).
^dThe majority of the studies had high risk of bias. Heterogeneity was substantial (I² = 63%) (risk of bias ‐1, inconsistency ‐1).
^eThe majority of the studies had high risk of bias (risk of bias ‐1).
^fThe majority of the studies had high risk of bias. Only three studies were included in this meta‐analysis (risk of bias ‐1, imprecision ‐1).
^gThe majority of the studies had high risk of bias. Heterogeneity was considerable (I² = 96%) (risk of bias ‐1, inconsistency ‐1).
^hThe majority of the studies had high risk of bias (risk of bias ‐1).
ⁱThe majority of the studies had high risk of bias (risk of bias ‐1).

Background

Description of the condition

Chronic obstructive pulmonary disease (COPD) is a chronic progressive lung condition characterised by exacerbations — acute distressing of symptoms, such as increased dyspnoea, cough and wheeze, and increased and altered sputum production (Anthonisen 1987; Rodriguez‐Roisin 2000) — that cause impaired health‐related quality of life (HRQoL), increased hospitalisations and mortality (GOLD 2021). COPD is both preventable and, although not fully reversible, treatable (GOLD 2021). In 2019, COPD ranked third in the leading causes of death globally (WHO 2020). It is predicted that by 2060, there will be over 5.4 million deaths annually from COPD and related conditions (WHO 2018). Apart from personal distress, COPD confers a substantial and increasing economic and social burden on society (GOLD 2021), with its exacerbations accounting for most direct costs (Toy 2010). The high and growing prevalence of COPD makes it a major problem of chronic morbidity and mortality in health care worldwide.

Description of the intervention

Management of COPD is complex and can be difficult for people with COPD due to heterogeneous disease progression, high symptom burden and fluctuation of symptoms (Agusti 2010; Donaldson 2005; Kessler 2011). Self‐management interventions have been described as an essential part of COPD disease management. They aim to help people develop skills to manage the disease more effectively, and have the goal to empower the individual during all stages of the disease (Bourbeau 2009; Effing 2012). This is important for people with chronic disease, such as COPD, who are responsible for their day‐to‐day care over the duration of the illness (Lorig 2003).

Lorig and Holman were amongst the first to suggest that a successful self‐management intervention should include the following essential skills and attributes: problem‐solving, decision‐making, resource utilisation, the formation of a partnership between participant and healthcare professional, action‐planning and self‐tailoring (Lorig 2003). Skills mastery, modelling, interpretation of symptoms and social persuasion are believed to contribute to enhanced self‐efficacy in people with COPD (Lorig 2003). Self‐efficacy is defined as having the confidence to effectively manage one's health, and has been recognised as a powerful factor in inducing new health behaviours in individuals, such as smoking cessation, regular exercise or physical activity, diet habits and coping with breathlessness (Bourbeau 2004; Effing 2012). The debate on the definition and most effective content of self‐management was ongoing during the previous update of this review in 2014 (Zwerink 2014).

In 2016, an international expert group reached consensus regarding a definition of a COPD self‐management intervention (Effing 2016):
"A COPD self‐management intervention is structured but personalised and often multi‐component, with goals of motivating, engaging, and supporting the patients to positively adapt their health behaviour(s) and develop skills to better manage their disease.

The ultimate goals of self‐management are: a) optimising and preserving physical health; b) reducing symptoms and functional impairments in daily life and increasing emotional well‐being, social well‐being, and quality of life; and c) establishing effective alliances with healthcare professionals, family, friends and community.

The process requires iterative interactions between patients and healthcare professionals who are competent in delivering self‐management interventions. These patient‐centred interactions focus on: 1) identifying needs, health beliefs, and enhancing intrinsic motivations; 2) eliciting personalised goals; 3) formulating appropriate strategies (e.g. exacerbation management) to achieve these goals; and if required 4) evaluating and re‐adjusting strategies. Behaviour change techniques are used to elicit patient motivation, confidence and competence. Literacy sensitive approaches are used to enhance comprehensibility.”

We developed our review inclusion criteria in line with the definition above.

Different frameworks have been developed to characterise the underlying mechanisms of changing the behaviour of an individual (Michie 2011). Behavioural change techniques (BCTs) are defined as “an observable, replicable, and irreducible component of an intervention designed to alter or redirect causal processes that regulate behaviour” (Michie 2013). These techniques are proposed to be an ‘active ingredient’ (e.g. feedback, self‐monitoring and reinforcement) and can be used alone or in combination, and in a variety of intervention forms (e.g. face‐to‐face, written or digital) (Michie 2013). BCTs are perceived as imperative to elicit motivation, confidence and competence of participants in COPD self‐management interventions (Effing 2016). Previous COPD self‐management intervention studies conclude that participant activation and long‐term behaviour change are crucial characteristics to achieve improvement of health status (Benzo 2012; Disler 2012; Effing 2012; Nici 2012; Nici 2014).

How the intervention might work

Self‐management interventions are directed towards behavioural change and include a variety of components, such as self‐treatment of exacerbations, symptom management, smoking cessation, physical activity and dietary intake. Due to a significant heterogeneity of content within self‐management interventions, the reported effects are diverse. This heterogeneity complicates the formulation of clear conclusions regarding effective intervention components and implementation in clinical practice. However, the current vision to personalise treatment based on participant characteristics will lead to more participant‐tailored treatment approaches, and heterogeneity will therefore in the future also be inevitable (Agusti 2014; Singh 2017; Trappenburg 2013). To be successful, a self‐management intervention has to lead to positive behaviour change in the individual behaviours targeted by the intervention (Bourbeau 2015). Primary health behaviour targets for COPD self‐management interventions are: adequate medication intake (e.g. adherence, inhalation technique), smoking cessation, increasing levels of physical activity and exercise, managing breathlessness, using energy conservation techniques, avoiding aggravating factors (e.g. smoke, pollution), and using stress management strategies (Bourbeau 2015). It is important to note that even when people with COPD are aware of the benefits of self‐management, this does not mean they will be motivated to positively adapt their health behaviour (Bourbeau 2015). However, if individuals are able to perceive the risks associated with a given condition, they may be more likely to seek health‐improving behaviours in order to prevent the condition or reduce its progression (Hayden 2009; Rosenstock 1974).

COPD self‐management interventions are associated with a reduced number of exacerbations days and hospitalisations and decreased healthcare costs, as well as improved HRQoL (Effing 2009; Lenferink 2017; Zwerink 2014). A 2017 Cochrane Review evaluated the effects of COPD self‐management interventions, including action plans for COPD exacerbations, compared with usual care (Lenferink 2017). In line with other COPD self‐management reviews, it concluded that COPD exacerbation action plans are associated with improvements in HRQoL and a lower probability of respiratory‐related hospital admissions (Lenferink 2017). Although no excessive all‐cause mortality risk was observed, results showed a small but significantly higher respiratory‐related mortality rate for self‐management (including an action plan for COPD exacerbations) compared to usual care (Lenferink 2017). Another systematic review published in 2017 found that COPD self‐management interventions generally improved HRQoL and, in addition, reduced emergency visits (Newham 2017). Furthermore, Newham and colleagues found that BCTs addressing mental health showed increased improvements in those outcomes (Newham 2017). Jonkman 2016 aimed to identify components of self‐management interventions for people with chronic conditions (chronic heart failure, COPD, type 2 diabetes mellitus) that affect improvements in HRQoL. They concluded that the duration of the intervention involving ongoing healthcare professional support showed positive associations with all‐cause hospital admissions. This conclusion reminds us that self‐management is not a time‐limited intervention, but an ongoing process of reviewing, problem‐solving, and collaboration between the healthcare professional and chronically ill person, which needs a whole systems approach for effective implementation (Jonkman 2016).

Why it is important to do this review

The original Cochrane Review regarding COPD self‐management interventions was published in 2003 (Monninkhof 2002; Monninkhof 2003). The first update of the review, published in 2007, concluded that self‐management interventions were associated with improved HRQoL and reduced hospital admissions with no indication of detrimental effects on the other health outcomes (Effing 2007). The second update of the review, published in 2014, strengthened the evidence for associations between the intervention and improved quality of life, reduced respiratory‐related hospitalisations and improved dyspnoea (Zwerink 2014). In addition, this update concluded that self‐management interventions were associated with reduced all‐cause hospitalisations (Zwerink 2014). However, because of heterogeneity amongst interventions, study populations, follow‐up time and outcome measures, it was not possible to formulate clear conclusions regarding effective components and characteristics of self‐management interventions (Zwerink 2014). The latest update of the review included studies until August 2011. Since then, multiple studies have been published and new opinions have been formed regarding the limitations and contents of self‐management interventions for people with COPD.

Previous systematic reviews regarding the effectiveness of COPD self‐management interventions recommended that further research should focus on: 1) identifying effective components of interventions and identifying participant‐specific factors that may modify these; and 2) characterisation of behavioural change theories and strategies that underpin COPD self‐management interventions (Jolly 2016; Jonkman 2016; Lenferink 2017; Newham 2017; Zwerink 2014). Therefore, in the current review, we intended to assess not only the effectiveness and safety of COPD self‐management interventions, but also tried to identify effective self‐management intervention characteristics (e.g. integration of various self‐management intervention components and behavioural change techniques).

Objectives

Primary objectives

• To evaluate the effectiveness of COPD self‐management interventions compared to usual care in terms of HRQoL and respiratory‐related hospital admissions.

• To evaluate the safety of COPD self‐management interventions compared to usual care in terms of respiratory‐related mortality and all‐cause mortality.

Secondary objectives

• To evaluate the effectiveness of COPD self‐management interventions compared to usual care in terms of other health outcomes and healthcare utilisation.

• To evaluate effective characteristics of COPD self‐management interventions.

Methods

Criteria for considering studies for this review

Types of studies

We included randomised controlled trials (RCTs) and cluster‐randomised trials (CRTs) assessing the effectiveness of self‐management interventions for people with COPD. For CRTs, we performed meta‐analyses only if they had been adjusted to account for clustering (or could be adjusted by ourselves). In line with the previous update, we excluded studies published before 1995, as we believe that the primary focus of self‐management interventions before 1995 consisted of improving knowledge through education rather than initiating and enabling sustained behavioural change (Zwerink 2014).

Types of participants

All included participants were required to have a diagnosis of COPD according to the GOLD classification criteria (that is, a post‐bronchodilator forced expiratory volume in one second (FEV1)/forced vital capacity (FVC) ratio < 0.70)(GOLD 2021), a measure used to diagnose COPD and to determine the severity of the disease, as shown by baseline FEV1/FVC ratio spirometry, or in case of uncertainty, confirmed by study authors. Therefore, inclusion using only e.g. International Classification of Diseases (ICD) codes (WHO 2019) was insufficient. We excluded participants with a primary diagnosis of asthma.

Types of interventions

To be included, self‐management interventions had to be defined as structured interventions for participants with COPD aimed at improvement of self‐health behaviours and self‐management skills using an iterative process in at least two of its intervention components (i.e. smoking cessation, self‐recognition of exacerbations, use of an exacerbation action plan, home‐based exercise or physical activity, diet, medication intake (e.g. adherence, inhalation technique), or coping with breathlessness). An iterative process was defined as an interaction between participants and healthcare professional(s), including at least two contact moments, in which goals were formulated and feedback was given to develop participants’ self‐management skills. Interventions needed to include techniques directed at achieving behavioural change. We included interventions only if they incorporated at least the two following BCT clusters, defined according to Michie 2013: ‘goals and planning’ and ‘feedback and monitoring’.

We excluded interventions classified as pulmonary rehabilitation or exercise classes offered in a hospital, at a rehabilitation centre or in a community‐based setting. We included interventions that incorporated unsupervised home‐based exercise programmes if they met all the other study criteria.

We included only studies with usual care as the comparator, defined as de facto routine clinical care.

Types of outcome measures

Primary outcomes

• Health‐related quality of life (HRQoL)

• Respiratory‐related hospital admissions

• Respiratory‐related mortality

• All‐cause mortality

Secondary outcomes

• All‐cause hospital admissions

• Use of (other) healthcare facilities (e.g. number of emergency department visits, number of all‐cause and respiratory‐related hospital admission days in total and per participant, general practitioner, number of nurse and specialist visits)

• Number of COPD exacerbations, based on:

  • COPD symptom scores (e.g. symptom diary)

  • Courses of oral corticosteroids or antibiotics, or both

• Health status (e.g. dyspnoea, impact of COPD on life, anxiety and/or depression)

• Self‐efficacy

• Days lost from work

• Exercise capacity and physical activity

• Self‐management behaviour

• Patient activation

• Health literacy

Search methods for identification of studies

Electronic searches

The previously published version of this review included searches up to August 2011 (Zwerink 2014). We re‐assessed all previously included studies for inclusion in this update. The search period for this update is 2011 to January 2020. Studies were identified from searches of the following databases and trials registries.

• Cochrane Airways Register, through the Cochrane Register of Studies (CRS).

• Cochrane Central Register of Controlled Trials (CENTRAL), through the CRS.

• MEDLINE (Ovid) ALL.

• EMBASE (Ovid).

• ClinicalTrials.gov (www.ClinicalTrials.gov).

• World Health Organization International Clinical Trials Registry Platform (ICTRP).

We searched all sources from 2011 up to 23 January 2020, with no restrictions on language or publication type. See Appendix 1 for details of the search strategies. We performed an updated database search from January 2020 to March 2021. We added potentially eligible studies from this search to 'Studies awaiting classification', and we will incorporate these into the review at the next update, if inclusion criteria are met.

Searching other resources

We checked reference lists of all primary studies, reviewed articles for additional references, and re‐evaluated the included studies from the previous version of this review against the updated inclusion and exclusion criteria.

Data collection and analysis

Selection of studies

Because of the large number of studies found, we used Cochrane's 'Screen4Me' workflow to help assess the results of our search for RCTs. Screen4Me includes three components: 1) known assessments: a service that matches records in the search results to records that have already been screened in Cochrane Crowd and have been labelled as 'RCT' or as 'not an RCT'; 2) the RCT classifier: a machine‐learning model that distinguishes RCTs from non‐RCTs; and if appropriate, 3) Cochrane Crowd: Cochrane's citizen science platform where 'the crowd' helps to identify and describe health evidence.

Following use of the Screen4Me workflow, any two of the team of review authors (JS, TE, AL, MB, JP, MZ or PV) independently assessed titles and abstracts of all references retrieved using Covidence software (Covidence 2016). Subsequently, two review authors (of JS, TE, AL, MB, JP or MZ) independently reviewed full‐text versions of potentially relevant reports to determine eligibility for inclusion based on the criteria stated above, using Covidence.

At the start of screening, we conducted calibration exercises to enhance the validity of the screening process. Therefore, all review authors independently assessed 50 titles and abstracts, and 10 full‐text articles. Subsequently, we compared screening results and discussed differences between review authors' judgements. We then updated a detailed worksheet to clarify the decision process. During the subsequent review process, any disagreements were resolved through discussion between the two review authors concerned. If consensus could not be reached, we consulted a third review author. Where necessary, we contacted authors of potentially eligible studies to ask for further information regarding inclusion criteria. Detailed information regarding this process can be found in the PRISMA flow diagram (Figure 1) and ‘Excluded studies’ section (Moher 2009).

1.

*FEV1/FVC ratio <0.7

**The previous version of this review included both RCTs and non‐RCTs

^#References included full‐text articles, trial register abstracts, conference abstracts, and sub‐studies

^##One study could not be included in any meta‐analyses because of insufficient available data

RCT: Randomised controlled trial; CRT: Cluster‐randomised controlled trial

Data extraction and management

Two review authors (of JS, TE, AL, MB, JP and MZ) independently extracted the following data from included studies using Covidence (Covidence 2016): relevant outcome measures, sample size, demographics of participants, disease severity, setting, duration and contents of the intervention. We used standard data extraction forms and spreadsheets for study characteristics and outcome data. At the start of the data extraction, review authors independently extracted data from five studies. We compared results, and discussed any differences between review authors. We then optimised the data extraction form for study characteristics and outcome data.

One review author (JS) transferred data into the Review Manager Web (RevMan Web) file(RevMan Web 2021). We double‐checked the accuracy of data entry for newly included studies by comparing data presented in the RevMan Web file with the data‐extraction forms (one of TE, AL, MB or JP).

Assessment of risk of bias in included studies

Two review authors (of JS, TE, AL, MB, JP and MZ) independently assessed risk of bias in the included studies using the Cochrane tool known as the 'risk of bias 2'(RoB 2) tool (Sterne 2019), as outlined in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2019, hereafter referred to as the Cochrane Handbook), for the following five domains.

• Bias arising from the randomisation process.

• Bias due to deviations from intended interventions.

• Bias due to missing outcome data.

• Bias in measurement of the outcome.

• Bias in selection of the reported result.

For the findings of each included study, two review authors (of JS, TE, AL, MB, JP and MZ) independently answered signalling questions to reach a risk of bias judgement related to each domain using a predefined algorithm. Subsequently, the overall risk of bias was assessed for each included study as 'high risk', 'low risk' or 'some concerns', using criteria detailed in the Cochrane Handbook (Higgins 2019). Again, we conducted calibration exercises at the start of the risk of bias assessments. Therefore, review authors independently assessed risk of bias in five studies. We compared results and discussed any differences between review authors. We resolved disagreements through discussion, and if necessary, involved a third review author (JS, TE, AL, MB, JP or MZ).

We report the grade of each potential bias per outcome of the included studies, together with a justification for our judgement, in the ‘Included studies’ section. In case of a CRT, we used a special variant of the RoB 2 tool (Sterne 2019), that focuses mainly on groups of participants from the clusters.

Assessment of bias in conducting the systematic review

We conducted this review according to our prespecified protocol. We detail deviations from the protocol in the ‘Differences between protocol and review’ section.

Measures of treatment effect

We synthesised study results using random‐effects modelling (REM) in RevMan Web (RevMan Web 2021), and displayed these in forest plots. For continuous outcomes, we reported mean differences (MDs) or the standardised mean differences (SMDs) with the 95% confidence intervals (CIs). We used final scores in our meta‐analyses if available, but if unavailable, we included the change from baseline scores. For dichotomous outcomes, we reported odds ratios (ORs) with corresponding 95% CIs or, in case of outcomes with few events, risk differences (RDs) with corresponding 95% CIs.

We determined the clinical relevance of treatment effects by using the minimal clinically important difference (MCID), when available. We calculated numbers needed to treat for an additional beneficial outcome (NNTB) for respiratory‐related hospitalisations, all‐cause hospitalisations, respiratory‐related mortality and all‐cause mortality, using pooled ORs and control group data from individual studies within the meta‐analysis with Visual Rx 4 (Visual Rx 2016). The calculation of NNTBs was performed in four steps: 1) we calculated the mean control event risks over the mean follow‐up duration of the studies with the highest and lowest baseline risks; 2) we calculated the usual care event risks per study (proportion (%) of participants who had at least one respiratory‐related hospital admission divided by the total number of usual care); 3) we made two equal groups, one including the studies with the highest baseline risks and one including the lowest baseline risks; and 4) we calculated the mean usual care event risk per group (using the same procedure as for calculating the risk per study).

Unit of analysis issues

The unit of analysis in the included RCTs was the participant. In case the unit of analysis was a cluster, we adjusted for this by inflating the standard errors, as outlined in Section 16.3.6 of the Cochrane Handbook (Higgins 2019). This method requires an intra‐cluster correlation coefficient (ICC). We ran sensitivity analyses for primary outcomes using adjustments of clustering assuming ICCs of 0.02 and 0.04. Furthermore, we included studies that compared more than two intervention groups in a meta‐analysis by making multiple pair‐wise comparisons. To avoid double‐counting of usual care group participants, we divided the usual care group number by two, to have two entries for the study in the meta‐analysis.

Dealing with missing data

In case of missing or incomplete data, we contacted study authors to request missing data. If study authors did not respond, we made a second ‐ and when necessary ‐ a third attempt to request missing data. If study authors did not respond after our third attempt, we analysed the available data and reported that data were missing. If we thought the missing data presented major bias, we took this into consideration in the GRADE (Grades of Recommendation, Assessment, Development and Evaluation) rating for affected outcomes (Guyatt 2011). We listed the study authors who have provided us with data for this and previous versions of the review in the ‘Acknowledgements’.

Assessment of heterogeneity

We explored variability among studies using the I² statistic (Higgins 2019). When substantial heterogeneity (I² > 50%) was detected, we discussed possible explanations and critically reconsidered the appropriateness of a meta‐analysis. Furthermore, in the meta‐analyses, we used a REM (estimated mean of a distribution of effects), rather than a fixed‐effect model (FEM), to account for heterogeneity.

Assessment of reporting biases

We explored possible reporting bias by assessing asymmetry in funnel plots to determine whether studies selectively reported as indicated in the paragraph, ‘Assessment of risk of bias in included studies’. We conducted a funnel plot when at least ten studies had been included.

Data synthesis

After exploring whether pooling of study outcomes was possible, we calculated a summary statistic for each study, to describe the observed intervention effect in the same way for every study. If appropriate, we performed a meta‐analysis using RevMan Web (RevMan Web 2021). We considered a meta‐analysis when at least three studies reported sufficient data for the outcome. Because of the nature of the intervention analysed in this review, we expected clinical heterogeneity between the studies. We planned to perform meta‐analyses using a REM if pooling was possible, but considered a FEM if the included interventions were very similar.

For primary outcomes, we performed primary and secondary analyses. The primary analysis included the final study endpoint outcome scores, regardless of length of follow‐up. The secondary analyses included short‐term (≤ 6 months), medium‐term (> 6 to ≤ 12 months), and long‐term (> 12 months) follow‐up. For the short‐ and medium‐term follow‐up, follow‐up scores closest to 6 and 12 months, respectively, were included. For the long‐term follow‐up, final scores were included if the follow‐up was longer than 12 months. For secondary outcomes, we only performed meta‐analyses including the final study end point outcome scores, regardless of length of follow‐up.

Subgroup analysis and investigation of heterogeneity

We performed preplanned subgroup analyses when at least three studies could be included in each subgroup. We defined the following subgroup analyses a priori to explain possible heterogeneity between study results.

• Duration of the intervention (< 8 weeks versus ≥ 8 weeks). No information is available regarding the most effective self‐management intervention duration. Effects of interventions with a shorter duration may well differ from those of longer duration.

• Inclusion of participants in the acute phase (having an acute exacerbation of COPD) versus stable state (at least four weeks post exacerbation and six weeks post hospitalisation). Acute exacerbations may hamper self‐management improvements. Awareness of the clinical sequelae of acute exacerbations of COPD enables approaches such as early post‐exacerbation rehabilitation to mitigate its negative effects (Goldstein 2014).

• COPD self‐management interventions delivered in different income countries (low‐ and middle‐income countries versus high‐income countries). We classified countries according to the World Bank list of economies (World Bank 2021). We expected a priori more room for improvement after the implementation of a self‐management intervention in low‐ and middle‐income countries compared to high‐income countries as we expected that some elements of self‐management interventions may already have been included as part of usual care in high‐income countries but not in low‐ and middle‐income countries. The latter also face challenges with COPD diagnosis and management, including poorly‐resourced primary care systems and lack of trained workforces (Mills 2014).

• COPD self‐management interventions delivered in different care settings: primary care versus secondary and tertiary care. Self‐management interventions delivered in primary care may appear to be less effective (Jolly 2018). This may be driven by large heterogeneity in COPD populations, interventions and outcomes. We therefore decided that it is important to look at the effects in different healthcare settings separately.

• Inclusion of the following self‐management intervention components:

  • COPD exacerbation action plan component (inclusion of a COPD exacerbation action plan component versus no COPD exacerbation action plan component in the self‐management intervention). An exacerbation action plan is defined as a guideline (a hard copy or via audiovisual media) for participants with COPD describing when and how to act in case of worsening COPD‐related symptoms, indicating (the onset of) an exacerbation. Inclusion of COPD exacerbation action plans may result in improved HRQoL and lower probability of respiratory‐related hospital admissions (Lenferink 2017).

  • Home‐based exercise or physical activity component (inclusion of a home‐based exercise or physical activity component in the self‐management intervention versus no exercise component in the self‐management intervention). Increased exercise capacity may result in better HRQoL and potentially fewer hospital admissions (McCarthy 2015).

  • Smoking cessation component (inclusion of a smoking cessation component in the self‐management intervention versus no smoking cessation component in the self‐management intervention). Smoking cessation may result in improved HRQoL (Cheruvu 2016; Van Eerd 2016).

  • Diet component (inclusion of a diet component in the self‐management intervention versus no diet component in the self‐management intervention) ‐ for example, evaluation and optimisation of participants' diet and nutritional intake.

  • COPD medication component (inclusion of a medication component in the self‐management intervention versus no medication component in the self‐management intervention) ‐ for example, advice about medication intake, adherence and inhalation technique.

  • Coping with breathlessness component (inclusion of a coping with breathlessness component in the self‐management intervention versus no coping with breathlessness component in the self‐management intervention).

  • Self‐recognition of COPD exacerbations component (inclusion of a self‐recognition of COPD exacerbations component in the self‐management intervention versus no self‐recognition of COPD exacerbations component in the self‐management intervention).

• The effects of COPD self‐management interventions with and without use of digital technology. We expected that COPD self‐management interventions with use of digital technology may have an added positive impact on HRQoL, hospital admissions and exercise capacity (McLean 2011; McCabe 2017).

• The integration of behavioural change techniques (BCTs) in COPD self‐management interventions. The BCT taxonomy (version 1) is a hierarchically structured, cross‐domain list of 93 distinct BCTs described in 16 different clusters: 1) Goals and planning, 2) Feedback and monitoring, 3) Social support, 4) Shaping of knowledge, 5) Natural consequences, 6) Comparison of behaviours, 7) Associations, 8) Repetition and substitution, 9) Comparison of outcomes, 10) Reward and threat, 11) Regulation, 12) Antecedents, 13) Identity, 14) Scheduled consequences, 15) Self‐belief, and 16) Covert learning (Michie 2013). BCTs applied in self‐management interventions were extracted by using the mobile BCT taxonomy application (BCT Taxonomy; Michie 2013). We only extracted data that were explicitly reported in published articles of included studies. We performed the following two subgroup analyses.

  • COPD self‐management interventions by integration of two BCTs versus less than two BCTs in the intervention (Michie 2013).

  • The number of BCT taxonomy clusters in COPD self‐management interventions: ‘lower or equal’ versus ‘higher’ than the median of BCT clusters found in all included interventions (Michie 2013).

We used the formal test for subgroup interactions in RevMan Web (RevMan Web 2021).

Sensitivity analysis

We planned the following sensitivity analyses, which we conducted using different assumptions to investigate the robustness of effect sizes found in this review.

• Assumption of small‐study effects: to identify whether review findings were dependent on study characteristics (e.g. studies with low and high numbers of included participants), by using REM versus FEM.

• Assumption of influencing outliers: to explore whether review findings were dependent on variation in results, by excluding those studies with outlying results from the analysis.

Summary of findings and assessment of the certainty of the evidence

Using the criteria outlined in the Cochrane Handbook (Higgins 2019), we created a Table 1 (SOF table), including key information concerning the quality of evidence, the magnitude of effect of the self‐management intervention and the sum of available data for the main outcomes. We used the five GRADE considerations (study limitations, consistency of effect, imprecision, indirectness and publication bias) to assess the quality of the body of evidence as it related to studies that contributed data to the meta‐analyses for prespecified outcomes. We used methods and recommendations described in Section 8.5 and Chapter 12 of the Cochrane Handbook (Higgins 2019), by using GRADEpro GDT software. In the SOF table footnotes and comments, we included justifications for decisions to downgrade the quality of studies, to aid the reader’s understanding of the review.

Results

Description of studies

See Characteristics of included studies.

Results of the search

Searches over the period January 2011 until January 2020 identified 6311 titles and abstracts (Figure 1). After de‐duplication and pre‐screening by Screen4Me workflow, 3576 records remained. We identified 458 potentially eligible articles, from which 21 studies were included. In addition, six of 29 studies included in our previous update (1995 to 2011) met the stricter inclusion criteria of this update (Figure 1). Therefore, a total of 27 studies (62 references) have been included in this review. One of these 27 studies could not be included in any quantitative syntheses (meta‐analyses) because of insufficient data (Emery 1998). Another study included two intervention groups versus one usual care group (Coultas 2005); all three study groups were included in meta‐analyses.

An update search in March 2021 identified 1280 titles and abstracts. After de‐duplication and prescreening by Screen4Me workflow, 640 records remained. We identified 55 potentially eligible articles. From these, 22 studies were excluded; six studies were classified as ongoing; 26 studies await classification; and one study ‐ Ozturk 2020 ‐ will be fully incorporated in a future update of this review, if the study criteria of this future update remain unchanged.

Included studies

We tabulated details of the 27 included studies (Benzo 2016; Bischoff 2012; Bösch 2007; Bourbeau 2003; Bringsvor 2018; Bucknall 2012; Coultas 2005; Emery 1998; Fan 2012; Ferrone 2019; Gallefoss 1999; Hernández 2015; Johnson‐Warrington 2016; Jolly 2018; Jonsdottir 2015; Kessler 2018; Lenferink 2019; Liang 2019; Martin 2004; Mitchell 2014; Rice 2010; Rose 2018; Sanchez‐Nieto 2016; Tabak 2014; Titova 2015; Walters 2013; Wang 2019). Participant characteristics can be found in Table 2, and the intervention and follow‐up details in Table 3. Coultas 2005 used two intervention groups and one usual care group. Two of the 27 included studies were CRTs (Liang 2019; Walters 2013); the others were RCTs. Because Walters 2013 did not adjust their reported outcomes for clustering, we manually adjusted the data using a calculated average cluster size of 5.8710 participants (i.e. 182 participants across 31 practices) and an ICC of 0.05, resulting in a design effect of 1.24.

1. Characteristics of participants in included studies.

Study	Randomised COPD participants (number)		Lost to follow‐up (%)		Age (years; mean (SD))		Gender (% male)		Current smokers (%)		FEV1/FVC ratio (SD)		FEV1% predicted (SD)
	Self‐management	Usual care	Self‐management	Usual care	Self‐management	Usual care	Self‐management	Usual care	Self‐management	Usual care	Self‐management	Usual care	Self‐management	Usual care
Benzo 2016	108	107	14.8	0.9	67.9 (9.8)	68.1 (9.2)	43	48	NR	NR	48.2 (13.8)	47.7 (13.8)	40.5 (17.1)	40.3 (17.2)
Bischoff 2012	55	55	10.9*	20*	65.5 (11.5)	63.5 (10.3)	67	51	29	33	43 (78)**	38 (69)**	66.3 (16.5)	67.0 (18.0)
Bösch 2007	38	12	21.1*	8.3*	63.8 (8.4)	64.6 (6.8)	NR	NR	13.3	27.3	NR	NR	45.9 (17.5)	47.8 (16.9)
Bourbeau 2003	96	95	10.4	16.8	69.4 (6.5)	69.6 (7.4)	52	59	25	26	46	45	NR	NR
Bringsvor 2018	92	90	31.5*	22.2	68.5 (8.16)	69.3 (9.02)	59	63	NR	NR	45.2 (12.4)	45.1 (12.7)	45.2 (14.4)	44.8 (16.2)
Bucknall 2012	232	232	16.8	21.1	70.0 (9.3)	68.3 (9.2)	38	35	39	39	46.4 (0.12)	45.4 (0.12)	41.2 (13.4)	39.8 (13.8)
Coultas 2005#,a	72	73	31.9	30.1	68.3 (6.6)	68.8 (10.4)	42.9	53.8	28.6	27.5	48.08 (12.35)	52.05 (12.99)	NR	NR
Coultas 2005#,b	72	73	29.2	30.1	70.1 (7.0)	68.8 (10.4)	32.7	53.8	23.5	27.5	49.85 (11.18)	52.05 (12.99)	NR	NR
Emery 1998	25	25	8.0*	0.0*	67.4 (5.9)	67.4 (5.9)	40	48	14	20	45 (11)	43 (12)	43 (18)	43 (18)
Fan 2012	209	217	17.2	9.2	66.2 (8.4)	65.8 (8.2)	97.6	96.3	28.2	27.2	47 (12)	47 (12)	38.2 (14.3)	37.8 (14.5)
Ferrone 2019	84	84	14.3	11.9	68.6 (9.6)	67.9 (9.8)	40.5	52.4	39.3	57.1	55.6 (11.8)	53.6 (10.4)	55.5 (14.5)	53.2 (14.7)
Gallefoss 1999	31	31	16.1	12.9	57 (9)	58 (10)	48	52	39	39	55 (9)	52 (10)	59 (9)	56 (11)
Hernández 2015	76	84	22.4	34.5	73 (8)	75 (9)	83	86	13	14	47 (13)	47 (15)	41 (19)	44 (20)
Johnson‐Warrington 2016	38	39	10.5	7.7	67.6 (8.5)	68.3 (7.7)	38.4	33.3	35.9	46.2	47.1 (14.0)	42.8 (10.5)	40.5 (15.7)	42.5 (11.7)
Jolly 2018	289	288	14.5	2.4	70.7 (8.8)	70.2 (7.8)	63	64	26	19	NR	NR	71.2 (18.9)	72.1 (18.7)
Jonsdottir 2015	60	59	20.0	11.9	59.4 (4.7)	58.7 (4.4)	39.6	51.9	50.0	69.2	NR	NR	54.0 (17.6)	60.9 (17.3)
Kessler 2018	172	173	20.3	26.0	67.3 (8.9)	66.6 (9.6)	69.4	69.8	21.7	21.0	45.7 (11.3)	43.7 (11.3)	37.8 (12.4)	36.4 (12.3)
Lenferink 2019	102	99	16.7	15.2	68.8 (9.0)	68.2 (8.9)	64.7	63.6	19.6	20.2	49.3 (14.3)	48.5 (12.2)	53.4 (16.1)	50.7 (14.3)
Liang 2019	157	115	28.0	33.0	66.6 (10.8)	61.7 (10.1)	60.5	62.6	53.5	71.3	57 (13)	57 (10)	69.0 (20.5)	70.8 (19.3)
Martin 2004	44	49	20.5	8.2	71.1 (68.7, 73.5)##	69.1 (63.5, 74.7)##	34.1	65.3	NR	NR	NR	NR	35.4 (31.6, 39.2)##	34.3 (31.2, 37.4)##
Mitchell 2014	89	89	27.0	18.0	69 (8)	69 (10.1)	60.7	49.5	20.2	22.1	49.8 (13.4)	50.5 (11.1)	56.0 (16.8)	60.0 (17.4)
Rice 2010	372	371	9.7	12.9	69.1 (9.4)	70.7 (9.7)	97.6	98.4	21.6	23.0	53.0 (14.5)	54.4 (14.3)	36.1 (14.5)	38.2 (14.4)
Rose 2018	237	238	12.7	19.7	71 (9.2)	71 (9.7)	50	44	23	26	50 (12.6)	52 (13.0)	43 (17.0)	45 (17.8)
Sanchez‐Nieto 2016	51	45	7.8$	15.6$	68.2 (7.2)	67.1 (6.8)	92.2	88.9	37.3	35.6	53 (17)	55 (10)	47.3 (14.4)	44.3 (11.9)
Tabak 2014	15	14	33.3$	71.4$	64.1 (9.0)	62.8 (7.4)	50.0	50.0	36.4	33.3	36.5 (29.5‐51.0)$$	33.5 (26.0‐52.0)$$	50.0 (33.3‐61.5)$$	36.0 (26.0‐53.5)$$
Titova 2015^	91	81	44.0	39.5	73.6 (9.2)	72.2 (9.4)	42.9	42.5	33.0	40.0	NR	NR	33.6 (9.9)	33.4 (9.4)
Walters 2013	90	92	17.8	13.0	68.2 (7.9)	67.3 (7.6)	54	51	48	36	56 (12)	50 (11)	54.0 (13.4)	56.4 (13.2)
Wang 2019	77	77	6.5	7.8	68.7 (6.2)	69.2 (6.1)	76.6	80.5	44.2	41.6	55.2 (18.2)	56.7 (16.9)	58.4 (17.3)	59.2 (18.2)

COPD: chronic obstructive pulmonary disease; FEV1: forced expiratory volume in one second; FVC: forced vital capacity; NR: not reported; SD: standard deviation.

^*No deaths reported and included in these data; ^**Post‐bronchodilator FEV1/FVC < lower limit of normal; ^#Study with one usual care group and two intervention groups, number of participants in usual care group halved in meta‐analyses; ^##mean (95% confidence interval); ^$unclear whether the deaths were included in these data (self‐management: 0; usual care: 2); ^$$median (interquartile range); ^{^}Different baseline data reported in 2015 and 2017 articles, data of 2017 article included in this Table; ^anurse‐assisted medical management intervention group; ^bnurse‐assisted collaborative management intervention group

2. Characteristics of interventions in included studies.

Study	Follow‐up (months)	Setting; provision intervention	Time period intervention (months); duration of sessions	Included components with iterative process
Benzo 2016	12	Outpatient clinic	12; 2 FTF individual sessions (first visit 120 min, second visit not reported) and 6 phone calls (mean duration 28.6 min (SD 10.0).	Self‐recognition of COPD exacerbations Use of a COPD exacerbation action plan Home‐based exercise or physical activity component Coping with breathlessness
Bischoff 2012	24	General practice	24; 2‐4 FTF individual sessions (60 min each) scheduled in 4 to 6 consecutive weeks, 6 phone calls	Smoking cessation (optional) Self‐recognition of COPD exacerbations Use of a COPD exacerbation action plan: Home‐based exercise or physical activity component (optional) Diet COPD medication intake (i.e. adherence, inhalation technique) Coping with breathlessness Managing anxiety and stress
Bösch 2007	12	Outpatient clinic; (University) hospital	12; 4 FTF group sessions (120 min each) and final session scheduled 6 weeks later	Smoking cessation (optional) Self‐recognition of COPD exacerbations Use of a COPD exacerbation action plan Home‐based exercise or physical activity component Diet COPD medication intake (i.e. adherence, inhalation technique) Coping with breathlessness Leisure activities and travelling
Bourbeau 2003	12 (24*)	(University) hospital	12; 7 FTF individual sessions (60 min each) scheduled in 7 to 8 consecutive weeks, 18 phone calls	Self‐recognition of COPD exacerbations Use of a COPD exacerbation action plan Home‐based exercise or physical activity component (optional) Diet COPD medication intake (i.e. adherence, inhalation technique) Coping with breathlessness Leisure activities and travelling Energy conservation during day‐by‐day activities Relaxation exercises Adopting a healthy lifestyle Long‐term oxygen (optional)
Bringsvor 2018	3	Meeting locations in the participants’ municipalities.	2.5; 11 FTF group sessions (120 min) scheduled weekly	Smoking cessation (optional) Self‐recognition of COPD exacerbations Use of a COPD exacerbation action plan Home‐based exercise or physical activity component Diet COPD medication intake (i.e. adherence, inhalation technique) Coping with breathlessness Psychological issues Information about the healthcare system, including local, regional and national “offers” for persons with COPD
Bucknall 2012	12	Home‐based	12; 4 FTF individual sessions (40 min each) in 2 months, at least 6 subsequent home visits, 828 phone calls intervention group	Smoking cessation Self‐recognition of COPD exacerbations Use of a COPD exacerbation action plan Diet (optional) COPD medication intake (i.e. adherence, inhalation technique) Coping with breathlessness
Coultas 2005a	6	Home‐based	6; 1 FTF individual session (mean duration 64 min (SD 23.1) and mean 6 (SD 1.8) phone calls (mean duration 10 min (SD 5.4)	Smoking cessation (optional) Self‐recognition of COPD exacerbations (optional) Use of a COPD exacerbation action plan (optional) COPD medication intake (i.e. adherence, inhalation technique) (optional)
Coultas 2005b	6	Home‐based	6; 1 FTF individual session (mean duration 64 min (SD 23.1) and mean 6 (SD 1.8) phone calls (mean duration 10 min (SD 5.4)	Smoking cessation (optional) Self‐recognition of COPD exacerbations (optional) Use of a COPD exacerbation action plan (optional) COPD medication intake (i.e. adherence, inhalation technique) (optional)
Emery 1998	2.5	Rehabilitation centre; (University) hospital	2.5; 26 FTF group sessions (16 lectures of 60 min and 10 management sessions of 60 min)	Self‐recognition of COPD exacerbations COPD medication intake (i.e. adherence, inhalation technique) Coping with breathlessness Relaxation exercises Coping skills training
Fan 2012	12#	Outpatient clinic	12; 4 FTF individual sessions (90 min each) scheduled weekly, 1 FTF group session, 6 phone calls (duration not specified)	Self‐recognition of COPD exacerbations Use of a COPD exacerbation action plan COPD medication intake (i.e. adherence, inhalation technique)
Ferrone 2019	12	General practice	9; 2 FTF individual sessions (first visit 60 min (baseline evaluation) and 5 to 7 min (encounter with physician) and second visit after 3 months of 45 min) and either a phone call or FTF visit at 6 and 9 months (15 to 30 min each)	Smoking cessation (optional) Self‐recognition of COPD exacerbations Use of a COPD exacerbation action plan Home‐based exercise or physical activity component COPD medication intake (i.e. adherence, inhalation technique) Coping with breathlessness
Gallefoss 1999	12	Outpatient clinic	1‐2; 1 or 2 FTF individual sessions by a nurse and 1 or 2 by physiotherapist (40 min each), 2 FTF group sessions (120 min each)	Smoking cessation (optional) Self‐recognition of COPD exacerbations COPD medication intake (i.e. adherence, inhalation technique) Coping with breathlessness
Hernández 2015	12 (84$)	Outpatient clinic; (University) hospital	12; Participants with no mobility problems: 1 FTF individual session (40 min) at home by primary care team, 3 FTF group sessions at outpatient clinic (2 x 90 min, 1 x 120 min) Participants with mobility problems: 4 FTF individual sessions (15 min each), 1 FTF individual session (120 min) or 1 FTF group session (40 min), all at home by primary care team All participants: Web based calls at least once per month (15 min each)	Smoking cessation (optional) Self‐recognition of COPD exacerbations Use of a COPD exacerbation action plan Home‐based exercise or physical activity component Diet COPD medication intake (i.e. adherence, inhalation technique) Coping with breathlessness Comorbid condition (no further explanation regarding the content)
Johnson‐Warrington 2016	3	(University) hospital; Home‐based	3; 1 FTF individual session (30 to 45 min) and 6 phone calls (5 to 20 min each)	Smoking cessation (optional) Self‐recognition of COPD exacerbations Use of a COPD exacerbation action plan Home‐based exercise or physical activity component COPD medication intake (i.e. adherence, inhalation technique) Coping with breathlessness
Jolly 2018	12	General practice; Home‐based	5.5; 4 individual phone calls (first call 35 to 60 min, other calls 15 to 20 min) scheduled at 3, 7 and 11 weeks	Smoking cessation (optional) Self‐recognition of COPD exacerbations (optional) Use of a COPD exacerbation action plan (optional) Home‐based exercise or physical activity component COPD medication intake (i.e. adherence, inhalation technique)
Jonsdottir 2015	12	Clinical research centre located on a university‐hospital	1‐2; 1 FTF group session (120 min), 3 to 4 FTF individual sessions (30 to 45 min), and 4 phone calls (5 to 10 min each)	Smoking cessation (optional) Self‐recognition of COPD exacerbations Home‐based exercise or physical activity component Diet COPD medication intake (i.e. adherence, inhalation technique) Coping with breathlessness Utilization of health care Prevent further decline of disease within the aim of enhancing health of patient and family Coping with feelings of shame and guilt
Kessler 2018	12	Outpatient clinic Home‐based	12; 1 FTF group session (90 to 120 min), 4 FTF individual sessions (60 to 90 min), and multiple phone calls (duration not specified)	Smoking cessation (optional) Self‐recognition of COPD exacerbations Use of a COPD exacerbation action plan Home‐based exercise or physical activity component (optional) Diet COPD medication intake (i.e. adherence, inhalation technique) Coping with breathlessness
Lenferink 2019	12	(University) hospital	12; 2‐3 FTF group sessions (120 to 240 min), 2 FTF individual sessions (60 min), and 3 phone calls (10 to 15 min each)	Self‐recognition of COPD exacerbations Use of a COPD exacerbation action plan COPD medication intake (i.e. adherence, inhalation technique) Coping with breathlessness Self‐recognition of increase in in comorbid symptoms and use of an action plan for these comorbidities (CHF, IHD, anxiety and depression)
Liang 2019	12	General practice	2; 3 FTF individual sessions (duration not specified), and 9 phone calls (duration not specified)	Smoking cessation (optional) Home‐based exercise or physical activity component COPD medication intake (i.e. adherence, inhalation technique)
Martin 2004	12	General practice; (University) hospital; Home‐based; Ambulance service	12 4 FTF individual sessions and respiratory nurse visits at 3, 6 and 12 months	Use of a COPD exacerbation action plan COPD medication intake (i.e. adherence, inhalation technique) Guidance regarding treatment for coexisting conditions (e.g. when/how to use oxygen therapy, and when to use diuretics)
Mitchell 2014	6	General practice; Home‐based	1; 1 FTF individual session (30 to 45 min) by a physiotherapist and 2 phone calls (duration not specified)	Smoking cessation (optional) Self‐recognition of COPD exacerbations Use of a COPD exacerbation action plan Home‐based exercise or physical activity component Management of psychological consequences (e.g. dealing with anger, depression, disease acceptance)
Rice 2010	12	Outpatient clinic	12; 1 FTF group session (60 to 90 min) by a respiratory therapist case manager, 12 monthly phone calls (10 to 15 min each)	Smoking cessation (optional) Self‐recognition of COPD exacerbations Use of a COPD exacerbation action plan COPD medication intake (i.e. adherence, inhalation technique)
Rose 2018	12	Outpatient clinic	12; 1 FTF individual session (40 min), 21 phone calls (duration not specified)	Smoking cessation (optional) Self‐recognition of COPD exacerbations Use of a COPD exacerbation action plan COPD medication intake (i.e. adherence, inhalation technique) Advance care planning
Sanchez‐Nieto 2016	12	(University) hospital	3; 1 FTF group session (40 min), and 3 FTF individual sessions (20 min each)	Self‐recognition of COPD exacerbations Use of a COPD exacerbation action plan Home‐based exercise or physical activity component COPD medication intake (i.e. adherence, inhalation technique)
Tabak 2014	9	Outpatient clinic; Primary care physiotherapy practices	9; 2 FTF group sessions (90 min each) by a nurse practitioner, 1 FTF individual session and 1 x intake by the physiotherapist, additional meetings after 1, 3, 6 and 9 months	Smoking cessation (optional) Self‐recognition of COPD exacerbations Use of a COPD exacerbation action plan Home‐based exercise or physical activity component
Titova 2015	24	Home‐based	24; 6 FTF individual sessions (1 x at discharge, 5 x home visits at 3 and 14 days, and at 6, 12, 24 months) by the specialist nurse, 1 e‐learning programme (15 min), at least 24 phone calls	Self‐recognition of COPD exacerbations Use of a COPD exacerbation action plan Coping with breathlessness
Walters 2013	12	Home‐based	12; 16 individual phone calls (30 min each)	Smoking cessation (optional) Self‐recognition of COPD exacerbations Use of a COPD exacerbation action plan Home‐based exercise or physical activity component (optional) Diet (optional) COPD medication intake (i.e. adherence, inhalation technique) (optional) Alcohol (optional) Psychosocial (optional)
Wang 2019	12	(University) hospital	3; 5 to 6 FTF individual sessions (45 min each), 3 home visits (45 to 60 min each), and weekly phone calls scheduled over 3 months (10 to 15 min each)	Smoking cessation (optional) Home‐based exercise or physical activity component COPD medication intake (i.e. adherence, inhalation technique) Coping with breathlessness Respiratory muscle training (pursed lip breathing and abdominal breathing) Coughing techniques Long‐term home oxygen therapy (optional)

COPD: chronic obstructive pulmonary disease; FTF: face‐to‐face; min: minute(s)

^*Second year data based on provincial health insurance and hospitalisation database records; ^#study was terminated early after a mean follow‐up time of 250 days; ^$two groups were passively followed up for 6 additional years; ^anurse‐assisted medical management intervention group; ^bnurse‐assisted collaborative management intervention group

Participants and recruitment

A total of 6008 participants (self‐management intervention n = 3074; usual care n = 2934) were assessed in the 27 included studies (Table 2). Dropout rates in the studies ranged from 0% to 71.4%, and a total of 5125 (85%) participants completed the study follow‐up.

Interventions

The content of self‐management interventions in the 27 included studies was diverse (Table 3). The follow‐up duration was three months or less in three (11%) studies, six months in two (7%), nine months in one (4%), 12 months in 18 studies (67%), and 24 months in three studies (11%). Self‐management interventions were delivered individually in 15 (56%) studies (Benzo 2016; Bischoff 2012; Bourbeau 2003; Bucknall 2012; Coultas 2005; Ferrone 2019; Johnson‐Warrington 2016; Jolly 2018; Liang 2019; Martin 2004; Mitchell 2014; Rose 2018; Titova 2015; Walters 2013; Wang 2019), in small groups in three studies (11%) (Bösch 2007; Bringsvor 2018; Emery 1998), and included both individual and group sessions in nine (33%) studies (Fan 2012; Gallefoss 1999; Hernández 2015; Jonsdottir 2015; Kessler 2018; Lenferink 2019; Rice 2010; Sanchez‐Nieto 2016; Tabak 2014). The median duration of the intervention, including self‐management reinforcement, was nine months (interquartile range (IQR) 3.0 to 12.0). The intervention duration was three months or less in nine (33%) studies (Bringsvor 2018; Emery 1998; Gallefoss 1999; Johnson‐Warrington 2016; Jonsdottir 2015; Liang 2019; Mitchell 2014; Sanchez‐Nieto 2016; Wang 2019), over three months and up to six months in two (7%) studies (Coultas 2005; Jolly 2018), nine months in two (7%) studies (Ferrone 2019; Tabak 2014), 12 months in 12 (44%) studies (Benzo 2016; Bösch 2007; Bourbeau 2003; Bucknall 2012; Fan 2012; Hernández 2015; Kessler 2018; Lenferink 2019; Martin 2004; Rice 2010; Rose 2018; Walters 2013), and 24 months in two (7%) studies (Bischoff 2012; Titova 2015).

A ‘COPD exacerbation action plan’ was part of the self‐management intervention in 23 (85%) studies; both ‘self‐recognition of COPD exacerbations’ and ‘a medication component’ were each part of 22 study interventions (81%); ‘home‐based exercise or physical activity component’ was part of 17 study interventions (63%); a ‘coping with breathlessness component’ was present in 16 study interventions (59%); a ‘smoking cessation component’ was part of 15 study interventions (56%); and a diet component was present in nine study interventions (33%).

A median of 4.0 (IQR 3.0 to 7.0) BCT clusters was detected per study intervention, with a minimum of two BCT clusters (Bischoff 2012; Bösch 2007; Bringsvor 2018; Coultas 2005) and a maximum 11 BCT clusters (Johnson‐Warrington 2016; Mitchell 2014). The BCT clusters that were integrated in COPD self‐management interventions groups to promote the uptake and optimal use of COPD self‐management behaviour patterns were: goals and planning (n = 28, all intervention groups of 27 studies, one study with two intervention groups); feedback and monitoring (n = 28, all intervention groups of 27 studies, one study with two intervention groups); shaping knowledge (n = 19, all but nine comparisons (Bischoff 2012; Bösch 2007; Bringsvor 2018; Coultas 2005; Emery 1998; Liang 2019; Rose 2018; Titova 2015)); social support (n = 16, all but twelve comparisons (Bischoff 2012; Bösch 2007; Bourbeau 2003; Bringsvor 2018; Coultas 2005; Emery 1998; Ferrone 2019; Gallefoss 1999; Kessler 2018; Martin 2004; Sanchez‐Nieto 2016; Tabak 2014)); natural consequences (n = 9 (Bucknall 2012; Fan 2012; Gallefoss 1999; Johnson‐Warrington 2016; Jolly 2018; Jonsdottir 2015; Liang 2019; Mitchell 2014; Wang 2019)); repetition and substitution (n = 9 (Benzo 2016; Bourbeau 2003; Bucknall 2012; Fan 2012; Johnson‐Warrington 2016; Jonsdottir 2015; Lenferink 2019; Mitchell 2014; Titova 2015)); regulation (n = 7 (Emery 1998; Fan 2012; Gallefoss 1999; Johnson‐Warrington 2016; Jolly 2018; Jonsdottir 2015; Mitchell 2014)); comparison of behaviour (n = 6 (Bourbeau 2003; Bucknall 2012; Hernández 2015; Johnson‐Warrington 2016; Jonsdottir 2015; Mitchell 2014)); associations (n = 6 (Bourbeau 2003; Fan 2012; Gallefoss 1999; Hernández 2015; Johnson‐Warrington 2016; Mitchell 2014)); antecedents (n = 6 (Bourbeau 2003; Emery 1998; Gallefoss 1999; Johnson‐Warrington 2016; Jolly 2018; Mitchell 2014)); identity (n = 3 (Johnson‐Warrington 2016; Jonsdottir 2015; Mitchell 2014)); self‐belief (n = 3 (Bucknall 2012; Jonsdottir 2015; Walters 2013)); comparison of outcomes (n = 1 (Jolly 2018)); and reward and threat (n = 1 (Liang 2019)). There were no scheduled consequences or covert learning reported in any of the self‐management interventions.

Adherence

Half of the studies (n = 13) reported details regarding participants’ adherence to the self‐management intervention. Of these, nine studies reported adherence as the number or percentage of sessions attended by participants. In Emery 1998, the self‐management group attended approximately 88% of both the education and stress management sessions. In Gallefoss 1999, they used a per‐protocol analysis and withdrew intervention group participants who did not attend the individual or group sessions (n = 5, 16%). In Bischoff 2012, the total number of sessions that were offered to participants depended on participants' needs, with a minimum of two. Participants in Bischoff 2012 received a mean of 3.4 (SD 1.5) sessions; 13% did not attend any sessions or telephone calls. Fan 2012 reported that, during the entire follow‐up period, eight of 209 participants in the self‐management intervention group and 10 of 217 participants in the usual care group either did not attend any scheduled visits or formally withdrew from the study. The study authors also reported that in the self‐management intervention group, 87% completed all four individual educational visits and 57% completed the scheduled group visit (Fan 2012). Early termination after the intervention was enforced by the Data and Safety Monitoring Committee and the apparently low attendance rate of the group visit may well be a consequence (Fan 2012).

Tabak 2014 reported that the self‐management module on the web portal, including the self‐treatment of COPD exacerbations, was used on 86% of treatment days per participant. Benzo 2016 reported that 85% of the participants in the self‐management intervention group received a complete intervention, defined as at least 70% of 21 phone calls completed. In Kessler 2018, 100% of the participants in the self‐management group completed all four initial individual home coaching sessions; 66.7% achieved at least 80% of their phone and group coaching; and 89% achieved at least 80% for weekly phone health status transmission, which demonstrated that most participants adhered to the intervention. Liang 2019 reported that only 31% of the participants completed the full self‐management intervention; 26% partially completed the intervention; and 43% did not receive the intervention. Rose 2018 reported that 29% of the participants were 100% compliant with all 12 weekly phone calls, and 31% of the participants were 100% compliant with all nine subsequent monthly phone calls.

Jolly 2018 reported adherence regarding medication. Participants in the self‐management intervention improved medication adherence in six months compared to baseline, with higher proportions having: an inhaler check (86% versus 55%); an agreed care plan with a healthcare provider (44% versus 30%); written advice about what to do if symptoms worsened (23% versus 17%); and an antibiotic rescue pack (37% versus 29%).

Comparisons

As per our inclusion criteria, self‐management interventions that included an iterative process for at least two of its intervention components were compared with usual care in 27 studies. Coultas 2005 used two intervention groups and one usual care group. In meta‐analyses, both intervention groups were compared with the same usual care group, resulting in one extra comparison.

Outcomes

See Additional Table 4 for details on the number of included studies reporting outcomes of interest.

3. Number of studies reporting outcomes of interest.

	Studies
Primary outcomes
Health‐related quality of life scores	23
Respiratory‐related hospital admissions	20
Respiratory‐related mortality	8
All‐cause mortality	24
Secondary outcomes
All‐cause hospital admissions	18
Respiratory‐related hospitalisation days	5
All‐cause hospitalisation days	8
Emergency department visits	13
General practitioner visits	7
Specialist visits	5
COPD exacerbations	8
Use of courses of oral steroids and antibiotics	11
Health status	6
Anxiety or depression, or both	13
Self‐efficacy	9
Days lost from work	1
Exercise capacity and physical activity	7
Self‐management behaviour	2
Patient activation	1
Health literacy	0

COPD: chronic obstructive pulmonary disease.

Missing data

We have listed the authors from whom we received responses to requests for additional data in the ‘Acknowledgements’. However, not all study authors were able to provide the requested additional information. If the requested data were not provided, we described the data that were available.

Excluded studies

For the period 2011 to January 2020, we excluded 355 studies (437 references) following the assessment of the full‐text articles (Figure 1). The most frequent exclusion reason was that studies could not be classified as a COPD self‐management intervention (n = 78, 22.0%). For the period 1995 to 2011, we excluded 19 of the previously included studies (Zwerink 2014), because of stricter inclusion criteria for the population target (i.e. COPD diagnosis) and the self‐management intervention.

Studies awaiting classification

Twelve studies await classification because we could not reach the study authors to verify whether the studies met our eligibility criteria (Abdulsalim 2017; Aboumatar 2017; Alharbey 2019; Efraimsson 2008; Ghanem 2010; Heidari 2018; Hill 2010; Jiang 2012; Khdour 2009; Li 2014; Li Z 2015; Liu 2013).

Ongoing studies

We identified 38 ongoing studies (Boer 2011; Bourne 2017; Cecere Feemster 2013; Chen 2018; ChiCTR1800018197; ChiCTR‐TRC‐12002559; Chien 2016; Costa 2015; Dewan 2011a; Ding 2019; Doheny 2013; Duran 2017; Ergan 2018; Fleehart 2015; Gonzalez 2015; Hernandez 2016; Imanalieva 2016; IRCT201504149014N61; IRCT2017030432764N2; James 2012; Ko 2015; Moreno 2017; NCT02258646; NCT02924870; NCT03012256; NCT03084874; NCT03216603; NCT03721315; NL3827 (NTR4009); Padilla‐Zarate 2013; Paquin 2014; Reguera 2017; Sano 2016; Siddharthan 2018; Sirichana 2014; Thomas 2019; NL5277 (NTR5558); Zanaboni 2016).

Risk of bias in included studies

We present an overview of our risk of bias assessment per outcome in Figure 2. We performed assessments based on the content of the study articles, with no extra information requested from study authors. Further details and the rationale for judgements for primary outcomes, per outcome and per study, can be found in the Risk of bias (tables) section.

2.

Risk of bias overview for each outcome according to authors' judgements

None of the included studies reported blinding of participants and personnel, and none provided sufficient information regarding the balance of non‐protocol interventions over the study groups. Also, none of the included studies provided sufficient information regarding analyses to estimate the effect of adhering to interventions. These limitations resulted in high‐risk scores for all studies in domain 2 – ‘deviations from the intended interventions’ – of the risk of bias assessment form. As a result, we had to consider the overall risk of bias in each assessed study as ‘high’.

Other potential sources of bias

We explored possible reporting bias by assessing asymmetry in funnel plots. The St. George’s Respiratory Questionnaire (SGRQ) and respiratory‐related hospital admissions funnel plots seem to show a gap on both the lower left and right side of the graphs (funnel plots not shown). This could indicate that smaller studies with effects in favour of both the self‐management intervention and usual care group are published less frequently. By contrast, the funnel plot of all‐cause mortality seems to show a gap on the left side of the graph, indicating that smaller studies and studies of moderate size with effects in favour of self‐management interventions are published less frequently. For the latter, the same could be suggested by the funnel plot of all‐cause hospital admissions (funnel plot not shown). We could not rule out the contribution of other study factors to funnel plot asymmetry.

Effects of interventions

See: Table 1

We included a ‘Table 1’ (SOF table) of the 27 included studies. This table reflects the endpoints related to HRQoL, hospital admissions, mortality, dyspnoea, emergency department visits, anxiety and depression, and COPD exacerbations.

Health‐related quality of life

St. George's Respiratory Questionnaire (SGRQ)

COPD‐specific HRQoL was measured by the SGRQ in 16 studies (Bourbeau 2003; Bucknall 2012; Coultas 2005; Fan 2012; Gallefoss 1999; Hernández 2015; Jolly 2018; Jonsdottir 2015; Kessler 2018; Liang 2019; Martin 2004; Rice 2010; Rose 2018; Titova 2015; Walters 2013; Wang 2019). For primary analysis, mean adjusted SGRQ total scores, regardless of length of follow‐up, of 14 studies with 2778 participants could be included in the meta‐analysis (Bourbeau 2003; Bucknall 2012; Coultas 2005; Fan 2012; Gallefoss 1999; Hernández 2015; Jolly 2018; Jonsdottir 2015; Kessler 2018; Liang 2019; Rice 2010; Titova 2015; Walters 2013; Wang 2019), in which 15 comparisons between self‐management interventions versus usual care could be made, as two intervention groups from Coultas 2005 were included. The meta‐analysis showed lower mean SGRQ total scores (MD ‐2.86, 95% CI ‐4.87 to ‐0.85; low‐quality evidence; Analysis 1.1; Figure 3), indicating likely better HRQoL in the intervention group compared to the usual care group, with substantial heterogeneity (I² = 60%). The pooled MD of ‐2.86 did not reach the MCID of four points (Jones 2005). Five individual studies reached the MCID of four points for the SGRQ score (Bucknall 2012; Hernández 2015; Rice 2010; Titova 2015; Wang 2019). Sensitivity analyses using FEM resulted in a similar effect size (MD ‐3.03, 95% CI ‐4.18 to ‐1.88) compared to REM. Liang 2019 showed a discrepancy with regard to SGRQ results. Whereas the authors described better HRQoL for the self‐management intervention group after follow‐up, they presented SGRQ scores that were higher (meaning worse). Because our contact attempts (in which we asked for clarification) remained unanswered, we included the presented data (worse HRQoL) in our primary analysis and performed a sensitivity analysis on SGRQ adjusted total score without the Liang 2019 study, resulting in a higher MD of ‐3.25 (95% CI ‐5.27 to ‐1.23) with lower heterogeneity (I² = 58%) compared to the primary analysis. Sensitivity analyses using ICCs of 0.02 and 0.04 for the CRT of Walters 2013 resulted in similar effect sizes (MD ‐2.86, 95% CI ‐4.87 to ‐0.85).

1.1. Analysis.

Comparison 1: Self‐management versus usual care (primary outcomes), Outcome 1: Health‐related quality of life (HRQoL): adjusted SGRQ total score (primary analysis)

3.

Forest plot of comparison: self‐management versus usual care, outcome: 1.1 HRQoL: adjusted SGRQ total score (primary analysis)

For the meta‐analysis of short‐term (≤ 6 months' follow‐up) effects, seven studies could be included (including the Coultas study with two intervention groups)(Bourbeau 2003; Coultas 2005; Jolly 2018; Liang 2019; Titova 2015; Walters 2013; Wang 2019). No difference in SGRQ total score between self‐management interventions and usual care was detected (MD ‐2.36, 95% CI ‐4.92 to 0.20; I² = 54%; Analysis 1.2; Figure 4). For medium‐term (> 6 to ≤ 12 months' follow‐up) effects, 13 studies could be included in the meta‐analysis (Bourbeau 2003; Bucknall 2012; Fan 2012; Gallefoss 1999; Hernández 2015; Jolly 2018; Jonsdottir 2015; Kessler 2018; Liang 2019; Rice 2010; Titova 2015; Walters 2013; Wang 2019), which showed probably lower SGRQ total scores for self‐management interventions compared to usual care (MD ‐2.65, 95% CI ‐4.78 to ‐0.52; I² = 65%; Analysis 1.2; Figure 4). Analysis for long‐term (> 12 months' follow‐up) effects could not be performed, because of an insufficient number of studies (n < 3).

1.2. Analysis.

Comparison 1: Self‐management versus usual care (primary outcomes), Outcome 2: Health‐related quality of life (HRQoL): adjusted SGRQ total score (secondary analysis)

4.

Forest plot of comparison: Self‐management versus usual care, outcome: 1.2 HRQoL: adjusted SGRQ total score (secondary analysis)

Two studies reported insufficient data for inclusion in the SGRQ meta‐analyses (Martin 2004; Rose 2018). Rose 2018 could not be included due to lack of per‐group participant numbers at different time points for SGRQ total scores. This study reported no change in SGRQ scores at six and 12 months' follow‐up. Martin 2004 also found no difference in SGRQ total score after 12 months' follow‐up.

Chronic Respiratory Questionnaire (CRQ)

Five studies measured COPD‐specific HRQoL with the Chronic Respiratory Questionnaire (CRQ) (Benzo 2016; Bischoff 2012; Johnson‐Warrington 2016; Lenferink 2019; Mitchell 2014). The CRQ consists of four domain scores: dyspnoea, fatigue, emotional function and mastery (sense of control over the disease) (Guyatt 1987). A higher CRQ domain score indicates better HRQoL and the MCID is reflected by a change in a CRQ domain score of at least 0.5 on a 7‐point scale (Jaeschke 1989; Redelmeier 1996).

For primary analyses, mean domain end point scores of five studies, with a total of 738 participants, could be included in the meta‐analyses. Whereas these studies showed higher mean domain CRQ scores in the self‐management intervention compared to usual care, the evidence suggests that self‐management interventions do not improve CRQ domain scores, with MDs of 0.13 (95% CI ‐0.10 to 0.35), 0.12 (95% CI ‐0.09 to 0.33), 0.23 (95% CI –0.01 to 0.47) and 0.20 (95% CI ‐0.06 to 0.46) for dyspnoea, mastery, fatigue and emotional function, respectively (Analysis 1.3).

1.3. Analysis.

Comparison 1: Self‐management versus usual care (primary outcomes), Outcome 3: Health‐related quality of life (HRQoL): CRQ domain scores (primary analysis)

For the meta‐analysis of short‐term (≤ 6 months' follow‐up) effects, three studies could be included (Johnson‐Warrington 2016; Lenferink 2019; Mitchell 2014). Domain scores showed no difference between self‐management interventions and usual care, with MDs of 0.23 (95% CI ‐0.15 to 0.61), 0.12 (95% CI ‐0.33 to 0.57), 0.11 (95% CI ‐0.55 to 0.77) and 0.22 (95% CI ‐0.37 to 0.82) for dyspnoea, mastery, fatigue and emotional function, respectively (Analysis 1.4). For medium‐term (> 6 to ≤ 12 months' follow‐up) effects, again, three studies could be included in the meta‐analysis (Benzo 2016; Lenferink 2019; Mitchell 2014). Domain scores showed no difference between self‐management interventions and usual care, with MDs of 0.13 (95% CI ‐0.9 to 0.34), 0.17 (95% CI ‐0.07 to 0.41), 0.22 (95% CI ‐0.15 to 0.59) and 0.12 (95% CI ‐0.27 to 0.51) for dyspnoea, mastery, fatigue and emotional function, respectively (Analysis 1.4). Analysis for long‐term (> 12 months' follow‐up) effects could not be performed, because of an insufficient number of studies (n < 3).

1.4. Analysis.

Comparison 1: Self‐management versus usual care (primary outcomes), Outcome 4: Health‐related quality of life (HRQoL): CRQ domain scores (secondary analyses)

Clinical COPD Questionnaire (CCQ)

In two studies with a total of 170 participants (Ferrone 2019; Tabak 2014), COPD‐specific HRQoL was measured with the Clinical COPD Questionnaire (CCQ). No meta‐analysis could be performed. A lower score indicates better HRQoL and the MCID of the CCQ total score is reflected by a change in score of 0.4 or more on a 6‐point scale (Kocks 2006). Ferrone 2019 reported clinically relevant lower CCQ total scores in the COPD self‐management intervention group (mean 1.89; SD 1.07) compared to the usual care group (CCQ total mean 2.79; SD 1.28). Tabak 2014 reported no differences between the small study groups after three months' follow‐up (self‐management intervention (n = 12): mean 1.8, SE 0.24; usual care (n = 12): mean 2.3, SE 0.26).

Other HRQoL measures

Coultas 2005 and Walters 2013 used the Short Form‐36 (SF‐36) to measure generic HRQoL. Both studies reported no effects on SF‐36 domain scores for the intervention group compared to usual care. Fan 2012 used the SF‐12, a reduced version of the SF‐36, and reported no improvement on SF‐12 domain scores for participants in the self‐management intervention group compared to the usual care group who completed 12 months of study visits.

Bucknall 2012 and Tabak 2014 reported generic HRQoL using EuroQol‐5 Dimensions (EQ‐5D). Both Bucknall 2012 and Tabak 2014 reported no differences in the EQ‐5D areas under the curve between the groups after follow‐up.

In Tabak 2014, the individual participants' HRQoL state was also reported using a vertical visual analogue scale (VAS): self‐management intervention (72.3; SE 3.1) and usual care (62.4; SE 3.5). No statistical test was performed.

Generic HRQoL was measured using the Illness Intrusiveness Rating Scale (IIRS) in Coultas 2005 and Jonsdottir 2015, the Short Form‐12 (SF‐12) in Fan 2012, and the Sickness Impact Profile (SIP) in Emery 1998. Both Coultas 2005 and Jonsdottir 2015 reported beneficial effects of the self‐management intervention on the total IIRS. Coultas 2005 found lower participant‐reported illness intrusiveness in the nurse‐assisted collaborative management group compared to usual care at six months' follow‐up (mean change ‐7.0, 95% CI ‐15.0 to ‐0.5). Jonsdottir 2015 found less perceived intrusiveness of COPD and its treatment for participants in the self‐management intervention, demonstrated by IIRS total score (self‐management intervention: mean 31.57, SD 17.31; usual care: mean 27.84, SD 14.5; P = 0.014) at 12 months' follow‐up. Finally, Emery 1998 found improvement in total function in the control group as measured by the SIP (baseline: mean 14.2, SD 8.6; after a 10‐week intervention: mean 10.4, SD 7.8; P < 0.001), whereas the self‐management intervention group showed no change.

Respiratory‐related hospital admissions

Respiratory‐related hospital admissions were reported in 19 studies (Benzo 2016; Bösch 2007; Bourbeau 2003; Bucknall 2012; Coultas 2005; Fan 2012; Gallefoss 1999; Hernández 2015; Johnson‐Warrington 2016; Jolly 2018; Lenferink 2019; Martin 2004; Mitchell 2014; Rice 2010; Sanchez‐Nieto 2016; Tabak 2014; Titova 2015; Walters 2013; Wang 2019). Coultas 2005 had two intervention groups included in the meta‐analysis.

For primary analyses, regardless of length of follow‐up, 15 studies including 3263 participants, could be included in a meta‐analysis (Benzo 2016; Bourbeau 2003; Bucknall 2012; Coultas 2005; Fan 2012; Gallefoss 1999; Hernández 2015; Johnson‐Warrington 2016; Lenferink 2019; Mitchell 2014; Rice 2010; Sanchez‐Nieto 2016; Tabak 2014; Titova 2015; Walters 2013). A lower probability of at least one respiratory‐related hospital admission was noted amongst participants receiving the self‐management intervention compared with those who received usual care (OR 0.75, 95% CI 0.57 to 0.98; very low‐quality evidence; Analysis 1.5; Figure 5). Pooled study results showed moderate heterogeneity (I² = 48%). Sensitivity analyses using FEM resulted in a similar effect size (OR 0.74, 95% CI 0.63 to 0.88) compared to REM. Sensitivity analyses using ICCs of 0.02 and 0.04 for the Walters 2013 CRT resulted in similar effect sizes (OR 0.75, 95% CI 0.57 to 0.98).

1.5. Analysis.

Comparison 1: Self‐management versus usual care (primary outcomes), Outcome 5: Healthcare utilisation: respiratory‐related hospital admissions (number of participants with at least one admission) (primary analysis)

5.

Forest plot of comparison: self‐management versus usual care, outcome: 1.4 Healthcare utilisation: respiratory‐related hospital admissions (number of participants with at least one admission)

The meta‐analysis of short‐term (≤ 6 months' follow‐up) effects included four studies (Coultas 2005; Johnson‐Warrington 2016; Mitchell 2014; Tabak 2014), and showed no difference between self‐management interventions and usual care (OR 0.84, 95% CI 0.45 to 1.55; Analysis 1.6). For medium‐term (> 6 to ≤ 12 months' follow‐up) effects, 11 studies were included (Benzo 2016; Bourbeau 2003; Bucknall 2012; Fan 2012; Gallefoss 1999; Hernández 2015; Lenferink 2019; Rice 2010; Sanchez‐Nieto 2016; Titova 2015; Walters 2013). No difference between self‐management and usual care was detected (OR 0.74, 95% CI 0.53 to 1.03; Analysis 1.6). Analysis for long‐term (> 12 months' follow‐up) effects could not be performed due to an insufficient number of studies (n < 3).

1.6. Analysis.

Comparison 1: Self‐management versus usual care (primary outcomes), Outcome 6: Healthcare utilisation: respiratory‐related hospital admissions (number of participants with at least one admission) (secondary analyses)

The Martin 2004 study could not be included in the meta‐analysis due to lack of SDs. In this study, more respiratory‐related hospitalisations were found in the intervention group (1.1 per patient per year) compared to usual care (0.7 per patient per year).

The study‐specific NNTBs for respiratory‐related hospital admissions ranged from 15 (95% CI 8 to 399) to 26 (95% CI 15 to 677). To calculate NNTB, the pooled effect on respiratory‐related hospital admissions (OR 0.75, 95% CI 0.57 to 0.98) was used, and this was applied to the mean usual care event risks over the mean follow‐up duration of the study comparisons with the highest and lowest baseline risks. The eight comparisons with the highest baseline risks for respiratory‐related hospital admissions had a mean control event risk (mean observed risk of the respiratory‐related hospital admissions in the usual care group) of 48.6 (Figure 6) (Benzo 2016; Bourbeau 2003; Bucknall 2012; Johnson‐Warrington 2016; Lenferink 2019; Sanchez‐Nieto 2016; Tabak 2014; Titova 2015). Over a mean of 9.75 months' follow‐up, 15 participants (95% CI 8 to 399) with high baseline risk of respiratory‐related hospital admissions needed to be treated to prevent one person with at least one respiratory‐related hospital admission. The eight comparisons with the lowest baseline risks for respiratory‐related hospital admissions had a mean usual care event risk of 17.6 (Figure 7) (Coultas 2005 (with two intervention groups); Fan 2012; Gallefoss 1999; Hernández 2015; Mitchell 2014; Rice 2010; Walters 2013). Over a mean of 9.75 months' follow‐up, 26 participants (95% CI 15 to 677) with low baseline risk of respiratory‐related hospital admissions needed to be treated to prevent one person with at least one respiratory‐related hospital admission.

6.

Cates plot of participants with COPD with high baseline risk of respiratory‐related hospital admissions. In the usual care group, 49 of 100 participants had at least one respiratory‐related hospital admission over a mean of 9.75 months, compared to 42 (95% CI 35 to 49) of 100 participants in the self‐management intervention group.

7.

Cates plot of participants with COPD with low baseline risk of respiratory‐related hospital admissions. In the usual care group, 18 of 100 participants had at least one respiratory‐related hospital admission over a mean of 9.75 months, compared to 14 (95% CI 11 to 18) of 100 participants in the self‐management intervention group.

Seven studies with 1572 participants were included in a meta‐analysis on the mean number of respiratory‐related hospital admissions (Bösch 2007; Bucknall 2012; Jolly 2018; Lenferink 2019; Tabak 2014; Titova 2015; Wang 2019). For primary analyses, regardless of length of follow‐up, no difference between self‐management interventions and usual care was found (MD ‐0.29, 95% CI ‐0.60 to 0.01; Analysis 1.7). Using FEM in the sensitivity analysis produced similar effects (MD ‐0.00, 95% CI ‐0.02 to 0.01).

1.7. Analysis.

Comparison 1: Self‐management versus usual care (primary outcomes), Outcome 7: Healthcare utilisation: respiratory‐related hospital admissions (mean number per participant) (primary analysis)

The meta‐analysis of short‐term (≤ 6 months' follow‐up) effects included three studies (Jolly 2018; Tabak 2014; Wang 2019), and showed no difference (MD ‐0.01, 95% CI ‐0.04 to 0.02; Analysis 1.8). For medium‐term (> 6 to ≤ 12 months' follow‐up) effects, six studies were included (Bösch 2007; Bucknall 2012; Jolly 2018; Lenferink 2019; Titova 2015; Wang 2019). No difference was detected (MD ‐0.33, 95% CI ‐0.68 to 0.01; Analysis 1.8). Analysis for long‐term (> 12 months' follow‐up) effects could not be performed, due to an insufficient number of studies (n < 3).

1.8. Analysis.

Comparison 1: Self‐management versus usual care (primary outcomes), Outcome 8: Healthcare utilisation: respiratory‐related hospital admissions (mean number per participant) (secondary analyses)

Mortality

Mortality was reported as an outcome measure in nine studies (Bucknall 2012; Fan 2012; Johnson‐Warrington 2016; Kessler 2018; Lenferink 2019; Rice 2010; Rose 2018; Sanchez‐Nieto 2016; Titova 2015). In addition to formal mortality data, we extracted mortality data from sections describing the participant flow and reasons for losses to follow‐up from 15 studies (Benzo 2016; Bourbeau 2003; Bringsvor 2018; Coultas 2005; Ferrone 2019; Gallefoss 1999; Hernández 2015; Jolly 2018; Jonsdottir 2015; Liang 2019; Martin 2004; Mitchell 2014; Tabak 2014; Walters 2013; Wang 2019). Three studies provided no information on mortality (Bischoff 2012, 110 participants; Bösch 2007, 50 participants; Emery 1998, 49 participants), and could not be included in the meta‐analyses.

Respiratory‐related mortality

We included data from eight studies in the meta‐analysis of respiratory‐related mortality (Bucknall 2012; Fan 2012; Gallefoss 1999; Johnson‐Warrington 2016; Lenferink 2019; Tabak 2014; Titova 2015; Wang 2019). No difference in mortality risk was found between self‐management intervention and usual care groups (risk difference (RD) 0.01, 95% CI ‐0.02 to 0.04; I² = 63%; 1572 participants ; low‐quality evidence; Analysis 1.9; Figure 8). Three studies reported no deaths in the self‐management and usual care groups after 12 months' follow‐up (Gallefoss 1999; Wang 2019), and after three months' follow‐up (Tabak 2014). Sensitivity analysis using a FEM produced similar results on respiratory‐related mortality (RD 0.03, 95% CI 0.00 to 0.05). It was not possible to calculate the NNTB for respiratory‐related mortality because the 95% CI of the pooled RD for respiratory‐related mortality included the possibilities of both benefit and harm.

1.9. Analysis.

Comparison 1: Self‐management versus usual care (primary outcomes), Outcome 9: Mortality: respiratory‐related mortality (primary analysis)

8.

Forest plot of comparison: self‐management versus usual care, outcome: 1.6 Mortality: respiratory‐related mortality)

For the meta‐analysis of short‐term (≤ 6 months' follow‐up) effects, four studies could be included (Gallefoss 1999; Johnson‐Warrington 2016; Tabak 2014; Wang 2019). No difference was found between self‐management interventions and usual care (RD ‐0.01, 95% CI ‐0.05 to 0.03; Analysis 1.10). For medium‐term (> 6 to ≤ 12 months' follow‐up) effects, seven studies could be included in the meta‐analysis (Bucknall 2012; Fan 2012; Gallefoss 1999; Kessler 2018; Lenferink 2019; Tabak 2014; Wang 2019). No difference between self‐management interventions and usual care was detected (RD 0.00, 95% CI ‐0.01 to 0.02; Analysis 1.10). Analysis for long‐term (> 12 months' follow‐up) effects could not be performed, due to an insufficient number of studies (n < 3).

1.10. Analysis.

Comparison 1: Self‐management versus usual care (primary outcomes), Outcome 10: Mortality: respiratory‐related mortality (secondary analyses)

All‐cause mortality

We included data from 24 studies with 5719 participants in the meta‐analysis for all‐cause mortality (Coultas 2005 had two intervention groups) (Benzo 2016; Bourbeau 2003; Bringsvor 2018; Bucknall 2012; Coultas 2005; Fan 2012; Ferrone 2019; Gallefoss 1999; Hernández 2015; Johnson‐Warrington 2016; Jolly 2018; Jonsdottir 2015; Kessler 2018; Lenferink 2019; Liang 2019; Martin 2004; Mitchell 2014; Rice 2010; Rose 2018; Sanchez‐Nieto 2016; Tabak 2014; Titova 2015; Walters 2013; Wang 2019). No difference in mortality risk was found between self‐management intervention and usual care group participants (RD ‐0.01, 95% CI ‐0.03 to 0.01; I² = 63%; low‐quality evidence; Analysis 1.11; Figure 9). Three studies reported no deaths in the self‐management and usual care groups (Gallefoss 1999; Tabak 2014; Wang 2019). Sensitivity analysis using a FEM resulted in a similar result on all‐cause mortality (RD ‐0.01, 95% CI ‐0.02 to 0.01). It was not possible to calculate the NNTB for all‐cause mortality, because the 95% CI of the pooled RD for all‐cause mortality included the possibilities of both benefit and harm. Sensitivity analyses using ICCs of 0.02 and 0.04 for the Walters 2013 CRT resulted in similar effect sizes (RD ‐0.01, 95% CI ‐0.03 to 0.01).

1.11. Analysis.

Comparison 1: Self‐management versus usual care (primary outcomes), Outcome 11: Mortality: all‐cause mortality (primary analysis)

9.

Forest plot of comparison: self‐management versus usual care, outcome: 1.7 Mortality: all‐cause mortality

For the meta‐analysis of short‐term (≤ 6 months' follow‐up) effects, nine studies could be included (Coultas 2005; Gallefoss 1999; Johnson‐Warrington 2016; Jolly 2018; Jonsdottir 2015; Liang 2019; Mitchell 2014; Tabak 2014; Wang 2019). No difference between self‐management interventions and usual care was detected (RD ‐0.00, 95% CI ‐0.01 to 0.01; Analysis 1.12). For medium‐term (> 6 to ≤ 12 months' follow‐up) effects, 21 studies could be included in the meta‐analyses (Benzo 2016; Bourbeau 2003; Bringsvor 2018; Bucknall 2012; Fan 2012; Ferrone 2019; Gallefoss 1999; Hernández 2015; Jolly 2018; Jonsdottir 2015; Kessler 2018; Lenferink 2019; Liang 2019; Martin 2004; Rice 2010; Rose 2018; Sanchez‐Nieto 2016; Tabak 2014; Titova 2015; Walters 2013; Wang 2019). Again, no difference between self‐management interventions and usual care was found (RD ‐0.01, 95% CI ‐0.02 to 0.01; Analysis 1.12). Analysis for long‐term (> 12 months' follow‐up) effects could not be performed, because an insufficient number of studies including outcomes of interest (n < 3) was available.

1.12. Analysis.

Comparison 1: Self‐management versus usual care (primary outcomes), Outcome 12: Mortality: all‐cause mortality (secondary analyses)

All‐cause hospital admissions

Ten studies, with a total of 2633 participants, were included in a meta‐analysis for number of participants with at least one all‐cause hospital admission (Benzo 2016; Bucknall 2012; Coultas 2005; Fan 2012; Hernández 2015; Johnson‐Warrington 2016; Lenferink 2019; Mitchell 2014; Rice 2010; Tabak 2014; the Coultas 2005 study had two intervention groups). There was no between‐group difference in all‐cause hospital admissions (OR 0.88, 95% CI 0.71 to 1.08; moderate‐quality evidence; Analysis 2.1). Heterogeneity was low (I² = 20%). Sensitivity analysis using FEM showed similar results (OR 0.85, 95% CI 0.72 to 1.01).

2.1. Analysis.

Comparison 2: Self‐management versus usual care (secondary outcomes), Outcome 1: Healthcare utilisation: all‐cause hospital admissions (number of participants with at least one admission)

Four studies could not be included in this meta‐analysis due to lack of required information (Bösch 2007; Kessler 2018; Martin 2004; Titova 2015). Two of these studies suggested a reduction in total number of all‐cause hospital admissions, favouring the self‐management intervention compared to usual care (Bösch 2007; Titova 2015). The other two studies reported no difference in all‐cause hospital admissions compared to usual care (Kessler 2018; Martin 2004). It was not possible to calculate the NNTB for all‐cause hospital admissions because the 95% CI of the pooled OR for at least one all‐cause hospital admission included the possibilities of both benefit and harm.

Seven studies reported the mean number of all‐cause hospital admissions (Bucknall 2012; Ferrone 2019; Jolly 2018; Lenferink 2019; Martin 2004; Rose 2018; Tabak 2014). No difference in this mean number was found (MD ‐0.01, 95% CI ‐0.06 to 0.04; Analysis 2.2). Sensitivity analysis using FEM showed similar results (MD ‐0.01, 95% CI ‐0.06 to 0.04). The Jolly 2018 study dominated the overall effect with a weight of 87% due to extremely small SEs and CIs. However, sensitivity analysis excluding Jolly 2018 showed similar results on all‐cause hospital admissions (MD ‐0.05, 95% CI ‐0.19 to 0.08). The Bourbeau 2003 study could not be included in the meta‐analysis because SDs were missing.

2.2. Analysis.

Comparison 2: Self‐management versus usual care (secondary outcomes), Outcome 2: Healthcare utilisation: all‐cause hospital admissions (mean number per participant)

Healthcare utilisation

Respiratory‐related hospitalisation days

Five studies assessed the number of respiratory‐related hospitalisation days per participant (Benzo 2016; Johnson‐Warrington 2016; Kessler 2018; Lenferink 2019; Sanchez‐Nieto 2016). Four of these, including 819 participants, could be included in the meta‐analysis (Benzo 2016; Kessler 2018; Lenferink 2019; Sanchez‐Nieto 2016). No differences were found between the self‐management intervention and usual care groups (MD ‐0.62, 95% CI ‐2.27 to 1.03; Analysis 2.3). Heterogeneity was substantial (I² = 68%). Sensitivity analyses using a FEM showed similar results (MD ‐0.57, 95% CI ‐1.18 to 0.05). The Johnson‐Warrington 2016 study could not be included in the meta‐analysis because insufficient data were reported. The difference in the median number of respiratory‐related hospitalisation days in this study was not different between the self‐management intervention and usual care group (median 12.0, IQR 9.0 to 33.8 versus median 15.0, IQR 3.5 to 32.0).

2.3. Analysis.

Comparison 2: Self‐management versus usual care (secondary outcomes), Outcome 3: Healthcare utilisation: respiratory‐related hospitalisation days (per participant)

All‐cause hospitalisation days

Eight studies assessed the number of all‐cause hospitalisation days per participant (Bourbeau 2003; Bucknall 2012; Hernández 2015; Johnson‐Warrington 2016; Kessler 2018; Lenferink 2019; Rice 2010; Rose 2018). Six studies with 2073 participants could be included in the meta‐analysis (Bourbeau 2003; Bucknall 2012; Hernández 2015; Kessler 2018; Lenferink 2019; Rice 2010). No between‐group differences were found (MD ‐0.51, 95% CI ‐1.85 to 0.84; Analysis 2.4). Heterogeneity was moderate (I² = 43%). Sensitivity analysis using a FEM showed similar results (MD ‐0.59, 95% CI ‐1.29 to 0.12). Two studies could not be included in the meta‐analysis, because insufficient data were reported (Johnson‐Warrington 2016; Rose 2018). The Johnson‐Warrington 2016 study reported lower median number of all‐cause bed days for readmission in the self‐management group (9.0, IQR 1.0 to 30.0) compared to the usual care group (16.5, IQR 3.8 to 39.8); however, the CI includes no difference. The Rose 2018 study reported lower median number of all‐cause bed days in the intervention group (8.0, IQR 4.0 to 22.0) compared to the usual care group (11.0, IQR 4.0 to 15.0), with a difference in hospitalisation day risk ratio of 0.84 (95% CI 0.78 to 0.90), favouring self‐management for those at risk.

2.4. Analysis.

Comparison 2: Self‐management versus usual care (secondary outcomes), Outcome 4: Healthcare utilisation: all‐cause hospitalisation days (per participant)

Emergency department (ED) visits

Thirteen studies reported ED visits (Bourbeau 2003; Bucknall 2012; Coultas 2005; Fan 2012; Ferrone 2019; Hernández 2015; Jolly 2018; Mitchell 2014; Rice 2010; Rose 2018; Sanchez‐Nieto 2016; Tabak 2014; Wang 2019). Five studies with 865 participants were included in a meta‐analysis for number of participants with at least one ED visit (Fan 2012; Ferrone 2019; Mitchell 2014; Sanchez‐Nieto 2016; Tabak 2014). Self‐management intervention participants may have a slightly lower probability of at least one ED visit compared to usual care participants (OR 0.53, 95% CI 0.32 to 0.87; low‐quality evidence; Analysis 2.5), with a low heterogeneity (I² = 34%). Sensitivity analysis using a FEM produced a similar result (OR 0.59, 95% CI 0.43 to 0.81).

2.5. Analysis.

Comparison 2: Self‐management versus usual care (secondary outcomes), Outcome 5: Healthcare utilisation: emergency department visits (number of participants with at least one visit)

Six studies with 1939 participants could be included in a meta‐analysis for the mean number of ED visits per participant (Bourbeau 2003; Bucknall 2012; Ferrone 2019; Jolly 2018; Rose 2018; Wang 2019). The meta‐analysis showed a slightly lower risk of ED visits for self‐management intervention participants compared to usual care participants (MD 0.52, 95% CI 0.89 to 0.15; Analysis 2.6), with considerable heterogeneity (I² = 96%). Sensitivity analysis using a FEM showed a similar result (MD ‐0.32, 95% CI ‐0.38 to ‐0.25). The outlier in this meta‐analysis was the study by Wang 2019, showing a very strong effect size favouring self‐management (MD ‐1.80, 95% CI ‐2.18 to ‐1.42). Sensitivity analysis without the Wang 2019 study showed no difference in risk of ED visits (MD ‐0.28, 95% CI ‐0.60 to 0.03).

2.6. Analysis.

Comparison 2: Self‐management versus usual care (secondary outcomes), Outcome 6: Healthcare utilisation: emergency department visits (mean number per participant)

Three studies could not be included in the meta‐analysis, because different methods were used to report the outcome (Coultas 2005; Hernández 2015; Rice 2010). Coultas 2005 reported respiratory‐related ED visits and reported no differences between groups in COPD‐related visits after six months' follow‐up. Hernández 2015 reported a lower mean number of respiratory‐related ED visits for self‐management intervention participants (mean 10, SD 12.11) compared to usual care participants (mean 23, SD 27.4). These data could not be included in a meta‐analysis because there were different approaches for co‐ordination of hospital admissions between study groups. Eighty percent of the admissions in the self‐management intervention group were co‐ordinated between primary care and the hospital team, and therefore bypassing the ED (Hernández 2015). All admissions in the usual care group were processed as unplanned through the ED (Hernández 2015). Rice 2010 found fewer visits in the self‐management intervention group compared to the usual care group (67.0 versus 91.2 per 100 person‐years, P = 0.02).

General practitioner (GP) visits

Seven studies reported GP visits (Bischoff 2012; Bourbeau 2003; Bucknall 2012; Gallefoss 1999; Jolly 2018; Martin 2004; Mitchell 2014). Four studies with 1113 participants were included in a meta‐analysis (Bucknall 2012; Gallefoss 1999; Jolly 2018; Martin 2004). No difference between the self‐management intervention and usual care groups was found (MD ‐0.21, 95% CI ‐0.86 to 0.25; Analysis 2.7). Sensitivity analyses using a FEM resulted in similar results (MD ‐0.11, 95% CI ‐0.25 to 0.02). Three studies could not be included in the meta‐analysis either because different methods were used to report the outcome (Bischoff 2012; Mitchell 2014), or because of missing SDs (Bourbeau 2003). Bourbeau 2003 reported fewer unscheduled GP visits for the self‐management intervention group compared to the usual care group. Nevertheless, the scheduled GP visits were comparable between the groups. CIs of group differences in the Bischoff 2012 study (self‐management intervention: n = 20, usual care: n = 18; OR 1.09, 95% CI 0.42 to 2.81) and the Mitchell 2014 study (self‐management intervention: n = 78, usual care: n = 73; OR 1.15, 95% CI 0.75 to 1.76) did not find a difference.

2.7. Analysis.

Comparison 2: Self‐management versus usual care (secondary outcomes), Outcome 7: Healthcare utilisation: GP visits (mean number per participant)

Specialist visits

Five studies reported data on specialist visits (Bourbeau 2003; Coultas 2005; Ferrone 2019; Martin 2004; Mitchell 2014). No meta‐analysis could be performed because different methods and definitions were used to report these visits. Bourbeau 2003 reported comparable unscheduled (self‐management intervention n = 24, usual care n = 26) and scheduled specialist visits (self‐management intervention n = 347, usual care n = 316) in both groups. Coultas 2005 reported no change in between‐group difference from baseline to six months' follow‐up (nurse‐assisted medical management versus usual care: mean change 1.3, 95% CI ‐1.5 to 4.1; nurse‐assisted collaborative management versus usual care: mean change 0.4, 95% CI ‐2.2 to 3.4).

Ferrone 2019 reported a lower number of urgent physician visits for the self‐management group compared to the usual care group after 12 months' follow‐up (between‐group difference: 1.46, 95% CI 0.90 to 2.02, P < 0.001). Martin 2004 reported a non‐significant higher number of all‐cause doctor and nurse visits in the self‐management intervention group compared to the usual care group (mean 15.6, SD 12.68 versus mean 11.6, SD 8.02). Mitchell 2014 observed a reduction in the number of nurse specialist home visits for respiratory reasons in the self‐management intervention compared to usual care group (OR 0.42, 95% CI 0.19 to 0.91).

Number of COPD exacerbations

A meta‐analysis, including 1401 participants from seven studies, of the mean number of exacerbations per participant (regardless of definition used) (Benzo 2016; Bischoff 2012; Bösch 2007; Fan 2012; Jonsdottir 2015; Kessler 2018; Lenferink 2019), resulted in lower mean exacerbations per participant for the self‐management intervention, but the CI includes no difference (MD ‐0.06, 95% CI ‐0.26 to 0.15; Analysis 2.8). Sensitivity analysis using a FEM produced similar results (MD ‐0.06, 95% CI ‐0.26 to 0.15).

2.8. Analysis.

Comparison 2: Self‐management versus usual care (secondary outcomes), Outcome 8: COPD exacerbations (mean number per participant)

Four studies reported COPD exacerbations based on symptoms (Bischoff 2012; Fan 2012; Kessler 2018; Lenferink 2019). Data from these studies were included in a meta‐analysis, with 1047 participants, resulting in no difference in mean COPD exacerbations between self‐management intervention and usual care (MD 0.05, 95% CI ‐0.22 to 0.31; Analysis 2.8), with low heterogeneity (I² = 0%).

Five studies reported the total number of exacerbations (Bischoff 2012; Bourbeau 2003; Fan 2012; Lenferink 2019; Tabak 2014). Bischoff 2012 reported 280 exacerbations in the self‐management intervention group (n = 55) and 235 in the usual care group (n = 55), with no between‐group difference (first year follow‐up rate ratio 1.10, 95% CI 0.86 to 1.40; second year follow‐up rate ratio 1.16, 95% CI 0.81 to 1.67). Bourbeau 2003 reported a difference in exacerbation rates between both groups, with 299 exacerbations in the self‐management intervention group (n = 96) and 362 exacerbations in the usual care group (n = 95) after 12 months' follow‐up (P = 0.06). Fan 2012 reported 600 self‐reported exacerbations in the self‐management intervention group (n = 209) and 610 in the usual care group (n = 217), with no between‐group difference during the first 12 months' follow‐up (rate ratio 1.03, 95% CI 0.97 to 1.10). Tabak 2014 reported 33 (median 2.0, IQR 1.0 to 3.0) exacerbations in the self‐management intervention group (n = 12); exacerbation data for the usual care group was not available. Lenferink 2019 reported 216 exacerbations in the self‐management intervention group (n = 102) and 230 exacerbations in the usual care group (n = 99), extracted from diary data, after 12 months' follow‐up. Whereas no difference in COPD exacerbation rates was found between both groups (rate per 100 person‐years 0.91, 95% CI 0.65 to 1.26), a shorter duration per COPD exacerbation was observed by Lenferink 2019 for the self‐management intervention group (median days 8.1, IQR 4.8 to 10.1) compared to the usual care group (median days 9.5, IQR 7.0 to 15.1) (P = 0.021).

Use of oral steroids and antibiotics

The majority of the studies (n = 16) did not report any data on the use of oral steroids or antibiotics and thus could not be included in a meta‐analysis (Bourbeau 2003; Bringsvor 2018; Bucknall 2012; Coultas 2005; Emery 1998; Hernández 2015; Johnson‐Warrington 2016; Jolly 2018; Jonsdottir 2015; Kessler 2018; Liang 2019; Rose 2018; Tabak 2014; Titova 2015; Walters 2013; Wang 2019). Two studies reported data on combined use of oral steroids and antibiotics (Benzo 2016; Bischoff 2012). Bischoff 2012 reported a similar number of participants who started prednisolone, antibiotics or both, to manage exacerbations in the self‐management intervention group (n = 16, 11%) compared to the usual care group (n = 13, 10%) in the first year of follow‐up. In the second year of follow‐up, a higher number of exacerbations in the self‐management intervention group were managed by starting prednisolone, antibiotics or both (OR 3.98, 95% CI 1.10 to 15.58). Benzo 2016 reported no difference between groups in the use of antibiotic‐prednisone combination at three, six and nine months after discharge. There was greater use of the written action plan from nine to 12 months, and therefore a greater use of antibiotic‐prednisone combination in the self‐management intervention group (n = 54, 65.9%) compared to the usual care group (n = 34, 43.0%) (P = 0.004).

Courses of oral steroids

Six studies reported the use of oral steroids (Fan 2012; Gallefoss 1999; Lenferink 2019; Martin 2004; Rice 2010; Sanchez‐Nieto 2016). The numbers of participants who used at least one course of oral steroids were available for three studies (Gallefoss 1999; Rice 2010; Sanchez‐Nieto 2016). Data from these studies were included in a meta‐analysis, with 881 participants, resulting in no difference in oral steroid use (OR 4.19, 95% CI 0.35 to 50.65; Analysis 2.9), with considerable heterogeneity (I² = 96%). Sensitivity analysis using a FEM resulted in a higher probability of using at least one course of oral steroids for self‐management interventions with a smaller CI (OR 8.98, 95% CI 5.95 to 13.56; I² = 96%), likely due to less beneficial small‐study effects. The Rice 2010 study included a very large population compared to Gallefoss 1999 and Sanchez‐Nieto 2016. The proportion of participants who received at least one course of oral steroids in the self‐management intervention group was relatively high (97.6%) compared to other studies (Gallefoss 1999 = 69.2%; Sanchez‐Nieto 2016 = 38.3%). The OR in Rice 2010 was 32.7 which is probably an overestimation of the risk ratio due to the fact that the event is common. An additional sensitivity analysis without the Rice 2010 study was not possible because of the limited number of studies. This meta‐analysis should therefore be interpreted with caution.

2.9. Analysis.

Comparison 2: Self‐management versus usual care (secondary outcomes), Outcome 9: Courses of oral steriods (number of participants who used at least one course)

Fan 2012 reported a higher mean of 2.5 exacerbations per patient‐year treated with prednisolone in the self‐management intervention group compared with 2.1 in the usual care group (rate ratio 1.25, 95% CI 1.05 to 1.48). In Martin 2004, no difference between both study groups was detected (self‐management intervention: 2.3 courses, 95% CI 1.4 to 3.2; usual care: 1.3 courses, 95% CI 0.8 to 1.8). Lenferink 2019 reported that a higher number of self‐management intervention participants (n = 34, 51.5%) than usual care participants (n = 23, 32.9%) with a COPD exacerbation initiated a course of oral prednisolone within two days from the COPD exacerbation start in at least 75% of the exacerbations.

Courses of antibiotics

Seven studies reported the use of antibiotics (Bösch 2007; Fan 2012; Lenferink 2019; Martin 2004; Mitchell 2014; Rice 2010; Sanchez‐Nieto 2016). Data regarding participants who used at least one course of antibiotics were available in three studies (Mitchell 2014; Rice 2010; Sanchez‐Nieto 2016). All three studies, with a total of 1012 participants, could be included in a meta‐analysis. Results show higher use of antibiotics in the self‐management group compared to usual care (OR 3.95, 95% CI 1.37 to 11.43; Analysis 2.10), with considerable heterogeneity (I² = 85%). Sensitivity analysis using a FEM also resulted in a higher probability of participants in the self‐management intervention group using at least one course of antibiotics for the self‐management interventions (OR 5.88, 95% CI 4.19 to 8.25) with similar considerable heterogeneity (I² = 85%). As with oral steroids, Rice 2010 reported much higher rates of antibiotic use in the self‐management intervention group compared to usual care.

2.10. Analysis.

Comparison 2: Self‐management versus usual care (secondary outcomes), Outcome 10: Courses of antibiotics (number of participants who used at least one course)

Bösch 2007 reported a reduction in the mean number of exacerbations that were treated with antibiotics in the self‐management intervention group (mean exacerbations 2.0, SD 1.4 to mean exacerbations 1.4; SD 1.6), with no changes observed in the usual care group. Fan 2012 reported a slightly higher mean of 2.7 exacerbations per patient‐year treated with an antibiotic in the self‐management intervention group compared with a mean of 2.5 in the usual care group, but the CI indicates no difference (rate ratio 1.11; 95% CI 0.97 to 1.27). In Martin 2004, there was no difference in the use of antibiotics between the groups after 12 months' follow‐up (self‐management intervention: 3.6, 95% CI 2.5 to 4.7, versus usual care: 2.5, 95% CI 1.7 to 3.3). Lenferink 2019 reported no difference in the number of patient‐reported antibiotics between the self‐management intervention group (rate per 100 person‐years: 1.3, n = 102) and usual care group (rate per 100 person‐years: 1.2, n = 99) (incidence rate ratio 1.08, 95% CI 0.70 to 1.67).

Health status

COPD assessment test (CAT)

Two studies reported data on the impact of COPD on a person’s life, measured with the CAT (Ferrone 2019; Lenferink 2019). Ferrone 2019 found an improvement in CAT score for the self‐management intervention group compared to usual care group after 12 months' follow‐up (adjusted difference: 9.3, 95% CI 7.8 to 10.8). Lenferink 2019 reported no between‐group differences in CAT score for self‐management interventions compared to usual care.

Dyspnoea

Five studies assessed the effect of self‐management interventions on dyspnoea as measured by the modified Medical Research Council questionnaire (mMRC) (Bösch 2007; Hernández 2015; Jolly 2018; Lenferink 2019; Liang 2019). Three studies, representing 356 participants, were included in a meta‐analysis (Bösch 2007; Hernández 2015; Lenferink 2019). No difference in dyspnoea scores was found (MD ‐0.31, 95% CI ‐1.23 to 0.60; Analysis 2.11). Sensitivity analyses using a FEM resulted in a lower effect on dyspnoea score for self‐management interventions, but the CI includes no difference (MD ‐0.04, 95% CI ‐0.25 to 0.17). Two studies could not be included in a meta‐analysis, because different methods were used to report the outcome (Jolly 2018; Liang 2019). Jolly 2018 reported no differences in the level of breathlessness for self‐management interventions compared to usual care (OR 1.1, 95% CI 0.7 to 1.5). Liang 2019 reported non‐significant median change from baseline differences between groups in mMRC grades (self‐management intervention: median 1 (IQR 1 to 2); usual care: median 1 (IQR 0 to 2); P = 0.74).

2.11. Analysis.

Comparison 2: Self‐management versus usual care (secondary outcomes), Outcome 11: Health status: modified Medical Research Council Dyspnoea Scale (mMRC)

Anxiety and depression

Thirteen studies assessed the effect of self‐management interventions on anxiety and depression, as measured by the Hospital Anxiety and Depression Scale (HADS)(Zigmond 1983), a 21‐unit scale in which higher scores indicate more severe symptoms (Bucknall 2012; Emery 1998; Hernández 2015; Johnson‐Warrington 2016; Jolly 2018; Jonsdottir 2015; Kessler 2018; Lenferink 2019; Liang 2019; Mitchell 2014; Rose 2018; Titova 2015; Walters 2013). Nine studies could be included in a meta‐analysis, with 1647 participants for the HADS–anxiety score, and 1653 participants for the HADS–depression score (Bucknall 2012; Hernández 2015; Johnson‐Warrington 2016; Jolly 2018; Jonsdottir 2015; Lenferink 2019; Mitchell 2014; Titova 2015; Walters 2013). The meta‐analyses showed probably better mean HADS–anxiety scores (MD ‐0.57, 95% CI ‐1.01 to ‐0.13; moderate‐quality evidence) and probably better mean HADS–depression scores (MD ‐0.45, 95% CI ‐0.80 to ‐0.10; moderate‐quality evidence) for the self‐management intervention compared to usual care (Analysis 2.12); heterogeneity was low for both HADS analyses (I² = 21% and 6%, respectively). Sensitivity analyses using a FEM produced similar results (HADS–anxiety score MD ‐0.57, 95% CI ‐0.94 to ‐0.20; HADS–depression score MD ‐0.45, 95% CI ‐0.79 to ‐0.12).

2.12. Analysis.

Comparison 2: Self‐management versus usual care (secondary outcomes), Outcome 12: Health status: Hospital Anxiety and Depression Scale (HADS)

Four studies could not be included in meta‐analyses on anxiety and depression because insufficient data were reported by authors or they did not use the HADS to assess the outcome (Emery 1998; Kessler 2018; Liang 2019; Rose 2018). Kessler 2018 and Liang 2019 reported no difference for HADS scores between the self‐management intervention and usual care groups (Kessler 2018 adjusted total score MD 0.2, 95% CI ‐0.3 to 0.7; Liang 2019 self‐management intervention: anxiety median 0 (IQR 0 to 4) and depression median 1 (IQR 0 to 3.25); usual care: anxiety median 0 (IQR 0 to 3) and depression median 0 (IQR 0 to 1.25). Rose 2018 could not be included in the meta‐analysis due to lack of per‐group participant numbers for both HADS anxiety and depression scores. They found no evidence that the self‐management intervention changed HADS scores at 6 and 12 months' follow‐up (no effects reported). Emery 1998 used different units of measurement to assess anxiety and depression. Anxiety was assessed by the anxiety subscales of the State‐Trait Anxiety Inventory (STAI) and the Hopkins Symptom Checklist. Depression was assessed by the Center for Epidemiological Studies‐Depression inventory (CES‐D), the depression subscale of the Hopkins Symptom Checklist and the Bradburn Affect‐Balance Scale. No differences in anxiety and depression were found between education and stress management (ESM) and waiting list groups (usual care) after 10 weeks' follow‐up.

Self‐efficacy

Nine studies reported data on self‐efficacy (Bischoff 2012; Bringsvor 2018; Bucknall 2012; Fan 2012; Johnson‐Warrington 2016; Jolly 2018; Lenferink 2019; Mitchell 2014; Walters 2013), but no meta‐analysis could be performed because of insufficient data and the use of different outcome measures. Three of these studies measured self‐efficacy by using the COPD Self‐Efficacy Scale (CSES). Bischoff 2012 reported differences in participants’ self‐efficacy between the intervention and control groups according to the CSES total (MD ‐0.17, 95% CI ‐0.64 to 0.30) and domain scores after 24 months' follow‐up. Bucknall 2012 also reported lower CSES total scores in the self‐management intervention group, but the CI indicates no difference (MD 2.65, 95% CI ‐5.85 to 11.14). Lenferink 2019 reported a reduction in the behavioural risk factors domain of the CSES (MD ‐0.26, 95% CI ‐0.52 to ‐0.01). Bringsvor 2018 measured self‐efficacy by using the General Self‐Efficacy Scale (GSE), but reported no mean change differences (P = 0.18). Fan 2012 measured participants’ self‐efficacy by a self‐developed 8‐item questionnaire and reported an improvement in the self‐management intervention group after 12 months' follow‐up (MD 0.65, 95% CI 0.02 to 1.29). In Johnson‐Warrington 2016 and Mitchell 2014, self‐efficacy was assessed with the Pulmonary Rehabilitation Adapted Index of Self‐Efficacy (PRAISE). Johnson‐Warrington 2016 found no difference between change in self‐efficacy in groups (self‐management intervention: mean change 0.54, SD 9.48; usual care: mean change 2.34, SD 8.73). Mitchell 2014 also reported no between‐group difference in PRAISE score (MD 1.47, 95% CI ‐0.65 to 3.60; P = 0.21). Jolly 2018 measured self‐efficacy with the Stanford self‐efficacy score, but found no difference between the self‐management intervention group compared to the usual care group after 12 months' follow‐up (MD 0.1, 95% CI ‐0.1 to 0.4). Walters 2013 assessed self‐efficacy with the Self‐Efficacy for Managing Chronic Disease (SE MCD); no improvement in self‐efficacy was reported (self‐management intervention versus usual care: β coefficient 0.41, 95% CI ‐0.56 to 1.37).

Days lost from work

Only Gallefoss 1999 reported days lost from work. No differences between groups were observed. Almost 50% of the participants with COPD in this study were employed. Only three of 14 (21%) participants in the self‐management intervention group and two of 13 (15%) in the usual care group reported absence from work.

Exercise capacity and physical activity

Six studies, with 772 participants, measured exercise capacity using the six‐minute walking test (6MWT) and could be included in the meta‐analysis (Bösch 2007; Bourbeau 2003; Hernández 2015; Kessler 2018; Tabak 2014; Wang 2019). Compared to usual care, self‐management interventions may result in an improvement in exercise capacity, with an MD of 45.14 meters (95% CI 9.16 to 81.13; Analysis 2.13). Heterogeneity was considerable (I² = 86%). Sensitivity analysis using a FEM resulted in similar effects (MD 46.57, 95% CI 33.73 to 59.41; I² = 86%). The pooled MD of 45.14 meters reached the MCID of 25 meters and therefore is considered clinically relevant (Holland 2010). Two studies – Tabak 2014 and Wang 2019 – found a very high mean distance walked for the self‐management intervention group compared to usual care (Tabak 2014: MD 99.66, 95% CI 66.45 to 132.75; Wang 2019: MD 89.00, 95% CI 63.36 to 114.64). The large variation in effect sizes seems to be the main contributor to the considerable heterogeneity in this analysis.

2.13. Analysis.

Comparison 2: Self‐management versus usual care (secondary outcomes), Outcome 13: Exercise capacity: six‐minute walk test (6MWT)

Two studies assessed exercise capacity with both the Incremental Shuttle Walk Test (ISWT) and Endurance Shuttle Walk Test (ESWT) (Johnson‐Warrington 2016; Mitchell 2014). In Johnson‐Warrington 2016, no differences were found between the self‐management intervention and usual care groups for both ISWT (self‐management intervention: mean change (metres (m)) 45, 95% CI 0 to 70; usual care: mean change (m) 30, 95% CI 0 to 95) and ESWT (self‐management intervention: mean change (seconds) 178.5, 95% CI ‐3.75 to 443.50; usual care: mean change (seconds) 155, 95% CI 21 to 618.50). Mitchell 2014 reported a between‐group difference in the change in distance walked on the ISWT at six weeks' follow‐up, but there were no differences reported at six months' follow‐up. Furthermore, the ESWT time improved in the self‐management intervention group compared to the usual care group at six weeks and was maintained between six weeks and six months.

Jonsdottir 2015 assessed self‐reported physical activity of different intensities (walking, moderate intensity, and vigorous intensity) using the International Physical Activity Questionnaire short version (IPAQ). Jonsdottir 2015 reported a higher score on the subscale ‘vigorous’ for the self‐management intervention group after 12 months' follow‐up (P = 0.02). Furthermore, they reported a lower score on the subscale ‘walking’ with time in both groups (P = 0.02).

Self‐management behaviour

Three studies reported data on self‐management behaviour (Bringsvor 2018; Lenferink 2019; Walters 2013). However, a meta‐analysis was not possible, because different outcome measures were used. Two of these studies measured self‐management behaviour and knowledge using the Partners in Health scale (PIH) (Lenferink 2019; Walters 2013). Walters 2013 reported an interaction of treatment group by time for the overall PIH score (β coefficient 0.15, 95% CI 0.03 to 0.29) and for the PIH knowledge domain (β coefficient 0.25, 95% CI 0.00 to 0.50). Furthermore, an increase over time in both groups for the PIH coping domain was observed (β coefficient 0.15, 95% CI 0.04 to 0.26). Lenferink 2019 reported no between‐group differences for the overall PIH score and PIH domain scores between self‐management intervention and usual care groups (between‐group difference: 0.28, 95% CI ‐2.43 to 3.00). Bringsvor 2018 used the 'Health education impact Questionnaire' (HeiQ 2) to measure eight self‐management domains: 1) Positive and active engagement in life; 2) Health‐directed activities; 3) Skill and technique acquisition; 4) Constructive attitudes and approaches; 5) Self‐monitoring and insight; 6) Health service navigation; 7) Social integration and support; and 8) Emotional distress. Positive changes were observed in intention‐to‐treat (ITT) analyses for the 'Constructive attitudes and approaches' domain (MD in change 0.14, 95% CI 0.00 to 0.27; P < 0.01) and 'Skill and technique acquisition' domain (MD in change 0.06, 95% CI ‐0.06 to 0.19; P = 0.04).

Patient activation

One study – Titova 2015 – measured patient activation using the Patient Activation Measure (PAM) and reported no differences between self‐management intervention and usual care in the mean values of the PAM scores at 6 months (MD in change 0.5, 95% CI ‐6.2 to 7.3), 12 months (MD in change 0.5, 95% CI ‐5.8 to 6.7) and 24 months' follow‐up (MD in change 3.1, 95% CI ‐4.5 to 10.8).

Health literacy

No studies reported data on health literacy.

Subgroup analyses

We performed 28 of the a priori 42 defined subgroup analyses for three outcomes: HRQoL, respiratory‐related hospital admissions, and all‐cause mortality (see Table 5). Fourteen subgroup analyses on outcomes of interest could not be performed due to an inadequate number of studies (n ≤ 2) in one of the two subgroups, and were therefore not presented.

4. Subgroup analyses.

	HRQoL		Respiratory‐related hospital admissions		All‐cause mortality
Group description	Group 1	Group 2	Group 1	Group 2	Group 1	Group 2
Duration of intervention: 1. Short (< 8 weeks) versus 2. Long (≥ 8 weeks)	2*	12*	2*	13*	3	21
COPD stability at inclusion: 1. Acute; versus 2. Stable	2*	7*	3	8	4	14
Income country: 1. Low/medium; versus 2. High	1*	13*	0*	15*	1*	23*
Care setting intervention: 1. Primary; versus 2. Secondary/tertiary	7	6	5	9	10	13
COPD exacerbation action plan component: 1. Yes; versus 2. No	11	3	15*	0*	21	3
Home‐based exercise/physical activity component: 1. Yes; versus 2. No	8	6	8	7	15	9
Smoking cessation component: 1. Yes; versus 2. No	9	5	6	9	13	11
Diet component: 1. Yes; versus 2. No	6	8	4	11	7	17
Medication component: 1. Yes; versus 2. No	13*	1*	10	5	19	5
Coping with breathlessness component: 1. Yes; versus 2. No	9	5	8	7	13	11
Self‐recognition of COPD exacerbations component: 1. Yes; versus 2. No	12*	2*	13*	2*	19	5
Use of digital technology: 1. Yes; versus 2. No	2*	12*	3	12	3	21
Integration of BCT clusters: 1. 2 BCTs; versus 2. > 2 BCTs	1*	14*	1*	15*	2*	23*
Integration of BCT clusters: 1. ≤ median of 4; versus 2. > median of 4	4	10	5	10	10	14

BCT: behaviour change technique; COPD: chronic obstructive pulmonary disease; HRQoL: health‐related quality of life

*Subgroup analysis could not be performed because of an insufficient number of studies in one of the subgroups.

Duration of intervention: short (< 8 weeks) versus longer (≥ 8 weeks)

No difference was found in all‐cause mortality between studies with short intervention duration (n = 3; RD ‐0.00, 95% CI ‐0.02 to 0.02) or longer intervention duration (n = 21; RD ‐0.02, 95% CI ‐0.04 to 0.01) (test for subgroup differences: Chi² = 0.93, degrees of freedom (df) = 1 (P = 0.33), I² = 0%) (forest plots not shown).

COPD stability at time of inclusion: acute phase versus stable phase

No difference was found in respiratory‐related hospital admissions between studies that included participants in acute phase (n = 3; OR 0.59, 95% CI 0.32 to 1.08) or stable phase (n = 8; OR 0.74, 95% CI 0.48 to 1.15) (test for subgroup differences: Chi² = 0.35, df = 1 (P = 0.55), I² = 0%) (forest plots not shown).

Also, no difference was found in all‐cause mortality between studies that included participants in acute phase (n = 4; RD ‐0.00, 95% CI ‐0.07 to 0.06) or stable phase (n = 14; RD 0.00, 95% CI ‐0.02 to 0.02) (test for subgroup differences: Chi² = 0.02, df = 1 (P = 0.89), I² = 0%) (forest plots not shown).

Country of intervention: low‐ and medium‐income versus high‐income

We classified included studies into a low‐, medium‐ or high‐income country according to the World Bank list of economies (World Bank 2021). Only one study was conducted in a middle‐income country (Wang 2019), while the other included studies were all conducted in high‐income countries. We were therefore not able to create subgroups of sufficient size to permit meta‐analyses on low‐ and middle‐income countries versus high‐income countries.

Care setting of intervention: primary care versus secondary and tertiary care

No difference for the effects on HRQoL was detected between studies that were conducted in a primary care setting (n = 7; MD ‐3.53, 95% CI ‐5.15 to ‐1.92) compared to a secondary and tertiary care setting (n = 6; MD ‐2.48, 95% CI ‐6.69 to 1.73) (test for subgroup differences: Chi² = 0.21, df = 1 (P = 0.65), I² = 0%) (forest plots not shown).

No difference was also found in respiratory‐related hospital admissions between studies conducted in a primary care setting (n = 5; OR 0.87, 95% CI 0.46 to 1.63) versus a secondary and tertiary care setting (n = 9; OR 0.64, 95% CI 0.52 to 0.79) (test for subgroup differences: Chi² = 0.82, df = 1 (P = 0.36), I² = 0%) (forest plots not shown).

No difference in the effects on all‐cause mortality was found for studies conducted in a primary care setting (n = 10; RD 0.01, 95% CI ‐0.01 to 0.02) versus a secondary and tertiary care setting (n = 13; RD ‐0.02, 95% CI ‐0.06 to 0.01) (test for subgroup differences: Chi² = 2.34, df = 1 (P = 0.13), I² = 57.2%) (forest plots not shown).

Use of self‐management intervention components:

Inclusion of a ‘COPD exacerbation action plan component’

We found no difference in HRQoL for studies with a COPD exacerbation action plan as a self‐management intervention component (n = 11; MD ‐2.57, 95% CI ‐4.11 to ‐1.04) versus studies without a plan (n = 3; MD ‐3.13, 95% CI ‐13.29 to 7.02) (test for subgroup differences: Chi² = 0.01, df = 1 (P = 0.92), I² = 0%) (forest plots not shown).

No difference for all‐cause mortality was found for studies with a COPD exacerbation action plan as a self‐management intervention component (n = 21; RD ‐0.01, 95% CI ‐0.03 to 0.01) versus studies without a plan (n = 3; RD 0.01, 95% CI ‐0.01 to 0.03) (test for subgroup differences: Chi² = 1.86, df = 1 (P = 0.17), I² = 46.2%) (forest plots not shown).

Inclusion of a ‘home‐based exercise programme or physical activity component’

For HRQoL, we found no difference between the subgroups (studies with a home‐based exercise or physical activity component: n = 8 (MD ‐2.51, 95% CI ‐5.78 to 0.77) versus studies without a home‐based exercise or physical activity component: n = 6 (MD ‐3.18, 95% CI ‐5.53 to ‐0.83) (test for subgroup differences: Chi² = 0.11, df = 1 (P = 0.74), I² = 0%) (forest plots not shown)).

For respiratory‐related hospital admissions, no difference was found between the inclusion of a home‐based exercise or physical activity component (n = 8; OR 0.82, 95% CI 0.50 to 1.33) versus the absence of a home‐based exercise or physical activity component (n = 7; OR 0.72, 95% CI 0.51 to 1.02) (test for subgroup differences: Chi² = 0.18, df = 1 (P = 0.67), I² = 0%) (forest plots not shown).

For all‐cause mortality, again, no difference was found between the inclusion of a home‐based exercise or physical activity component (n = 15; RD ‐0.02, 95% CI ‐0.04 to 0.00) versus the absence of a home‐based exercise or physical activity component (n = 9; RD 0.01, 95% CI ‐0.03 to 0.05) (test for subgroup differences: Chi² = 1.74, df = 1 (P = 0.19), I² = 42.6%) (forest plots not shown).

Inclusion of a ‘smoking cessation component’

No effect was observed in a subgroup analysis on HRQoL (studies including a smoking cessation component: n= 9 (MD ‐3.37, 95% CI ‐5.95, ‐0.79) versus studies without a smoking cessation component: n = 5 (MD 0.99, 95% CI ‐3.16 to 1.18) (test for subgroup differences: Chi² = 1.91, df = 1, (P = 0.17), I² = 47.8%) (forest plots not shown)).

A subgroup analysis on respiratory‐related hospital admissions showed a between‐group difference of studies with a smoking cessation component (n = 6; OR 1.10, 95% CI 0.68 to 1.79) versus studies without a smoking cessation component (n = 9; OR 0.60, 95% CI 0.44 to 0.82) (test for subgroup differences: Chi² = 4.20, df = 1 (P = 0.04)), including a substantial variability in effect estimates from the different subgroups (I² = 76.2%) (Analysis 3.1; Figure 10).

3.1. Analysis.

Comparison 3: Subgroup analyses: self‐management versus usual care, Outcome 1: Healthcare utilisation: respiratory‐related hospital admissions (subgroup by smoking cessation component)

10.

Forest plot of comparison: self‐management versus usual care, outcome: 3.1 Healthcare utilisation: respiratory‐related hospital admissions (subgroup by smoking cessation component)

For all‐cause mortality, no difference was found between the group of studies that included a smoking cessation component (n = 13; RD ‐0.02, 95% CI ‐0.04 to 0.01) versus the group of studies that did not include a smoking cessation component (n = 11; RD 0.01, 95% CI ‐0.03 to 0.04) (test for subgroup differences: Chi² = 1.23, df = 1 (P = 0.27), I² = 18.4%) (forest plots not shown).

Inclusion of a ‘diet component’

For HRQoL, no effect was observed between the subgroups (including diet component: n = 6 (MD ‐1.84, 95% CI ‐3.94 to 0.25); no diet component: n = 8 (MD ‐3.55, 95% CI ‐6.54 to ‐0.57) (test for subgroup differences: Chi² = 0.84, df = 1 (P = 0.36), I² = 0%) (forest plots not shown)).

We found no subgroup differences in respiratory‐related hospital admissions between studies with a diet component (n = 4; OR 1.22, 95% CI 0.57 to 2.58) and without a diet component (n = 11; OR 0.65, 95% CI 0.52 to 0.80) (test for subgroup differences: Chi² = 2.49, df = 1 (P = 0.11), I² = 59.8%) (forest plots not shown).

For all‐cause mortality, no difference was observed between studies including a diet component (n = 7; RD ‐0.03, 95% CI ‐0.07 to 0.02) versus without a diet component (n = 17; RD ‐0.00, 95% CI ‐0.02 to 0.01) (test for subgroup differences: Chi² = 0.88, df = 1 (P = 0.35), I² = 0%) (forest plots not shown).

Inclusion of a ‘medication component’

For respiratory‐related hospital admissions, no difference was observed between the subgroups (studies including a medication component: n = 10 (OR 0.82, 95% CI 0.59 to 1.16) versus studies with no medication component: n = 5 (MD 0.60; 95% CI 0.40 to 0.91) (test for subgroup differences: Chi² = 1.29, df = 1 (P = 0.26), I² = 22.8%) (forest plots not shown)).

We also did not find a difference for all‐cause mortality for studies with a medication component (n = 19; RD ‐0.01, 95% CI ‐0.03 to 0.01) versus without a medication component (n = 5; RD ‐0.01, 95% CI ‐0.07 to 0.05) (test for subgroup differences: Chi² = 0.00, df = 1 (P = 0.97), I² = 0%) (forest plots not shown).

Inclusion of a ‘coping with breathlessness component’

No effect was found in a subgroup analysis on HRQoL for studies with a 'coping with breathlessness' component (n = 9; MD ‐3.73, 95% CI ‐6.93 to ‐0.52) versus studies without such a component (n = 5; MD ‐1.89, 95% CI ‐4.41 to 0.63) (test for subgroup differences: Chi² = 0.78, df = 1 (P = 0.38), I² = 0%) (forest plots not shown).

No effect was observed in a subgroup analysis on respiratory‐related hospital admissions for studies with a 'coping with breathlessness' component (n = 8; OR 0.77, 95% CI 0.48 to 1.24) versus studies without such a component (n = 7; OR 0.71, 95% CI 0.56 to 0.91) (test for subgroup differences: Chi² = 0.08, df = 1 (P = 0.77), I² = 0%) (forest plots not shown).

No difference was found in effects on all‐cause mortality between studies with a 'coping with breathlessness' component (n = 13; RD ‐0.01, 95% CI ‐0.04 to 0.01) versus studies without such a component (n = 11; RD ‐0.00, 95% CI ‐0.03 to 0.02) (test for subgroup differences: Chi² = 0.40, df = 1 (P = 0.53), I² = 0%) (forest plots not shown).

Inclusion of a ‘self‐recognition of COPD exacerbations component’

For all‐cause mortality, we found no difference between studies with a 'self‐recognition of COPD exacerbations' component (n = 19; RD ‐0.01, 95% CI ‐0.03 to 0.01) and studies without such a component (n = 5; RD ‐0.00, 95% CI ‐0.03 to 0.02) (test for subgroup differences: Chi² = 0.27, df = 1 (P = 0.60), I² = 0%) (forest plots not shown).

Use of digital technology

No difference was observed for the probability of respiratory‐related hospital admissions between studies that incorporated digital technology (n = 3; OR 0.85, 95% CI 0.17 to 4.15) versus studies without digital technology (n = 12; OR 0.74, 95% CI 0.58 to 0.95) (test for subgroup differences: Chi² = 0.03, df = 1 (P = 0.87), I² = 0%) (forest plots not shown).

There was also no difference observed in the risk of all‐cause mortality for studies with digital technology (n = 3; RD ‐0.00, 95% CI ‐0.15 to 0.15) versus without digital technology (n = 21; RD ‐0.01, 95% CI ‐0.03 to 0.01) (test for subgroup differences: Chi² = 0.01, df = 1 (P = 0.92), I² = 0%) (forest plots not shown).

Integration of behavioural change technique (BCT) clusters: high number of BCT clusters (> median of 4) versus low number of BCT clusters (≤ median of 4)

We observed no difference for the effects on HRQoL amongst studies with a high number of BCT clusters (n = 10; MD ‐2.62, 95% CI ‐5.37 to 0.13) versus a low number of BCT clusters (n = 4; MD ‐3.79, 95% CI ‐6.02 to ‐1.56) (test for subgroup differences: Chi² = 0.42, df = 1 (P = 0.52), I² = 0%) (forest plots not shown).

Subgroup analyses on BCT clusters integrated in the self‐management intervention also showed no differences in respiratory‐related hospital admissions in studies with a high number of BCT clusters (n = 10; OR 0.86, 95% CI 0.60 to 1.22) compared to studies with a low number of BCT clusters (n = 5; OR 0.57, 95% CI 0.43 to 0.76) (test for subgroup differences: Chi² = 3.08, df = 1 (P = 0.08)), including a substantial variability in effect estimates from different subgroups (I² = 67.6%) (forest plots not shown).

No difference between subgroups on all‐cause mortality was found amongst studies with a low number of BCT clusters (n = 10; RD ‐0.03, 95% CI ‐0.06 to 0.01) and studies with a high number of BCT clusters (n = 14; RD 0.00, 95% CI ‐0.01 to 0.02) (test for subgroup differences: Chi² = 2.78, df = 1 (P = 0.10)), including a substantial variability in effect estimates from different subgroups (I² = 64.1%) (forest plots not shown).

Recently published studies

The updated search from January 2020 to March 2021 identified one study (Ozturk 2020), that met all the current inclusion criteria. This study included several COPD self‐management intervention components; namely, smoking cessation, exercise and physical activity, coping with breathlessness, energy‐saving techniques, psychological assessment and nutritional training. They observed beneficial effects in the self‐management intervention compared to usual care in HRQoL measured by the total SGRQ score (P = 0.02), health status measured by the CAT score (P < 0.001), anxiety and depression symptoms measured by the HADS score (anxiety: P = 0.01; depression: P = 0.01) (Ozturk 2020).

Discussion

Summary of main results

This is an update of a review previously published in 2014 (Zwerink 2014). We systematically evaluated 25 RCTs and two CRTs (described in 38 articles) on the effectiveness of COPD self‐management interventions compared to usual care. Compared to the previous review update, we included 21 new self‐management studies. We had to exclude 19 of 29 previously included studies, due to the application of stricter inclusion criteria regarding COPD self‐management interventions and COPD diagnosis (Effing 2016), and because some of the previously included studies were not RCTs. Positive effects of COPD self‐management interventions on HRQoL and respiratory‐related hospitalisations were detected (Effing 2007; Monninkhof 2002; Monninkhof 2003; Zwerink 2014). The lack of observed effects regarding respiratory‐related and all‐cause mortality has strengthened the view that COPD self‐management interventions are unlikely to cause harm. Applying stricter inclusion criteria has led to less heterogeneity in interventions, and has also resulted in fewer studies being eligible for review. Most predefined subgroup analyses did not show differences; the small numbers of studies in many of these subgroup analyses may have contributed to this. As a result, we have been unsuccessful in identifying potentially effective (intervention) characteristics that can be linked to COPD self‐management intervention outcomes.

We observed a beneficial effect for COPD self‐management interventions on HRQoL, measured by the SGRQ adjusted total score which did not reach the MCID of four points (Jones 2005). We determined a priori that, in addition to analysing the final outcome points, if possible, we would also analyse short‐term (≤ 6 months' follow‐up), medium‐term (> 6 to ≤ 12 months' follow‐up), and long‐term (> 12 months' follow‐up) effects of the primary outcomes in this review. For the SGRQ, these analyses could be performed for short‐ and medium‐term effects. Only the analysis for the medium‐term effects showed a pooled beneficial effect favouring self‐management. This could suggest that HRQoL may further improve when people with COPD develop their self‐management skills over time. However, due to the highly overlapping CIs of short‐ and medium‐term SGRQ effects, and the impossibility of performing a subgroup difference test, this interpretation should be treated with caution, and no conclusions can be drawn from our data regarding the influence of time on HRQoL improvements.

A beneficial self‐management effect was observed for respiratory‐related hospital admissions. Participants in self‐management intervention study arms were at a lower risk for at least one respiratory‐related hospital admission compared to participants who received usual care. Fifteen participants with high baseline risk and 14 participants with low baseline risk needed to be treated to prevent one respiratory‐related hospital admission over a mean follow‐up of 9.75 months.

We observed no difference between self‐management interventions and usual care for the risk of all‐cause mortality.

No effect was found on respiratory‐related mortality (RD 0.01, 95% CI ‐0.02 to 0.04). In a review regarding the effect of COPD self‐management interventions, including exacerbation action plans, published in 2017 (Lenferink 2017), a very small, but higher respiratory‐related mortality rate was found in the self‐management intervention group compared to the usual care group (RD 0.03, 95% CI 0.05 to 0.05). The two studies that dominated this negative effect were also included in the current analysis (Bucknall 2012; Fan 2012), but with more studies and participants included (1572 versus 1219 participants), no detrimental effect was detected.

In the current update, we did not find a difference in the probability of all‐cause hospital admissions in the self‐management intervention group compared to the usual care group. The previous 2014 update reported that the probability of having one or more all‐cause hospital admissions was higher in the self‐management group (Zwerink 2014). The included studies in our meta‐analysis (n = 10) included four of the six previously included studies. Six newly included studies showed heterogeneity of effect sizes. The lack of effect on all‐cause hospitalisations in our review may be explained by the fact that most COPD self‐management intervention studies are still predominantly directed towards COPD and do not include treatment components directed towards frequently existing comorbidities (Zwerink 2014).

A beneficial difference was observed for self‐management interventions on ED visits, both for the number of participants with at least one visit and the mean number of visits per participant. However, the lack of clear definitions of ED visits in most studies and considerable heterogeneity in the mean number of ED visits per participant meaning that these results should be interpreted with caution.

A higher use of oral corticosteroids and antibiotics was reported in the self‐management groups. The higher probability of using at least one course of oral steroids was found after sensitivity analysis and should be interpreted with caution. Only three studies were included in this sensitivity analysis (Gallefoss 1999; Rice 2010; Sanchez‐Nieto 2016), with the Rice 2010 study having a large population and high proportion of events. No sensitivity analysis excluding Rice 2010 could be performed because of the limited number of studies. Higher use of both corticosteroids and antibiotics in the self‐management groups may have been triggered by the use of exacerbation action plans encouraging earlier initiation of self‐treatment with corticosteroids, and when necessary, antibiotics. However, the differences in medication use could as likely be caused by actual undertreatment of exacerbations in the usual care group.

Anxiety and depression, measured by the HADS, were both reduced in participants assigned to the self‐management intervention. Whereas the presence of a mental health component in most of the interventions would have been a very plausible explanation, this was not the case, as only two of the nine studies included in the meta‐analyses had a mental health component (Jonsdottir 2015; Walters 2013). However, six of the nine studies included a ‘coping with breathlessness’ component (Bucknall 2012; Hernández 2015; Jonsdottir 2015; Lenferink 2019; Titova 2015; Walters 2013), which may have contributed to the positive effects on anxiety and depression. Respiratory health and especially breathlessness in COPD can trigger anxiety symptoms (Heslop‐Marshall 2014). Importantly, the baseline levels of anxiety and depression in most of the included studies were quite high (mean total HADS scores > 11 in both study groups) (Spinhoven 1997; Zigmond 1983), resulting in room for improvement for this parameter.

The MD of 45 meters between self‐management intervention and usual care groups in the 6MWT was clinically relevant, favouring self‐management. As only two of the six studies in this meta‐analysis included a ‘home‐based exercise component’ in their self‐management intervention, it is not plausible that this component was the only contributor to the improvement in walking distance. The ‘COPD exacerbation action plan component’ – included in five of the six studies – may also have played a role in this improvement as it encourages prompt treatment of exacerbations and therefore may have led to less severe exacerbations, a faster recovery, and possibly a better physical condition.

Subgroup analyses

The total number of included studies provided the opportunity to perform several subgroup analyses to try to gain greater insight into the “black box” of COPD self‐management interventions. However, only one subgroup analysis showed a difference between effects in the subgroups. A limited number of studies in the majority of subgroup analyses may have contributed to the lack of effects.

Studies without a smoking cessation component showed a lower probability on respiratory‐related hospital admissions favouring self‐management compared to studies with a smoking cessation component. This is the opposite effect of what was expected, as smoking cessation is associated with a reduction in hospitalisations (Godtfredsen 2002). A possible explanation may be that participants in both the self‐management and usual care groups of studies without a smoking cessation component may have already quit smoking before entering the study, and may therefore have already achieved the beneficial effect on respiratory‐related hospital admission before study entry, leading to less room for improvement by the self‐management intervention. The numbers of current smokers in study groups included in this subgroup analysis were however fairly comparable at baseline (range of current smokers in self‐management intervention groups: 13.0% to 53.5%; range of current smokers in usual care groups: 14.0% to 71.3%). Unfortunately, no data were reported regarding the participants that actually stopped smoking during the intervention, and we cannot rule out differences in other variables between the two groups that may explain this effect. A meta‐regression analysis would have provided us with the possibility to adjust for potential effect modifiers (Higgins 2019). Unfortunately, due to an insufficient number of studies in our review, we could not use this very promising statistical technique, as at least 10 studies for each characteristic modelled are needed. However, the technique should be considered if future reviews include significantly more studies.

Overall completeness and applicability of evidence

Our review showed beneficial effects on HRQoL and respiratory‐related hospital admissions. Additionally, beneficial effects were detected for ED visits, anxiety, depression and exercise capacity. Also, no increase in mortality was detected in the COPD self‐management interventions, which strengthens the view that these interventions are unlikely to cause harm. Lastly, our results showed higher use of antibiotic courses in the self‐management group.

This review included 6008 participants with COPD having a post‐bronchodilator FEV1 to FVC ratio of less than 0.7. We included studies conducted in 13 different countries on four different continents (15 in Europe, eight in North America, one in Asia, and four in Oceania; with one study conducted in both Europe and Oceania), suggesting that our findings can be generalised across various high‐income healthcare settings. Ideally, more studies from Asia would have been included, but we encountered problems with gaining required inclusion information from six potentially eligible Asian studies (Abdulsalim 2017; Alharbey 2019; Ghanem 2010; Li Z 2015; Liu 2013; Lou 2015). Having a better distribution of included studies over all continents would certainly increase the generalisability of the review results. Our searches were current up to January 2020.

There are some limitations to the generalisability of our results. We had difficulties collecting essential information regarding 12 studies (Abdulsalim 2017; Aboumatar 2017; Alharbey 2019; Efraimsson 2008; Ghanem 2010; Heidari 2018; Hill 2010; Jiang 2012; Khdour 2009; Li 2014; Liu 2013; Lou 2015). Based on the information provided in the publications, we were not able to check whether the studies included only participants that met our COPD diagnosis criteria, and whether they had at least two self‐management intervention components provided to all included participants using an iterative process. We made at least three attempts to request information from the authors of these studies. Unfortunately, we received no response from the authors.

Three of the included studies (11%) had follow‐up periods of three months or less (Bringsvor 2018; Emery 1998; Johnson‐Warrington 2016). Depending on the time of participant enrolment (e.g. during summer), seasonal variation may have influenced the outcomes in these studies (e.g. the number of exacerbations). This may have resulted in an under‐ or overestimation of the actual effect. The study by Fan 2012 was prematurely stopped with a mean follow‐up of 250 days, because of a higher number of deaths in the intervention group compared with the control group that could not be explained satisfactorily by the study authors. It is therefore uncertain if a true effect was observed. The results of this study need to be interpreted with caution.

Moreover, we were not able to perform a meta‐analysis on physical activity outcomes, because of limited studies including this component. Many studies that incorporated a sole physical activity component were excluded from this review, as inclusion required at least two components with an iterative process.

Over the span of 25 years, views about what is required for COPD self‐management interventions have changed. Nowadays, it is agreed that self‐management interventions should incorporate BCTs and encourage activation of participants. In addition, social support and digital technology have more frequently been integrated. Whereas usual care is diverse across countries, it is likely that usual care has been optimised over the years, and that self‐management approaches are increasingly embedded in usual care. This leads to the expectation that the observed benefits of self‐management interventions compared to usual care will be diminishing.

Quality of the evidence

All 27 included studies in our review were judged as having an overall high risk of bias for several reasons. Due to the nature of COPD self‐management interventions, it is not possible to blind participants and personnel during RCTs and CRTs. In addition, none of the studies provided detailed information regarding the distribution of non‐protocol interventions to the study groups during the follow‐up period of the study. Finally, not a single study reported whether an appropriate analysis was used to estimate the effect of adhering to interventions. As a result, all studies scored ‘high risk’ in the same domain of the risk of bias 2 (ROB 2) tool (i.e. domain 2 – ‘deviations from the intended interventions’)(Sterne 2019), and consequently scored ‘high’ on overall risk of bias. This directly affected the GRADE score (Guyatt 2011); outcomes were downgraded from high‐ to moderate‐quality evidence, from moderate‐ to low‐quality evidence, and from low‐ to very low‐quality evidence.

With the assigned overall high risk of bias score and moderate heterogeneity, the quality of evidence for HRQoL was graded as low. The improvement in HRQoL, measured by the adjusted SGRQ total score, did not reach the MCID. Therefore, we need to consider this carefully as the positive effects may only have been clinically relevant for part of the population. Furthermore, the overall high risk of bias score, moderate heterogeneity of included studies and a wide 95% CI resulted in very low‐quality evidence for respiratory‐related hospital admissions. We graded the quality of evidence for all‐cause and respiratory‐related mortality as low because of the high risk of bias in all included studies, and substantial heterogeneity resulting in inconsistency in both mortality outcomes. Finally, we graded the quality of evidence for all other secondary outcomes as moderate to low; assessments were based on fewer studies or smaller sample sizes, or both.

Because of the nature of the self‐management intervention, blinding of personnel and participants to group assignment is complicated and will be very unlikely in future studies. Improvement in the risk of bias due to deviations from intended interventions may be achieved if studies: a) provide better descriptions of non‐protocol interventions used in study groups (e.g. by using online repositories); and b) detail intervention implementation failures and non‐adherence to the intervention; or c) apply and describe appropriate analyses to estimate the effect of adhering to interventions. By doing this, the overall risk of bias will be reduced, and consequently, the overall quality of the evidence will improve.

Serious inconsistency in effect sizes affected the quality of evidence of more than half of the outcomes. Heterogeneity in intervention content may have contributed to this inconsistency. In this review, we have tried to decrease this heterogeneity by applying stricter inclusion criteria in line with the most recent published definition of COPD self‐management interventions (Effing 2016). However, because of the nature of COPD self‐management interventions and because individual tailoring is desirable, heterogeneity in future interventions will be inevitable; it will never be a 'one size fits all' intervention.

Potential biases in the review process

We observed heterogeneity in clinical diversity (e.g. care setting, intervention components, BCTs, intensity and duration), (primary) outcome measures, and statistical diversity (e.g. variability in intervention effects). As inclusion of studies in this review was not based on reported outcome measures, we observed a broad spectrum of outcome measures with various methods for assessment (e.g. different questionnaires for the same outcome measure) and various calculations (e.g. mean number versus the percentage of participants). We could therefore not perform all predefined meta‐analyses due to insufficient (< 3 studies) similarly‐reported outcome data.

Studies were only eligible if self‐management interventions had at least two intervention components that were offered to all included participants. Several self‐management intervention studies that incorporated intervention components and characteristics tailored to the individual (e.g. personalised care plans), could not confirm that at least two specific components were offered to each included participant, and therefore could not be included in this review.

Furthermore, studies were only included when data or authors confirmed that all included participants met COPD spirometry FEV1/FVC criteria (GOLD 2021). However, crucial information on COPD diagnosis remained missing, even after several contact attempts with authors. We have therefore been unable to include all potentially eligible studies.

In this review update, we classified the Titova 2015 study as an RCT. However, we acknowledge that this classification is tentative, as it remains unclear whether random sequence allocation was performed at the participant or health centre level. If random sequence allocation was performed solely at health centre level, the study should have been classified as a CRT.

A priori, we expected to see heterogeneity amongst studies due to the nature of the self‐management intervention. We therefore decided to use a REM for the meta‐analyses. This model weighs by study, rather than number of participants, when heterogeneity is present. However, when only a few large studies and many small studies are included, this may result in bias introduced by small‐study effects. We therefore performed several sensitivity analyses using FEM meta‐analysis. However, bias introduced by small‐study effects was unlikely, as the observed effect sizes in FEM and REM were comparable, except for the meta‐analysis on the use of oral corticosteroids where a non‐significant REM resulted in a significant FEM analysis.

Self‐management interventions were not always described in sufficient detail to allow coding of all applied BCTs. It is possible that some BCTs were present in interventions but were not adequately described by study authors. For standardisation purposes, we decided to use only data that were explicitly reported in published articles of included studies, and we coded BCTs by using the mobile BCT Taxonomy application (BCT Taxonomy). So, we have not used any extra information that was provided by authors for determining BCTs (e.g. unpublished protocols). Our approach has almost certainly led to an underestimation of the number (and variety) of BCTs and may therefore have contributed to the lack of found effects in subgroup analyses regarding BCTs. There is a significant need for providing more detailed, uniformly and transparently reported data on BCTs used in self‐management interventions in future studies – for example, by using online journal repositories – to increase the meaningfulness of the BCT subgroup analysis.

Agreements and disagreements with other studies or reviews

A previous Cochrane Review by Lenferink and colleagues on COPD self‐management interventions that include action plans for exacerbations of COPD (Lenferink 2017), reported similar beneficial effects on HRQoL, measured by SGRQ (MD ‐2.69, 95% CI ‐4.49 to ‐0.90), and respiratory‐related hospital admissions (OR 0.69, 95% CI 0.51 to 0.94). However, they observed a small negative effect on respiratory‐related mortality (RD 0.03, 95% CI 0.05 to 0.05), which was not observed in the present review. Their review highlighted that self‐management interventions including exacerbation action plans with a smoking cessation programme contributed to significant improvements in HRQoL (Lenferink 2017). However, this could not be confirmed by subgroup analyses in our review.

A review by Jonkman 2016 aimed to quantify the diversity in components of self‐management interventions and aimed to identify intervention components that improve HRQoL, measured by SGRQ, in chronically ill participants (i.e. COPD, chronic heart failure, diabetes), by conducting individual patient data analysis. They found that self‐management interventions improve HRQoL in participants with COPD at 12 months (SMD 0.08, 95% CI 0.00 to 0.16), but not at 6 months (SMD 0.05, 95% CI ‐0.05 to 0.15). This finding could not be confirmed in our review. Their subgroup analyses did not identify any intervention components that were associated with the intervention effects (Jonkman 2016). Furthermore, a risk reduction was found at 12 months in respiratory‐related hospital admissions (RR 0.77, 95% CI 0.64 to 0.93) and all‐cause hospital admissions (RR 0.84, 95% CI 0.73 to 0.96). It was also observed that a longer duration of self‐management interventions conferred a reduction in respiratory‐related hospital admissions (hazard ratio 0.79, 95% CI 0.66 to 0.94) and all‐cause hospitalisations (hazard ratio 0.80, 95% CI 0.69 to 0.92). These results strengthen the finding that self‐management interventions with longer follow‐up duration should be recommended for clinical practice, rather than interventions with short‐term follow‐up. Whether the observed lower risks are clinically relevant is unclear, because there is no MCID for hospitalisations.

Jordan 2015 conducted a review, including ten RCTs, on the effectiveness of supported self‐management interventions that were delivered to participants with COPD who had recently been discharged from hospital. As in the current review, a beneficial effect on HRQoL, measured by SGRQ, was reported (MD 3.48, 95% CI 1.29 to 6.40), but the Jordan 2015 review included studies that showed high rates of loss to follow‐up. Furthermore, no distinct beneficial effects were found on mortality, anxiety and depression, and exercise capacity.

Another review on COPD self‐management interventions by Newham 2017 observed that self‐management interventions were significantly more effective than usual care. In line with the current review, they found significant improvements in HRQoL, measured by SGRQ, and a reduced number of ED visits. In addition, self‐management interventions that tackle mental health concerns were considered to be more effective than those that focused on symptom management alone. As in the current review, they used an established taxonomy – Michie 2013 – to assess the integration of BCTs into self‐management interventions of included studies. Also in line with our review, they observed no significant association between number of BCTs and improvement in HRQoL (Newham 2017).

Furthermore, a review by Jolly 2018 on the effectiveness of community‐based self‐management interventions reported no effects on HRQoL (measured by SGRQ) or anxiety and depression. This review only included studies conducted in primary care, and therefore only included participants with mild or moderate COPD. In our review, we performed a subgroup analysis on HRQoL in studies delivered in primary care settings (n = 7) versus secondary and tertiary care settings (n = 6) and found no significant between‐group differences in effects. Jolly 2018 argued that people with COPD in primary care may still benefit from self‐management support, although it may be ineffective in its current form. Further research in this primary care COPD population is recommended to identify suitable and effective self‐management interventions for the less severe primary care population.

Finally, a recent review by Song 2021 aimed to evaluate 'blended' (i.e. eHealth combined with individual face‐to‐face) self‐management interventions compared to: 1) eHealth interventions with and without usual care; 2) face‐to‐face interventions with or without usual care; and 3) usual care only, in participants with COPD and asthma. In participants with COPD, they revealed beneficial effects of the blended self‐management intervention on HRQoL (measured by SGRQ, CAT and CRQ), exercise capacity and hospital admissions. However, the comparator was unclear.

Authors' conclusions

Implications for practice.

Self‐management interventions in people with COPD are associated with improvement in HRQoL, as measured by the SGRQ; a reduction in both respiratory‐related admissions and ED visits; a likely improvement in both anxiety and depression symptoms, and exercise capacity; and probably more use of antibiotics. No differences were found in other outcome parameters. In addition, the lack of observed effects regarding respiratory‐related and all‐cause mortality strengthens the view that COPD self‐management interventions are unlikely to cause harm. By using stricter inclusion criteria, we have decreased the heterogeneity amongst studies, but also reduced the number of studies that could be included in this review and therefore our capacity to do subgroup analyses. Consequently, the data are insufficient to permit clear conclusions about effective (intervention) characteristics of COPD self‐management interventions (e.g. duration of the intervention, intervention components). Because tailoring of self‐management interventions to individuals is desirable, heterogeneity is and will in all likelihood remain present in self‐management interventions.

Future clinical practice may focus on the following strategy:

• Ensuring that the self‐management interventions meet the criteria of the definition of COPD self‐management interventions (Effing 2016) (e.g. include multiple intervention components with an iterative process between healthcare provider(s) and participants, directed towards behaviour change).

Implications for research.

Future studies and systematic reviews of studies should focus on the following points and improvements:

• Providing more detailed, uniformly and transparently reported data on the self‐management intervention components and the BCTs used.

• Achieving greater homogeneity in outcome measures, with greater attention to behavioural outcome measures.

• Assessing outcomes over the long term (> 12 months' follow‐up), as COPD self‐management interventions are directed towards behavioural change, which is often not achieved in a short period of time and if this is the case, it would be of interest to know whether this change is maintained.

• Providing more detailed information regarding deviations from intended interventions to allow for higher certainty of evidence of outcomes (i.e. information regarding distribution of non‐protocol interventions, as well as analyses used to estimate the effect of adhering to intervention).

What's new

Date	Event	Description
13 March 2023	Amended	Correction to typo in the Results section (subgroup analyses, page 32)

History

Protocol first published: Issue 2, 2001
Review first published: Issue 1, 2003

Date	Event	Description
14 January 2022	Amended	Title changed
27 January 2020	New citation required and conclusions have changed	Complete rewrite of the review conducted. 21 new studies added. New risk of bias assessment completed for all included studies. References in background updated. Literature search run in March 2021 and not fully incorporated see Studies awaiting classification.
27 January 2020	New search has been performed	New literature search run.
7 July 2014	Amended	We amended the data for all‐cause hospitalisations. This outcome now favours the self‐management group (OR 0.60 95%CI 0.45 to 0.89) and the review was revised accordingly.
31 August 2011	New search has been performed	New literature search run
31 August 2011	New citation required and conclusions have changed	Complete rewrite of the review conducted. Summary of findings table added. 14 new studies added. New risk of bias assessment completed for all included studies. References in background updated Change of title—'education' removed
25 March 2008	Amended	Converted to new review format.
21 August 2007	New citation required and conclusions have changed	New studies: N = 7 (Bourbeau 2003; Boxall 2005; Coultas 2005a; Coutas 2005b; Martin 2004; Monninkhof 2003; Rea 2004) What these studies have added: Data on health related quality of life; exacerbations (hospitalisations, requirement for oral steroids); lung function (FEV1). Quality of life scores and respiratory‐related hospital admission now show significant benefits. Lung function parameters do not show a significant difference. Steroid‐treated exacerbations were not significantly different. How this has changed the review: The review now demonstrates that from the self‐management interventions assessed in the studies assembled in the review, patients were less likely to require hospital admissions when treated with this type of intervention. There was a small improvement in total quality of life scores measured by the St George's Respiratory Questionnaire. There were no indications of detrimental effects in other outcome parameters. The effects of different components of self‐management interventions and their requisite intensity requires more research.

Risk of bias

Risk of bias for analysis 1.1 Health‐related quality of life (HRQoL): adjusted SGRQ total score (primary analysis).

Study

Bias

Randomisation process

Deviations from intended interventions

Missing outcome data

Measurement of the outcome

Selection of the reported results

Overall

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Gallefoss 1999

Low risk of bias

Comment: At inclusion they signed a written consent and were then randomised to an intervention group or a control group.

High risk of bias

Effect of assignment: Comment: No information reported about blinding, but it is almost impossible to blind these patients because of the nature of the intervention. Furthermore, no information was reported whether deviations arose from the intended intervention because of the trial context. Effect of adhering: "Forty‐six percent of those educated (12 of 26) received a standard treatment plan. Non‐standard treatment plans, incorporating the use of oral steroids as the first line of action in the yellow zone, were followed in some cases, for example, if the patient already used high dosages of inhaled steroids as maintenance therapy, or could report that a double or triple increase in inhaled steroids previously had shown marginal effect on the course of attacks/exacerbations. Among those 14 patients receiving non‐standard treatment plans, eight patients did not want to or were not able to use peak flow monitoring as a basis for change in medication. For those patients, symptom‐only based treatment plans were issued." Gallefoss 2002, p. 425. "In the intervention group, four patients failed to complete the educational program (social problems (n = 1), unannounced emigration (n = 1), failure to meet at educational group sessions for unknown reasons (n = 1) and serious myocardial infarction (n = 1))." Gallefoss 2002, p. 427. Comment: No information reported about adherence to the programme of the other patients (who completed the programme).

High risk of bias

“In the intervention group, four patients failed to complete the educational programme (social problems (n = 1), unannounced emigration (n = 1), failure to meet at educational group sessions for unknown reasons (n = 1) and serious myocardial infarction (n = 1)). Another patient was withdrawn from the study during the follow‐up due to lymphoma (n = 1). This left us with 26 patients (81%) for a 1‐year follow‐up. The patients who were withdrawn from the intervention group did not, to our knowledge, have any serious deterioration in their obstructive lung disease, and none were hospitalised. In the control group four patients were withdrawn (lack of co‐operation (n = 2), diagnosis of rectal cancer (n = 1) and emigration (n = 1)). Two of the withdrawn control group patients were hospitalised for exacerbations of their COPD. This left us with 27 patients (84%) for the 1‐year follow‐up” Gallefoss 2002, p. 427. Comment: The number of dropouts was relatively low, and reasons for dropout were comparable over groups

High risk of bias

"Prebronchodilator spirometry was performed before randomization and at 12 mo follow‐up by standard methods using a Jaeger MasterLab Body Box (Würzburg, Germany). The technical staff did not know whether the patients belonged to the control or intervention groups." Gallefoss 1999, p. 2003. Comment: No information reported regarding the assessors of the primary outcome. Number of prednisolone courses was retrospectively self‐reported at 12 months of follow‐up.

Some concerns

"We defined a priori the patient as compliant when dispensed regular medication was greater than 75% of prescribed regular medication during the study period." Gallefoss 1999, p. 2002.

High risk of bias

Unpredictable direction of the outcome.

Bourbeau 2003

Low risk of bias

“… central computergenerated list of random numbers. Randomization was stratified per center and in blocks of 6, and patients were assigned to the self‐management program (intervention group) or to usual care.” p. 586. “The blocking factor was not known by the investigators or their staff in each participating center." p. 586. “The use of respiratory medications was similar between study groups, except that oral steroids were used less commonly in the intervention group (7%) than in the usual care group (13%). None of the disease severity characteristics were otherwise different between the study groups.” p. 586. Comment: Random sequence allocation was adequately performed.

High risk of bias

Effect of assignment: "Since a double‐blind design was impossible..." p. 586. Comment: Participants and personnel were not blinded. ITT analysis was performed. Effect of adhering: "Since a double‐blind design was impossible..." p. 586. Comment: Participants and personnel were not blinded. Comment: There might be some imbalance in non‐protocol interventions such as exercise evaluation (not mandatory). However, no information available.

Low risk of bias

‐

Low risk of bias

"...an independent evaluator unaware of the patient assignment was responsible for the evaluation process in each center. The evaluator was cautioned not to ask about the workbook modules and types of contact.” p. 586. Comment: Outcome assessment was blinded.

Some concerns

Comment: No protocol or prespecified analysis plan available.

High risk of bias

Unpredictable direction of the outcome.

Coultas 2005

Low risk of bias

“Patients were randomly assigned to one of three intervention groups (usual care (UC), nurse‐assisted medical management (MM), or nurse‐assisted collaborative management (CM)) using a computer‐generated random list.” p. 2018. “Although the distribution of most of the baseline characteristics of the three groups was similar (Tables 2 and 3), only the distribution of gender and the tangible domain of social support were significantly different among the intervention groups.” p. 2020. Comment: It is unclear how allocation concealment was guaranteed. Observed imbalances are probably compatible with chance. Random sequence generation was adequately performed.

High risk of bias

Effect of assignment: “Of the 217 patients enrolled in the study, 151 (69.6%) completed the 6‐month intervention and follow‐up data collection.” p. 2020. “The reasons for the failure to complete the study were patient was unavailable for follow‐up (26.7%) and death (3.7%) [Fig 1]. The frequency of patients being unavailable for follow‐up was equally distributed among the three intervention groups (Fig 1). Overall, the demographic characteristics of the patients who dropped out of the study were similar to those who completed the study (data not shown). However, patients who dropped out of the study had more severe airflow obstruction, higher levels of distress, and lower quality of life, as measured with the SGRQ, compared with the patients who had completed the study (data not shown).” p. 2020. Comment: Not reported whether deviations arose because of the trial context. Only patients who completed the 6‐month intervention and follow‐up data collection were analysed (per‐protocol analyses). Effect of adhering: Comment: Patients and personnnel were probably aware of assignment to the intervention during the trial.

High risk of bias

“The reasons for the failure to complete the study were patient was unavailable for follow‐up (26.7%) and death (3.7%).” p. 2020. “However, patients who dropped out of the study had more severe airflow obstruction, higher levels of distress, and lower quality of life, as measured with the SGRQ, compared with the patients who had completed the study (data not shown).” p. 2020. Comment: Only patients who completed the 6‐month intervention and follow‐up data collection were analysed.

Low risk of bias

“To limit interviewer bias, each interviewer who obtained the baseline and 6‐month outcome data was blinded to the patient’s treatment group.” p. 2023. "Although self‐reported health‐care utilization may be subject to recall bias, it has been shown to be an acceptable method of measurement and may actually result in an underestimation of utilization with more frequent use.” p. 2023.

Some concerns

Comment: No protocol available. Measures and analysis intentions were only reported in the method section of the paper.

High risk of bias

Unpredictable direction of the outcome.

Coultas 2005

Low risk of bias

“Patients were randomly assigned to one of three intervention groups (usual care (UC), nurse‐assisted medical management (MM), or nurse‐assisted collaborative management (CM)) using a computer‐generated random list.” p. 2018. “Although the distribution of most of the baseline characteristics of the three groups was similar (Tables 2 and 3), only the distribution of gender and the tangible domain of social support were significantly different among the intervention groups.” p. 2020. Comment: It is unclear how allocation concealment was guaranteed. Observed imbalances are probably compatible with chance. Random sequence generation was adequately performed.

High risk of bias

Effect of assignment: “Of the 217 patients enrolled in the study, 151 (69.6%) completed the 6‐month intervention and follow‐up data collection.” p. 2020. “The reasons for the failure to complete the study were patient was unavailable for follow‐up (26.7%) and death (3.7%) [Fig 1]. The frequency of patients being unavailable for follow‐up was equally distributed among the three intervention groups (Fig 1). Overall, the demographic characteristics of the patients who dropped out of the study were similar to those who completed the study (data not shown). However, patients who dropped out of the study had more severe airflow obstruction, higher levels of distress, and lower quality of life, as measured with the SGRQ, compared with the patients who had completed the study (data not shown).” p. 2020. Comment: Not reported whether deviations arose because of the trial context. Only patients who completed the 6‐month intervention and follow‐up data collection were analysed (per‐protocol analyses). Effect of adhering: Comment: Patients and personnnel were probably aware of assignment to the intervention during the trial.

High risk of bias

“The reasons for the failure to complete the study were patient was unavailable for follow‐up (26.7%) and death (3.7%).” p. 2020. “However, patients who dropped out of the study had more severe airflow obstruction, higher levels of distress, and lower quality of life, as measured with the SGRQ, compared with the patients who had completed the study (data not shown).” p. 2020. Comment: Only patients who completed the 6‐month intervention and follow‐up data collection were analysed.

Low risk of bias

“To limit interviewer bias, each interviewer who obtained the baseline and 6‐month outcome data was blinded to the patient’s treatment group.” p. 2023. "Although self‐reported health‐care utilization may be subject to recall bias, it has been shown to be an acceptable method of measurement and may actually result in an underestimation of utilization with more frequent use.” p. 2023.

Some concerns

Comment: No protocol available. Measures and analysis intentions were only reported in the method section of the paper.

High risk of bias

Unpredictable direction of the outcome.

Rice 2010

Some concerns

"We assigned subjects in equal proportions to each of the two treatment arms by permuted‐block randomisation." Appendix 1, p. 3. "The baseline characteristics of the two treatment groups were generally well matched, with differences in age, lung function, and ED visits (not significant)." p. 891. Comment: Information on the method of allocation concealment was not reported.

High risk of bias

Effect of assignment: "We performed a randomized, adjudicator‐blinded, controlled, 1‐year trial." p. 890. "The status of all 743 patients was determined after 1 year. Electronic medical records were available for all hospitalizations and ED visits that occurred within the VA medical system. The success rate for obtaining discharge summaries and ED records for self‐reported events that occurred outside the VA was 99% in the usual care group and 98% in the disease management group." p. 891. Effect of adhering: Comment: 336 out of 372 (90.3%) patients completed disease management 323 out of 371 (87%) completed usual care.

Low risk of bias

"55% of patients in the usual care group and 60% of patients in the disease management group returned a completed the Saint George’s Respiratory Questionnaire in response to a single mailing at the end of the study. low response rates on SGRQ leading to a high risk of bias. However, data on healthcare utilisation seem complete with no risk of bias." p. 892.

Low risk of bias

"Blinded pulmonologists independently reviewed all discharge summaries and ED reports and assigned a primary cause for each." p. 891.

High risk of bias

Comment: Primary outcome in the paper is reported as a combined outcome measure, whereas it was planned to analyse the two outcomes separately.

High risk of bias

Unpredictable direction of the outcome.

Bucknall 2012

Low risk of bias

“We used a minimisation technique to stratify randomisation of participants by demographic factors (deprivation category of area of residence,11 age and sex, FEV1 per cent predicted at the time of randomisation, smoking status, participation in pulmonary rehabilitation within two years, and number of previous admissions) to control for key aspects of disease severity and predictors of readmission. We constructed a computer generated sequence by using the method of randomised permuted blocks of length four, with two allocations being made at random and two by minimisation. Treatment group allocations were obtained by telephone, after baseline assessments had been made.” p 2. Comment: Random sequence allocation was adequately performed.

High risk of bias

Effect of assignment: Comment: No blinding of patients and personnel. Data for the primary outcome (time to first acute COPD hospital admission or death) were available for all patients and analysed with a Cox proportional hazards model adjusting for the stratification variables. Effect of adhering: Comment: No blinding of patients and personnel. Seventy‐five (42%) of 180 intervention group participants who completed 12 months of follow‐up were judged to have become successful self managers.

High risk of bias

Comment: For secondary outcome measures (e.g. SGRQ, HADS, CSES) not all data were available for all patients randomised (Table 2).

Low risk of bias

‐

Some concerns

‐

High risk of bias

Unpredictable direction of the outcome.

Fan 2012

Low risk of bias

“Randomization lists were generated on the basis of random, permuted blocks of variable size to ensure approximate balance over time.” p. 674. “The CSP Coordinating Center in Boston, Massachusetts, randomly assigned eligible patients in equal numbers to 2 groups, stratifying patients per site to allow for possible regional differences in patient characteristics and clinical practice patterns.” p. 674. Comment: Random sequence generation was adequately performed. The allocation was adequately concealed. Authors didn’t report whether there were any significant differences. Most characteristics are equally balanced, but there is a bit larger difference in the following characteristics (>5%): Currently married Completed high school Long acting inhaled B‐agonist Inhaled corticosteroid Home oxygen use IHD CHF

High risk of bias

Effect of assignment: “The 2 groups differed on the basis of a complex behavioral intervention that made blinding impossible.” p. 674. Comment: No blinding of participants and personnel. "For the primary intention‐to‐treat analysis, the null hypothesis was that the COPD hospitalization hazard rate for intervention did not differ from that for standardized care."p. 675. "Also, with respect to time to first COPD hospitalization, we conducted a futility analysis to assess what might have resulted if the trial had continued." p. 676. "When the study was stopped, the 1‐year cumulative incidence of COPD‐related hospitalization was 27% in the intervention group and 24% in the usual care group (hazard ratio, 1.13 [95% CI, 0.70 to 1.80]; P 0.62)." p. 677. Effect of adhering: "During the entire follow‐up period extending to September 2009, a total of 8 patients in the intervention group and 10 in the usual care group either did not attend scheduled study visits or formally withdrew from the study. Among patients in the intervention group, 87% completed all 4 individual educational visits and 57% completed the group visit. A review of the recorded sessions found that case managers covered 77% to 89% of the educational items in each session. The ranges of average scores across sites that provided audio recordings for each session were 18.3 to 31 for session 1 (13 sites), 12 to 19 for session 2 (12 sites), 12.7 to 23 for session 3 (10 sites), and 6 to 18 for session 4 (10 sites)." p. 677.

High risk of bias

“This multisite, randomized, controlled trial of an educational and acute care management program was stopped early when a safety monitoring board noted more deaths in the intervention group.” p. 674 Comment: There is incomplete outcome data due to early termination of the study.

Low risk of bias

“Telephone‐based ascertainment of study outcomes (COPD hospitalizations and exacerbations) was performed by centralized research staff blinded to assignment. All outcomes were collected by centralized staff blinded to study group, and all hospitalizations were adjudicated by a committee that was also blinded to study group.” p. 674. Comment: Research staff blinded to study group contacted patients every 2 months to determine whether they developed symptoms of a COPD exacerbation, along with details of treatment and health care use. Deaths were ascertained by staff at the study sites, by review of VA electronic medical and administrative records and by a search of the Social Security Death Index. Extensive multivariate analyses failed to identify any factors alone or combined that could plausibly explain the difference in mortality. Two of the authors who were blinded to study group reviewed all available medical information and did not find anything in the deaths that would not be expected in frail patients with COPD.

Some concerns

Comment: The primary and secondary outcomes were reported, only healthcare costs as (secondary objective) were not reported.

High risk of bias

Predicted directon: Favours comparator, because more deaths in the intervention group were observed. There is still a chance that this was because of chance.

Walters 2013

High risk of bias

Randomisation process: "After recruitment, practices were randomised using a code generated by investigators from a random numbers table stratified in blocks of four by Rural, Remote and Metropolitan Areas (RRMA) classifications in Tasmania. Allocation occurred independently using sequentially numbered, opaque and sealed envelopes." p. 2. Timing of identification or recruitment of participants: GPs identified patients with COPD aged over 45 years seen within the previous 12 months through database searches, based on a diagnostic code for COPD or prescription of tiotropium. Comment: Most variables are equally distributed, but more smokers (not statistically significant, p=0.102) and more previous admissions in the intervention group (statistically significant, p=0.028), so potentially more room for improvement. Although, this small number of differences identified as ‘statistically significant’ at the conventional 0.05 threshold should usually be considered to be compatible with chance.

High risk of bias

Effect of assignment: ‐ Effect of adhering: "The limitations of the study were the high rate of misclassification of COPD in general practices, which impacted on enrolment, and higher than expected withdrawal rates, especially in the HM group, which reduced the intended statistical power." p. 6. Comment: Nine patients discontinued and 8 patients withdrew in the intervention group.

Low risk of bias

Comment: Figure 1: 9 intervention patients discontinued, but they attended follow‐up. In Table 2, no patient numbers are mentioned.

Low risk of bias

Comment: Blinding of participants or research officers was not possible given the nature of the study.

Low risk of bias

Comment: Data that produced this result analysed in accordance with a pre‐specified analysis plan.

High risk of bias

Unpredictable direction of the outcome.

Hernández 2015

Low risk of bias

"A computer‐generated list of random numbers with no restrictions and administered by personnel who were not involved in the study ensured blinded randomisation." p. 2. "(…) and administered by personnel who were not involved in the study." p. 2. Comment: Random sequence generation was adequately performed. No blinding of participants or personnel. The patients of the two groups showed similar characteristics, except that influenza and pneumococcal vaccination were more prevalent in the IC group (P<0.01).

High risk of bias

Effect of assignment: Comment: It was not possible to blind the treatment for patients and people delivering the treatment. Intention‐to‐treat analysis performed. Effect of adhering: "As a sensitivity analysis, we repeated all analyses after excluding patients who had a previous contact with any IC programme (n = 15)." p. 2. "25 patients (22%) died and 16 (14%) were lost during the 1‐year follow‐up, and the remaining 114 patients (IC, n = 59 and UC, n = 55) completed the 12‐month assessment visit. Patients lost to follow‐up exhibited a higher affectation of activities of daily living and health‐related quality of life at baseline than those who were followed up, without differences in other sociodemographic, clinical, functional and medical care‐related variables." p. 3. Comment: It is not mentioned in which way these patients were distributed over the two groups, nor are the results of the sensitivity analysis reported. Although the approach was not adequately adopted at the community level and thus the intervention was not continued after the 1 year of follow‐up, this was not a failure of implementing the intervention, since the intervention stopped at 1 year according to the protocol.

High risk of bias

Comment: Refusal of n=5 in the intervention group. Lost to follow‐up 12 in intervention group and 29 in usual care group. So in total 22% died and 14% were lost to follow‐up. Patients lost to follow‐up exhibited a higher affectation of activities of daily living and health‐related quality of life at baseline than those who were followed up, (no differences in other sociodemographic, clinical, functional and medical care‐related variables) this could have biased the results between the intervention group and usual care group.

Low risk of bias

"A blind evaluation of the study group carried out before randomisation and after the 12‐month follow‐up consisted of a patient interview and analysis of medical records, self‐administered questionnaires and lung function testing." p. 2.

Some concerns

"Although the study was officially registered as part of the EU Grant (CIP‐ICT‐PSP‐2007.225025), the RCT was not included in the clinicaltrials.gov registry because at that time it was not compulsory." p. 5. Comment: No protocol available.

High risk of bias

Direction: favours experimental. Comment: More patients were lost to follow‐up in the usual care group and patients lost to follow‐up exhibited a higher affectation of activities of daily living and health‐related quality of life at baseline than those who were followed up.

Jonsdottir 2015

Low risk of bias

"After accepting the invitation, prior to entering the research centre, the research nurses randomized the patients to experimental and control groups in accordance with CONSORT requirements using a computer‐generated list of random numbers." p. 2638.

High risk of bias

Effect of assignment: Comment: Not possible to blind the treatment for patients and people delivering the treatment. Intention‐to‐treat analysis was not carried out as stated in the article, but in table 3 all patients that completed the follow‐up seem to be included. Effect of adhering: "To minimize treatment bias the data were concealed from the nurses providing the treatment and the treatment notes were concealed from the nurses conducting data collection." p. 2640. "Both groups followed standardized protocols and had regular meetings to maintain consistency. Few family members participated. This actually changed the nature of the study from being a family‐centered intervention as planned, into becoming focused on patients, with family involvement when possible." p. 2640. "Attendance at treatment sessions in the experimental group was more than 80% in the patient/family conversations (43/48 attended three or four out of the four possible meetings) and the group meeting (40/48). Half of the patients in the experimental group smoked (24/48). Of those, 58% (14/24) participated in the smoking cessation treatment." p. 2641.

High risk of bias

"Three patients died in the experimental group... The dropout rate was 20% (n = 12) in the experimental group and 12% (n = 7) in the control group. Also additional missing data on outcome variables as displayed in table 3. The dropout rate for family members was 31% (n = 5) and 50% (n = 7) in the experimental and control groups respectively." p. 2640. Comment: No info on dropout reasons or characteristics of dropouts.

Low risk of bias

"To minimize treatment bias the data were concealed from the nurses providing the treatment and the treatment notes were concealed from the nurses conducting data collection." p. 2640.

High risk of bias

No information, not registered in a trial register.

High risk of bias

Unpredictable direction of the outcome.

Titova 2015

High risk of bias

Randomisation process: “They were randomly allocated to either integrated care (IC) or usual care (UC) based on their address of permanent residence. In order to create two pairs of districts with approximately equal numbers of citizens, a pair‐wise matching of districts was carried out. It was decided by lottery that participants from District Pair 1 were assigned to the UC group, and participants from District Pair 2 were assigned to the IC group.” Titova 2015, p. 2. “There were no significant differences in baseline characteristics between the IC and the UC groups, but when comparing those who died during follow‐up with those who were still alive after two years, those who died had a significantly higher PaCO2 and lower BMI at baseline (Table 1). Those who died during follow‐up also had significantly more HA and HD during the year prior to study start.” Titova 2015, p. 4. Comment: No information provided about the allocation concealment. Timing of identification or recruitment of participants: "They were randomly allocated to either IC or UC group based on their address of permanent residence. Patient data were registered at discharge and 6, 12, and 24 months after discharge from the TUH." Titova 2017, p 3. Comment: If allocation was known before study inclusion, suitable patients could be selected for the intervention.

High risk of bias

Effect of assignment: “The study was a prospective, open, single‐centre intervention study.” Titova 2015, p. 2. "The principal comparisons of changes were performed on an intention‐to treat basis." Titova 2017, p. 4. Comment: No blinding of participants and personnel. Effect of adhering: ‐

High risk of bias

“172 patients were randomly allocated to the IC group (n = 91) or the UC group (n = 81), respectively. After two years of follow‐up, a total of 100 patients (58% of the included patients), 51 patients in the IC group and 49 patients in the UC group, were available for evaluation (Figure 1).” Titova 2017, p. 5. "Those who died had a significantly higher PaCO2 and lower BMI at baseline (Table 1). Those who died during follow‐up also had significantly more HA and HD during the year prior to study start." Titova 2015, p. 4. Comment: No mortality reported; however, Figure 1 shows higher mortality for the IC group, n=35 (38.4%), compared to the UC group, n=21 (25.9%).

High risk of bias

“Information concerning the number and duration of the COPD exacerbations, as well as the time from onset of symptoms until the start of self‐initiated treatment is insufficient due to many incomplete registrations in “My COPD book.” Titova 2015, p. 9. "“Data concerning HA (hospital admissions) and HD (hospital days) were collected from the hospital registry database’s medical charts.” Titova 2015, p. 3. Comment: Unclear who was the outcome assessor.

Some concerns

Trial register reported number of hospitalisations as the primary outcome. Secondary outcomes were quality of life, lung function, use of medication, cost‐effectiveness. However, the article reported not all secondary outcomes (e.g. cost‐effectiveness).

High risk of bias

Unpredictable direction of the outcome.

Jolly 2018

High risk of bias

"When a patient is identified as eligible for the study, and has given written, informed consent to take part, the researcher randomises the patient using a web‐based programme hosted by the Primary Care Clinical Trials Unit (PC‐CTU), University of Birmingham. Centre specific randomisation lists were produced by a statistician at the trials unit. Participants are individually randomised, stratified by centre. The four recruitment centres are Birmingham and West‐Midlands South; Greater Manchester; North West Midlands and Oxfordshire/Gloucestershire. Only the PC‐CTU have access to the allocation sequence." Sidhu 2015 (protocol paper), p. 4. "The study groups were generally well balanced in terms of patient characteristics, although there was a higher proportion of current smokers in the telephone health coaching group. The usual care group reported a higher level of self reported moderate and vigorous physical activity, but this was not reflected in the accelerometry data at baseline." Jolly 2018 (results paper), p. 5.

High risk of bias

Effect of assignment: Comment: Patients were informed of their allocation at the end of the recruitment appointment; they were not masked to treatment allocation. All data were analysed by intention to treat. Effect of adhering: "We achieved good fidelity of delivery of the intervention, with 75% of intervention participants receiving all four calls and only four patients receiving none." p. 11. "The dose and coverage of intervention delivery were high: 86.4% (999/1156) of the scheduled calls were delivered and 75.4% (218/289) of participants received all four calls." No information regarding pulmonary rehabilitation, oxygen use.

High risk of bias

"There was imbalance in the follow‐up rates between telephone health coaching (82.7%; 37 withdrawals) and usual care (96.2%; 7 withdrawals)." p. 4. "Participants who did not provide data at 12 months were more likely to be in GOLD stage 3, to be smokers, had lower levels of self‐reported physical activity, and to live alone than responders. " p. 5. Comment: For health‐related quality of life, data were available for 75% of intervention patients (217 of 289), although some sensitivity analyses were performed.

Some concerns

"Patients were informed of their allocation at the end of the recruitment appointment; they were not masked to treatment allocation. Data were entered into the study database by researchers at the University of Birmingham who were masked to the treatment allocation." p. 3. Comment: No information about outcome assessors.

Low risk of bias

‐

High risk of bias

Unpredictable direction of the outcome.

Kessler 2018

Low risk of bias

"Patients were allocated to groups in a 1:1 fashion according to a pre‐specified randomisation list generated before the study by a partial‐minimisation computer algorithm under supervision of the study sponsor. Patients were assigned a randomisation number by study staff at each centre in sequential numerical order through a telephone‐based interactive voice response system. Randomisation was stratified by smoking status (current or former), need for respiratory assistance (none, or on LTOT and/or HMV), and centre." p. 3.

High risk of bias

Effect of assignment: "For practical reasons, the study was open; neither the patients nor the investigators were blinded to the COPD management strategy." p. 3. "The primary analysis of the primary outcome was performed in the intention‐to‐treat (ITT) population (all patients who entered the follow‐up period). The number of hospitalisation days fitting the established criteria was normalised on a yearly basis for each patient and compared between both groups using the non‐parametric Wilcoxon rank‐sum test." p. 4. "The open design of the study, though necessary owing to the extensive differences between the two disease management programmes, may have led to some bias because doctors could have had a tendency to discharge patients in the disease management arm earlier than those in the usual management arm." p. 12. Effect of adhering: Comment: 15 patients (<10%) received <25% self‐management training.

Low risk of bias

"…leaving 157 (91%) and 162 (94%) patients in the ITT population, 137 (80%) and 128 (74%) of whom completed at least 12 months of follow‐up." p. 5.

Low risk of bias

"Because hospitalisation decisions might have been subject to local practices, social considerations, bed availability, and so on, all hospitalisations were reviewed by an independent, three‐member end‐point validation committee (EVC) to assess the reason and appropriateness of each hospitalisation, determine whether the hospitalisation fulfilled the protocol definition, and determine the extent to which it was to be considered in the efficacy assessments (fully, partially, not at all)." p. 3.

Low risk of bias

"During the study, the protocol was amended to accelerate recruitment and allow study completion (supplementary material)." p. 3. Comment: Inclusion criteria were expanded from COPD patients receiving LTOT to patients with severe COPD, with or without LTOT; the minimum number of patients was reduced from 520 to 306 by retaining the initial assumptions for sample size and removing a multiple comparisons adjustment; the follow‐up period was reduced from 2 years to 1 year; and the contact frequency between DM patients and CMs was increased to improve patient training (either group or phone session, once per month). Prior to these changes, 72 patients on LTOT were enrolled in the study.

High risk of bias

Unpredictable direction of the outcome.

Wang 2019

Low risk of bias

"The allocation sequence was generated and released to the interventionist on a case‐by‐case basis by an independent department that specialised in supplying randomly generated sequences for research. The randomisation sequence was recorded." p. 151. "Participants were randomly assigned to groups after consent was obtained and baseline data were collected." p. 151. "There were no significant differences in demographic and clinical data between the two groups at baseline (Table 1)." p. 151.

High risk of bias

Effect of assignment: Comment: It is not possible to blind the treatment for patients and people delivering the treatment. Intention‐to‐treat analysis was used. Effect of adhering: "154 participants were enrolled, among whom, 143 completed the 12‐month study program. The attrition rate was 7.14%." p. 151. "...adherence to the self‐management intervention was reported by the participants but not monitored by the nurses." p. 155. Comment: No info on reasons of adherence by patients reported in the current publication.

Low risk of bias

Comment: Relatively low attrition rate.

Low risk of bias

"To minimize researcher bias, the interventionists who provided care were blinded to the participants’ baseline and allocation sequence. The statistician was blinded to the participants’ results during the trial." p. 151. Comment: Not possible to blind the treatment for patients and people delivering the treatment.

Some concerns

Comment: No pre‐specified analysis plan published which can be checked. However they mention: The statistician was blinded to the participants’ results during the trial.

High risk of bias

Unpredictable direction of the outcome.

Liang 2019

High risk of bias

"Group or solo GP clinics with ≥1000 patients on their databases were approached directly or with assistance from primary care organisations. After obtaining signed agreement, clinics were block randomised (block sizes of four and six) to usual care or intervention arms using an externally managed web‐based randomisation programme. Clinics were notified of their allocation." p. 3.

High risk of bias

Effect of assignment: ‐ Effect of adhering: "67 participants (43%) did not receive any part of the intervention." p. 7. "A key limitation is the low uptake of the intervention components." p. 10.

High risk of bias

Comment: Data were not available for all patients randomised.

Low risk of bias

"The effective analysis was according to an intention‐to‐treat principle." p. 4.

Low risk of bias

Results were analysed according to the published design paper.

High risk of bias

Predicted direction: Favours comparator, because general practices could exclude patients from selection to participate in the intervention due to the cluster randomisation approach. Adherence to the intervention was low.

Risk of bias for analysis 1.3 Health‐related quality of life (HRQoL): CRQ domain scores (primary analysis).

Study

Bias

Randomisation process

Deviations from intended interventions

Missing outcome data

Measurement of the outcome

Selection of the reported results

Overall

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Subgroup 1.3.1 HRQoL: CRQ ‐ dyspnoea

Bischoff 2012

Some concerns

"We randomised participants by using a computer generated two block randomisation procedure with stratification on severity of COPD (mild or moderate v severe airflow obstruction), smoking status (current v former smoker), and frequency of exacerbations in the previous 24 months (<2 v ≥ 2 exacerbations)". p. 2. "Table 1 shows that at inclusion in the study, patients’ characteristics were well balanced between the three study groups, except for sex.". p. 7. Comment: Random sequence generation was adequately performed. No information on who performed the allocation. There is another paper published, with different patient numbers (Uijen, British Journal of General Practice, June 2012), this paper shows a higher number of patients in the intervention group (n= 64) compared to the Bischoff’s paper (n= 55), but a lower number of patients in the usual care group (n= 49) compared to the Bischoff’s paper (n= 55). This raised some concerns about the randomisation process. Clarification could not be given by the authors.

High risk of bias

Effect of assignment: "Our primary analysis was based on the intention to treat principle and included all available data for all participants. We compared self management with usual care and routine monitoring with usual care. We did not impute any missing data." p. 3. Comment: It is not possible to blind patients / carers /people delivering the intervention due to the nature of this intervention. Effect of adhering: "“We prevented potential treatment contamination caused by provision of self management, routine monitoring, and usual care within the same." p. 2. "Seven (13%) patients did not receive any session or telephone contact, and 48 (87%) received two or more sessions." p. 4. "Six (11%) patients did not receive any consultation, and 42 (76%) patients received two or more consultations." p. 4. Comment: There was some non‐adherence in intervention group (at least 10%), this may all have led to some effect in the usual care group and a smaller difference between the intervention and usual care group.

Some concerns

Comment: Almost 16% of the participants dropped out during follow‐up (intervention 11%; usual care 20%). However, baseline characteristics did not differ between dropouts and participants who finished follow‐up. Exclusion is well described in flow chart. Intention‐to‐treat analyses were used.

Low risk of bias

Comment: The outcome assessor was blinded (p. 2). Outcome assessment was performed with standardised questionnaires and a telephonic exacerbation assessment system (TEXAS).

Some concerns

Comments: The trial register includes outcomes that haven’t been reported in this paper, but the outcomes that are reported in this paper have been mentioned in the trial register.

High risk of bias

Unpredictable direction of the outcome.

Mitchell 2014

Low risk of bias

“Patients were assigned to either usual care or SPACE FOR COPD via a web‐based, concealed allocation programme, using simple randomisation codes prepared by the trial statistician (J. Bankart).” p. 1539. "Randomisation was conducted by the trial investigator responsible for administering the intervention (K.E. Mitchell).” p. 1539. Comment: Random sequence allocation was adequately performed. The method of allocation concealment was not reported, but they say it is concealed. The only significant baseline difference between treatment groups was the CRQ‐SR dyspnoea score; therefore, this was corrected for in subsequent analyses.

High risk of bias

Effect of assignment: “Lack of participant blinding may have increased motivation when receiving the treatment and attempts to satisfy the researchers might have increased the observed treatment effects in the intervention arm. We cannot, therefore, rule out the possible impact of attention.” p. 1546. Comment: No blinding of participants. Analysis was carried out on an intention‐to‐treat basis. Effect of adhering: Comment: Withdrawal around intervention group: 27% and UC: 17%. The adherence of using the manual in by the intervention patients was not reported.

Low risk of bias

Comment: Data were not available for all randomised patients (withdrawal intervention: 27%, usual care: 17%). However, they found no significant differences in demographics or baseline variables between those who completed and those who did not complete the study.

Low risk of bias

“The assessments at week 6 and 6 months were conducted by a member of the research team who was blind to randomisation allocation (V. Johnson‐Warrington)." p. 1540 Comment: The outcome assessment was blinded.

Low risk of bias

Comment: No signs for selective outcome reporting. The primary outcome measure and most of the secondary outcome measures were reported.

High risk of bias

Unpredictable direction of the outcome.

Johnson‐Warrington 2016

Low risk of bias

"Participants were randomized to receive usual care or SPACE for COPD via a web‐based, concealed allocation program (www.sealedenvelope.com) using simple random permuted block 1:1 randomization by VJ‐W. Randomization was performed after the participants completed the baseline assessment, with treatment allocation prior to hospital discharge." p. 1162.

Some concerns

Effect of assignment: "Analysis was conducted on an intention‐to‐treat basis." p. 1163. Effect of adhering: Patients and personnel were not blinded. No information regarding adherence reported.

High risk of bias

Some patients were missing: n= 78 were randomized (n= 39 to each arm). Included in the analysis: Intervention: n= 35 and usual care: n= 36.

Low risk of bias

"All outcomes were measured at baseline (during admission but as close to discharge as possible) and 3 months after randomization, by a clinician blinded to treatment allocation." p. 1163.

Some concerns

No protocol including a pre‐specified analysis plan was available.

High risk of bias

Unpredictable direction of the outcome.

Benzo 2016

Low risk of bias

Comment: Patients were randomly assigned, using an online, computer‐generated, simple binomial randomization program to one of the two groups, stratified by center.

High risk of bias

Effect of assignment: "85% of the individuals in the intervention group received the intervention as intended, defined as 15 (.70%) of 21 calls completed. Reasons for not completing the intervention were death during the study period (n = 3), unable to contact (n = 6), and refusal to complete the scheduled calls (n = 7)." p. 674. Comment: It was possible to blind the treatment for patients and people delivering the treatment. Intention‐to‐treat analysis was used. Effect of adhering: "Attendance at pulmonary rehabilitation visits anytime in the first 12 months after discharge tended to be higher in the intervention group (53% vs. 43%; P = 0.056)." p. 675. "In the intervention group, 85% of the individuals received a complete intervention, defined as 15 (.70%) of 21 calls completed." p. 674.

Low risk of bias

‐

Low risk of bias

Comment: Outcome assessors were blinded.

High risk of bias

Comment: No analysis plan available in the trial register. Furthermore, the trial register mentions that the outcomes will be assessed in the 12 months time frame, but not mentioned how and when they were assessed. Although 12 months analyses show no differences, the additional time points that were reported (not prespecified) are in favour of experimental.

High risk of bias

Predicted direction: Favours experimental.

Lenferink 2019

Low risk of bias

Comment: Allocation was stratified on potential confounders. Only a baseline difference in 6MWD was reported.

High risk of bias

Effect of assignment: Comment: It was not possible to blind the treatment for patients and people delivering the treatment. Intention‐to‐treat analyses were performed. Effect of adhering: "The 12‐month follow‐up was completed by 83.3% of the individuals in the self‐management group and 84.8% in the UC group (figure 1). Patients completed 81.3% (self‐management: 82.3%; UC: 80.2%) of the diary data." p. 6. Comment: The effect of adherence to the intervention was not analysed. However, it was observed that a high number of patients in completed follow‐up and their symptom diaries. Adherence to action plans was measured, but not yet reported in the current publication.

Low risk of bias

Comment: Relatively low number of drop‐outs. The imputated data analyses showed similar results as the observed data analyses.

Low risk of bias

"Wherever possible, though, assessors of outcomes were blinded to the treatment group." p. 3.

Low risk of bias

Comment: Some small changes in final analyses compared to analysis plan that was published with the study protocol.

High risk of bias

Additional bias: "Based on the exacerbation rates/patient/day from the COPE‐II study (intervention: 0.116; control: 0.176, both with an estimated standard deviation of 0.17) and allowing for overdispersion and an attrition rate of 10%, 105 patients per group provided 80% power to detect an effect of this size, with a two‐tailed p<0.05." p. 4. Comment: The study included less patients (intervention: n= 102; usual care: n= 99) than aimed by power calculation (both groups n= 105). Unpredicted direction of the outcome.

Subgroup 1.3.2 HRQoL: CRQ ‐ mastery

Bischoff 2012

Some concerns

"We randomised participants by using a computer generated two block randomisation procedure with stratification on severity of COPD (mild or moderate v severe airflow obstruction), smoking status (current v former smoker), and frequency of exacerbations in the previous 24 months (<2 v ≥ 2 exacerbations)". p. 2. "Table 1 shows that at inclusion in the study, patients’ characteristics were well balanced between the three study groups, except for sex.". p. 7. Comment: Random sequence generation was adequately performed. No information on who performed the allocation. There is another paper published, with different patient numbers (Uijen, British Journal of General Practice, June 2012), this paper shows a higher number of patients in the intervention group (n= 64) compared to the Bischoff’s paper (n= 55), but a lower number of patients in the usual care group (n= 49) compared to the Bischoff’s paper (n= 55). This raised some concerns about the randomisation process. Clarification could not be given by the authors.

High risk of bias

Effect of assignment: "Our primary analysis was based on the intention to treat principle and included all available data for all participants. We compared self management with usual care and routine monitoring with usual care. We did not impute any missing data." p. 3. Comment: It is not possible to blind patients / carers /people delivering the intervention due to the nature of this intervention. Effect of adhering: "“We prevented potential treatment contamination caused by provision of self management, routine monitoring, and usual care within the same." p. 2. "Seven (13%) patients did not receive any session or telephone contact, and 48 (87%) received two or more sessions." p. 4. "Six (11%) patients did not receive any consultation, and 42 (76%) patients received two or more consultations." p. 4. Comment: There was some non‐adherence in intervention group (at least 10%), this may all have led to some effect in the usual care group and a smaller difference between the intervention and usual care group.

Some concerns

Comment: Almost 16% of the participants dropped out during follow‐up (intervention 11%; usual care 20%). However, baseline characteristics did not differ between dropouts and participants who finished follow‐up. Exclusion is well described in flow chart. Intention‐to‐treat analyses were used.

Low risk of bias

Comment: The outcome assessor was blinded (p. 2). Outcome assessment was performed with standardised questionnaires and a telephonic exacerbation assessment system (TEXAS).

Some concerns

Comments: The trial register includes outcomes that haven’t been reported in this paper, but the outcomes that are reported in this paper have been mentioned in the trial register.

High risk of bias

Unpredictable direction of the outcome.

Mitchell 2014

Low risk of bias

“Patients were assigned to either usual care or SPACE FOR COPD via a web‐based, concealed allocation programme, using simple randomisation codes prepared by the trial statistician (J. Bankart).” p. 1539. "Randomisation was conducted by the trial investigator responsible for administering the intervention (K.E. Mitchell).” p. 1539. Comment: Random sequence allocation was adequately performed. The method of allocation concealment was not reported, but they say it is concealed. The only significant baseline difference between treatment groups was the CRQ‐SR dyspnoea score; therefore, this was corrected for in subsequent analyses.

High risk of bias

Effect of assignment: “Lack of participant blinding may have increased motivation when receiving the treatment and attempts to satisfy the researchers might have increased the observed treatment effects in the intervention arm. We cannot, therefore, rule out the possible impact of attention.” p. 1546. Comment: No blinding of participants. Analysis was carried out on an intention‐to‐treat basis. Effect of adhering: Comment: Withdrawal around intervention group: 27% and UC: 17%. The adherence of using the manual in by the intervention patients was not reported.

Low risk of bias

Comment: Data were not available for all randomised patients (withdrawal intervention: 27%, usual care: 17%). However, they found no significant differences in demographics or baseline variables between those who completed and those who did not complete the study.

Low risk of bias

“The assessments at week 6 and 6 months were conducted by a member of the research team who was blind to randomisation allocation (V. Johnson‐Warrington)." p. 1540 Comment: The outcome assessment was blinded.

Low risk of bias

Comment: No signs for selective outcome reporting. The primary outcome measure and most of the secondary outcome measures were reported.

High risk of bias

Unpredictable direction of the outcome.

Benzo 2016

Low risk of bias

Comment: Patients were randomly assigned, using an online, computer‐generated, simple binomial randomization program to one of the two groups, stratified by center.

High risk of bias

Effect of assignment: "85% of the individuals in the intervention group received the intervention as intended, defined as 15 (.70%) of 21 calls completed. Reasons for not completing the intervention were death during the study period (n = 3), unable to contact (n = 6), and refusal to complete the scheduled calls (n = 7)." p. 674. Comment: It was possible to blind the treatment for patients and people delivering the treatment. Intention‐to‐treat analysis was used. Effect of adhering: "Attendance at pulmonary rehabilitation visits anytime in the first 12 months after discharge tended to be higher in the intervention group (53% vs. 43%; P = 0.056)." p. 675. "In the intervention group, 85% of the individuals received a complete intervention, defined as 15 (.70%) of 21 calls completed." p. 674.

Low risk of bias

‐

Low risk of bias

Comment: Outcome assessors were blinded.

High risk of bias

Comment: No analysis plan available in the trial register. Furthermore, the trial register mentions that the outcomes will be assessed in the 12 months time frame, but not mentioned how and when they were assessed. Although 12 months analyses show no differences, the additional time points that were reported (not prespecified) are in favour of experimental.

High risk of bias

Predicted direction: Favours experimental.

Johnson‐Warrington 2016

Low risk of bias

"Participants were randomized to receive usual care or SPACE for COPD via a web‐based, concealed allocation program (www.sealedenvelope.com) using simple random permuted block 1:1 randomization by VJ‐W. Randomization was performed after the participants completed the baseline assessment, with treatment allocation prior to hospital discharge." p. 1162.

Some concerns

Effect of assignment: "Analysis was conducted on an intention‐to‐treat basis." p. 1163. Effect of adhering: Patients and personnel were not blinded. No information regarding adherence reported.

High risk of bias

Some patients were missing: n= 78 were randomized (n= 39 to each arm). Included in the analysis: Intervention: n= 35 and usual care: n= 36.

Low risk of bias

"All outcomes were measured at baseline (during admission but as close to discharge as possible) and 3 months after randomization, by a clinician blinded to treatment allocation." p. 1163.

Some concerns

No protocol including a pre‐specified analysis plan was available.

High risk of bias

Unpredictable direction of the outcome.

Lenferink 2019

Low risk of bias

Comment: Allocation was stratified on potential confounders. Only a baseline difference in 6MWD was reported.

High risk of bias

Effect of assignment: Comment: It was not possible to blind the treatment for patients and people delivering the treatment. Intention‐to‐treat analyses were performed. Effect of adhering: "The 12‐month follow‐up was completed by 83.3% of the individuals in the self‐management group and 84.8% in the UC group (figure 1). Patients completed 81.3% (self‐management: 82.3%; UC: 80.2%) of the diary data." p. 6. Comment: The effect of adherence to the intervention was not analysed. However, it was observed that a high number of patients in completed follow‐up and their symptom diaries. Adherence to action plans was measured, but not yet reported in the current publication.

Low risk of bias

Comment: Relatively low number of drop‐outs. The imputated data analyses showed similar results as the observed data analyses.

Low risk of bias

"Wherever possible, though, assessors of outcomes were blinded to the treatment group." p. 3.

Low risk of bias

Comment: Some small changes in final analyses compared to analysis plan that was published with the study protocol.

High risk of bias

Additional bias: "Based on the exacerbation rates/patient/day from the COPE‐II study (intervention: 0.116; control: 0.176, both with an estimated standard deviation of 0.17) and allowing for overdispersion and an attrition rate of 10%, 105 patients per group provided 80% power to detect an effect of this size, with a two‐tailed p<0.05." p. 4. Comment: The study included less patients (intervention: n= 102; usual care: n= 99) than aimed by power calculation (both groups n= 105). Unpredicted direction of the outcome.

Subgroup 1.3.3 HRQoL: CRQ ‐ fatigue

Bischoff 2012

Some concerns

"We randomised participants by using a computer generated two block randomisation procedure with stratification on severity of COPD (mild or moderate v severe airflow obstruction), smoking status (current v former smoker), and frequency of exacerbations in the previous 24 months (<2 v ≥ 2 exacerbations)". p. 2. "Table 1 shows that at inclusion in the study, patients’ characteristics were well balanced between the three study groups, except for sex.". p. 7. Comment: Random sequence generation was adequately performed. No information on who performed the allocation. There is another paper published, with different patient numbers (Uijen, British Journal of General Practice, June 2012), this paper shows a higher number of patients in the intervention group (n= 64) compared to the Bischoff’s paper (n= 55), but a lower number of patients in the usual care group (n= 49) compared to the Bischoff’s paper (n= 55). This raised some concerns about the randomisation process. Clarification could not be given by the authors.

High risk of bias

Effect of assignment: "Our primary analysis was based on the intention to treat principle and included all available data for all participants. We compared self management with usual care and routine monitoring with usual care. We did not impute any missing data." p. 3. Comment: It is not possible to blind patients / carers /people delivering the intervention due to the nature of this intervention. Effect of adhering: "“We prevented potential treatment contamination caused by provision of self management, routine monitoring, and usual care within the same." p. 2. "Seven (13%) patients did not receive any session or telephone contact, and 48 (87%) received two or more sessions." p. 4. "Six (11%) patients did not receive any consultation, and 42 (76%) patients received two or more consultations." p. 4. Comment: There was some non‐adherence in intervention group (at least 10%), this may all have led to some effect in the usual care group and a smaller difference between the intervention and usual care group.

Some concerns

Comment: Almost 16% of the participants dropped out during follow‐up (intervention 11%; usual care 20%). However, baseline characteristics did not differ between dropouts and participants who finished follow‐up. Exclusion is well described in flow chart. Intention‐to‐treat analyses were used.

Low risk of bias

Comment: The outcome assessor was blinded (p. 2). Outcome assessment was performed with standardised questionnaires and a telephonic exacerbation assessment system (TEXAS).

Some concerns

Comments: The trial register includes outcomes that haven’t been reported in this paper, but the outcomes that are reported in this paper have been mentioned in the trial register.

High risk of bias

Unpredictable direction of the outcome.

Mitchell 2014

Low risk of bias

“Patients were assigned to either usual care or SPACE FOR COPD via a web‐based, concealed allocation programme, using simple randomisation codes prepared by the trial statistician (J. Bankart).” p. 1539. "Randomisation was conducted by the trial investigator responsible for administering the intervention (K.E. Mitchell).” p. 1539. Comment: Random sequence allocation was adequately performed. The method of allocation concealment was not reported, but they say it is concealed. The only significant baseline difference between treatment groups was the CRQ‐SR dyspnoea score; therefore, this was corrected for in subsequent analyses.

High risk of bias

Effect of assignment: “Lack of participant blinding may have increased motivation when receiving the treatment and attempts to satisfy the researchers might have increased the observed treatment effects in the intervention arm. We cannot, therefore, rule out the possible impact of attention.” p. 1546. Comment: No blinding of participants. Analysis was carried out on an intention‐to‐treat basis. Effect of adhering: Comment: Withdrawal around intervention group: 27% and UC: 17%. The adherence of using the manual in by the intervention patients was not reported.

Low risk of bias

Comment: Data were not available for all randomised patients (withdrawal intervention: 27%, usual care: 17%). However, they found no significant differences in demographics or baseline variables between those who completed and those who did not complete the study.

Low risk of bias

“The assessments at week 6 and 6 months were conducted by a member of the research team who was blind to randomisation allocation (V. Johnson‐Warrington)." p. 1540 Comment: The outcome assessment was blinded.

Low risk of bias

Comment: No signs for selective outcome reporting. The primary outcome measure and most of the secondary outcome measures were reported.

High risk of bias

Unpredictable direction of the outcome.

Benzo 2016

Low risk of bias

Comment: Patients were randomly assigned, using an online, computer‐generated, simple binomial randomization program to one of the two groups, stratified by center.

High risk of bias

Effect of assignment: "85% of the individuals in the intervention group received the intervention as intended, defined as 15 (.70%) of 21 calls completed. Reasons for not completing the intervention were death during the study period (n = 3), unable to contact (n = 6), and refusal to complete the scheduled calls (n = 7)." p. 674. Comment: It was possible to blind the treatment for patients and people delivering the treatment. Intention‐to‐treat analysis was used. Effect of adhering: "Attendance at pulmonary rehabilitation visits anytime in the first 12 months after discharge tended to be higher in the intervention group (53% vs. 43%; P = 0.056)." p. 675. "In the intervention group, 85% of the individuals received a complete intervention, defined as 15 (.70%) of 21 calls completed." p. 674.

Low risk of bias

‐

Low risk of bias

Comment: Outcome assessors were blinded.

High risk of bias

Comment: No analysis plan available in the trial register. Furthermore, the trial register mentions that the outcomes will be assessed in the 12 months time frame, but not mentioned how and when they were assessed. Although 12 months analyses show no differences, the additional time points that were reported (not prespecified) are in favour of experimental.

High risk of bias

Predicted direction: Favours experimental.

Johnson‐Warrington 2016

Low risk of bias

"Participants were randomized to receive usual care or SPACE for COPD via a web‐based, concealed allocation program (www.sealedenvelope.com) using simple random permuted block 1:1 randomization by VJ‐W. Randomization was performed after the participants completed the baseline assessment, with treatment allocation prior to hospital discharge." p. 1162.

Some concerns

Effect of assignment: "Analysis was conducted on an intention‐to‐treat basis." p. 1163. Effect of adhering: Patients and personnel were not blinded. No information regarding adherence reported.

High risk of bias

Some patients were missing: n= 78 were randomized (n= 39 to each arm). Included in the analysis: Intervention: n= 35 and usual care: n= 36.

Low risk of bias

"All outcomes were measured at baseline (during admission but as close to discharge as possible) and 3 months after randomization, by a clinician blinded to treatment allocation." p. 1163.

Some concerns

No protocol including a pre‐specified analysis plan was available.

High risk of bias

Unpredictable direction of the outcome.

Lenferink 2019

Low risk of bias

Comment: Allocation was stratified on potential confounders. Only a baseline difference in 6MWD was reported.

High risk of bias

Effect of assignment: Comment: It was not possible to blind the treatment for patients and people delivering the treatment. Intention‐to‐treat analyses were performed. Effect of adhering: "The 12‐month follow‐up was completed by 83.3% of the individuals in the self‐management group and 84.8% in the UC group (figure 1). Patients completed 81.3% (self‐management: 82.3%; UC: 80.2%) of the diary data." p. 6. Comment: The effect of adherence to the intervention was not analysed. However, it was observed that a high number of patients in completed follow‐up and their symptom diaries. Adherence to action plans was measured, but not yet reported in the current publication.

Low risk of bias

Comment: Relatively low number of drop‐outs. The imputated data analyses showed similar results as the observed data analyses.

Low risk of bias

"Wherever possible, though, assessors of outcomes were blinded to the treatment group." p. 3.

Low risk of bias

Comment: Some small changes in final analyses compared to analysis plan that was published with the study protocol.

High risk of bias

Additional bias: "Based on the exacerbation rates/patient/day from the COPE‐II study (intervention: 0.116; control: 0.176, both with an estimated standard deviation of 0.17) and allowing for overdispersion and an attrition rate of 10%, 105 patients per group provided 80% power to detect an effect of this size, with a two‐tailed p<0.05." p. 4. Comment: The study included less patients (intervention: n= 102; usual care: n= 99) than aimed by power calculation (both groups n= 105). Unpredicted direction of the outcome.

Subgroup 1.3.4 HRQoL: CRQ ‐ emotional function

Bischoff 2012

Some concerns

"We randomised participants by using a computer generated two block randomisation procedure with stratification on severity of COPD (mild or moderate v severe airflow obstruction), smoking status (current v former smoker), and frequency of exacerbations in the previous 24 months (<2 v ≥ 2 exacerbations)". p. 2. "Table 1 shows that at inclusion in the study, patients’ characteristics were well balanced between the three study groups, except for sex.". p. 7. Comment: Random sequence generation was adequately performed. No information on who performed the allocation. There is another paper published, with different patient numbers (Uijen, British Journal of General Practice, June 2012), this paper shows a higher number of patients in the intervention group (n= 64) compared to the Bischoff’s paper (n= 55), but a lower number of patients in the usual care group (n= 49) compared to the Bischoff’s paper (n= 55). This raised some concerns about the randomisation process. Clarification could not be given by the authors.

High risk of bias

Effect of assignment: "Our primary analysis was based on the intention to treat principle and included all available data for all participants. We compared self management with usual care and routine monitoring with usual care. We did not impute any missing data." p. 3. Comment: It is not possible to blind patients / carers /people delivering the intervention due to the nature of this intervention. Effect of adhering: "“We prevented potential treatment contamination caused by provision of self management, routine monitoring, and usual care within the same." p. 2. "Seven (13%) patients did not receive any session or telephone contact, and 48 (87%) received two or more sessions." p. 4. "Six (11%) patients did not receive any consultation, and 42 (76%) patients received two or more consultations." p. 4. Comment: There was some non‐adherence in intervention group (at least 10%), this may all have led to some effect in the usual care group and a smaller difference between the intervention and usual care group.

Some concerns

Comment: Almost 16% of the participants dropped out during follow‐up (intervention 11%; usual care 20%). However, baseline characteristics did not differ between dropouts and participants who finished follow‐up. Exclusion is well described in flow chart. Intention‐to‐treat analyses were used.

Low risk of bias

Comment: The outcome assessor was blinded (p. 2). Outcome assessment was performed with standardised questionnaires and a telephonic exacerbation assessment system (TEXAS).

Some concerns

Comments: The trial register includes outcomes that haven’t been reported in this paper, but the outcomes that are reported in this paper have been mentioned in the trial register.

High risk of bias

Unpredictable direction of the outcome.

Mitchell 2014

Low risk of bias

“Patients were assigned to either usual care or SPACE FOR COPD via a web‐based, concealed allocation programme, using simple randomisation codes prepared by the trial statistician (J. Bankart).” p. 1539. "Randomisation was conducted by the trial investigator responsible for administering the intervention (K.E. Mitchell).” p. 1539. Comment: Random sequence allocation was adequately performed. The method of allocation concealment was not reported, but they say it is concealed. The only significant baseline difference between treatment groups was the CRQ‐SR dyspnoea score; therefore, this was corrected for in subsequent analyses.

High risk of bias

Effect of assignment: “Lack of participant blinding may have increased motivation when receiving the treatment and attempts to satisfy the researchers might have increased the observed treatment effects in the intervention arm. We cannot, therefore, rule out the possible impact of attention.” p. 1546. Comment: No blinding of participants. Analysis was carried out on an intention‐to‐treat basis. Effect of adhering: Comment: Withdrawal around intervention group: 27% and UC: 17%. The adherence of using the manual in by the intervention patients was not reported.

Low risk of bias

Comment: Data were not available for all randomised patients (withdrawal intervention: 27%, usual care: 17%). However, they found no significant differences in demographics or baseline variables between those who completed and those who did not complete the study.

Low risk of bias

“The assessments at week 6 and 6 months were conducted by a member of the research team who was blind to randomisation allocation (V. Johnson‐Warrington)." p. 1540 Comment: The outcome assessment was blinded.

Low risk of bias

Comment: No signs for selective outcome reporting. The primary outcome measure and most of the secondary outcome measures were reported.

High risk of bias

Unpredictable direction of the outcome.

Benzo 2016

Low risk of bias

Comment: Patients were randomly assigned, using an online, computer‐generated, simple binomial randomization program to one of the two groups, stratified by center.

High risk of bias

Effect of assignment: "85% of the individuals in the intervention group received the intervention as intended, defined as 15 (.70%) of 21 calls completed. Reasons for not completing the intervention were death during the study period (n = 3), unable to contact (n = 6), and refusal to complete the scheduled calls (n = 7)." p. 674. Comment: It was possible to blind the treatment for patients and people delivering the treatment. Intention‐to‐treat analysis was used. Effect of adhering: "Attendance at pulmonary rehabilitation visits anytime in the first 12 months after discharge tended to be higher in the intervention group (53% vs. 43%; P = 0.056)." p. 675. "In the intervention group, 85% of the individuals received a complete intervention, defined as 15 (.70%) of 21 calls completed." p. 674.

Low risk of bias

‐

Low risk of bias

Comment: Outcome assessors were blinded.

High risk of bias

Comment: No analysis plan available in the trial register. Furthermore, the trial register mentions that the outcomes will be assessed in the 12 months time frame, but not mentioned how and when they were assessed. Although 12 months analyses show no differences, the additional time points that were reported (not prespecified) are in favour of experimental.

High risk of bias

Predicted direction: Favours experimental.

Johnson‐Warrington 2016

Low risk of bias

"Participants were randomized to receive usual care or SPACE for COPD via a web‐based, concealed allocation program (www.sealedenvelope.com) using simple random permuted block 1:1 randomization by VJ‐W. Randomization was performed after the participants completed the baseline assessment, with treatment allocation prior to hospital discharge." p. 1162.

Some concerns

Effect of assignment: "Analysis was conducted on an intention‐to‐treat basis." p. 1163. Effect of adhering: Patients and personnel were not blinded. No information regarding adherence reported.

High risk of bias

Some patients were missing: n= 78 were randomized (n= 39 to each arm). Included in the analysis: Intervention: n= 35 and usual care: n= 36.

Low risk of bias

"All outcomes were measured at baseline (during admission but as close to discharge as possible) and 3 months after randomization, by a clinician blinded to treatment allocation." p. 1163.

Some concerns

No protocol including a pre‐specified analysis plan was available.

High risk of bias

Unpredictable direction of the outcome.

Lenferink 2019

Low risk of bias

Comment: Allocation was stratified on potential confounders. Only a baseline difference in 6MWD was reported.

High risk of bias

Effect of assignment: Comment: It was not possible to blind the treatment for patients and people delivering the treatment. Intention‐to‐treat analyses were performed. Effect of adhering: "The 12‐month follow‐up was completed by 83.3% of the individuals in the self‐management group and 84.8% in the UC group (figure 1). Patients completed 81.3% (self‐management: 82.3%; UC: 80.2%) of the diary data." p. 6. Comment: The effect of adherence to the intervention was not analysed. However, it was observed that a high number of patients in completed follow‐up and their symptom diaries. Adherence to action plans was measured, but not yet reported in the current publication.

Low risk of bias

Comment: Relatively low number of drop‐outs. The imputated data analyses showed similar results as the observed data analyses.

Low risk of bias

"Wherever possible, though, assessors of outcomes were blinded to the treatment group." p. 3.

Low risk of bias

Comment: Some small changes in final analyses compared to analysis plan that was published with the study protocol.

High risk of bias

Additional bias: "Based on the exacerbation rates/patient/day from the COPE‐II study (intervention: 0.116; control: 0.176, both with an estimated standard deviation of 0.17) and allowing for overdispersion and an attrition rate of 10%, 105 patients per group provided 80% power to detect an effect of this size, with a two‐tailed p<0.05." p. 4. Comment: The study included less patients (intervention: n= 102; usual care: n= 99) than aimed by power calculation (both groups n= 105). Unpredicted direction of the outcome.

Risk of bias for analysis 1.5 Healthcare utilisation: respiratory‐related hospital admissions (number of participants with at least one admission) (primary analysis).

Study

Bias

Randomisation process

Deviations from intended interventions

Missing outcome data

Measurement of the outcome

Selection of the reported results

Overall

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Gallefoss 1999

Low risk of bias

Comment: At inclusion they signed a written consent and were then randomised to an intervention group or a control group.

High risk of bias

Effect of assignment: Comment: No information reported about blinding, but it is almost impossible to blind these patients because of the nature of the intervention. Furthermore, no information was reported whether deviations arose from the intended intervention because of the trial context. Effect of adhering: "Forty‐six percent of those educated (12 of 26) received a standard treatment plan. Non‐standard treatment plans, incorporating the use of oral steroids as the first line of action in the yellow zone, were followed in some cases, for example, if the patient already used high dosages of inhaled steroids as maintenance therapy, or could report that a double or triple increase in inhaled steroids previously had shown marginal effect on the course of attacks/exacerbations. Among those 14 patients receiving non‐standard treatment plans, eight patients did not want to or were not able to use peak flow monitoring as a basis for change in medication. For those patients, symptom‐only based treatment plans were issued." Gallefoss 2002, p. 425. "In the intervention group, four patients failed to complete the educational program (social problems (n = 1), unannounced emigration (n = 1), failure to meet at educational group sessions for unknown reasons (n = 1) and serious myocardial infarction (n = 1))." Gallefoss 2002, p. 427. Comment: No information reported about adherence to the programme of the other patients (who completed the programme).

High risk of bias

“In the intervention group, four patients failed to complete the educational programme (social problems (n = 1), unannounced emigration (n = 1), failure to meet at educational group sessions for unknown reasons (n = 1) and serious myocardial infarction (n = 1)). Another patient was withdrawn from the study during the follow‐up due to lymphoma (n = 1). This left us with 26 patients (81%) for a 1‐year follow‐up. The patients who were withdrawn from the intervention group did not, to our knowledge, have any serious deterioration in their obstructive lung disease, and none were hospitalised. In the control group four patients were withdrawn (lack of co‐operation (n = 2), diagnosis of rectal cancer (n = 1) and emigration (n = 1)). Two of the withdrawn control group patients were hospitalised for exacerbations of their COPD. This left us with 27 patients (84%) for the 1‐year follow‐up” Gallefoss 2002, p. 427. Comment: The number of dropouts was relatively low, and reasons for dropout were comparable over groups

High risk of bias

"Prebronchodilator spirometry was performed before randomization and at 12 mo follow‐up by standard methods using a Jaeger MasterLab Body Box (Würzburg, Germany). The technical staff did not know whether the patients belonged to the control or intervention groups." Gallefoss 1999, p. 2003. Comment: No information reported regarding the assessors of the primary outcome. Number of prednisolone courses was retrospectively self‐reported at 12 months of follow‐up.

Some concerns

"We defined a priori the patient as compliant when dispensed regular medication was greater than 75% of prescribed regular medication during the study period." Gallefoss 1999, p. 2002.

High risk of bias

Unpredictable direction of the outcome.

Bourbeau 2003

Low risk of bias

“… central computergenerated list of random numbers. Randomization was stratified per center and in blocks of 6, and patients were assigned to the self‐management program (intervention group) or to usual care.” p. 586. “The blocking factor was not known by the investigators or their staff in each participating center." p. 586. “The use of respiratory medications was similar between study groups, except that oral steroids were used less commonly in the intervention group (7%) than in the usual care group (13%). None of the disease severity characteristics were otherwise different between the study groups.” p. 586. Comment: Random sequence allocation was adequately performed.

High risk of bias

Effect of assignment: "Since a double‐blind design was impossible..." p. 586. Comment: Participants and personnel were not blinded. ITT analysis was performed. Effect of adhering: "Since a double‐blind design was impossible..." p. 586. Comment: Participants and personnel were not blinded. Comment: There might be some imbalance in non‐protocol interventions such as exercise evaluation (not mandatory). However, no information available.

Low risk of bias

‐

Low risk of bias

"...an independent evaluator unaware of the patient assignment was responsible for the evaluation process in each center. The evaluator was cautioned not to ask about the workbook modules and types of contact.” p. 586. Comment: Outcome assessment was blinded.

Some concerns

Comment: No protocol or prespecified analysis plan available.

High risk of bias

Unpredictable direction of the outcome.

Coultas 2005

Low risk of bias

“Patients were randomly assigned to one of three intervention groups (usual care (UC), nurse‐assisted medical management (MM), or nurse‐assisted collaborative management (CM)) using a computer‐generated random list.” p. 2018. “Although the distribution of most of the baseline characteristics of the three groups was similar (Tables 2 and 3), only the distribution of gender and the tangible domain of social support were significantly different among the intervention groups.” p. 2020. Comment: It is unclear how allocation concealment was guaranteed. Observed imbalances are probably compatible with chance. Random sequence generation was adequately performed.

High risk of bias

Effect of assignment: “Of the 217 patients enrolled in the study, 151 (69.6%) completed the 6‐month intervention and follow‐up data collection.” p. 2020. “The reasons for the failure to complete the study were patient was unavailable for follow‐up (26.7%) and death (3.7%) [Fig 1]. The frequency of patients being unavailable for follow‐up was equally distributed among the three intervention groups (Fig 1). Overall, the demographic characteristics of the patients who dropped out of the study were similar to those who completed the study (data not shown). However, patients who dropped out of the study had more severe airflow obstruction, higher levels of distress, and lower quality of life, as measured with the SGRQ, compared with the patients who had completed the study (data not shown).” p. 2020. Comment: Not reported whether deviations arose because of the trial context. Only patients who completed the 6‐month intervention and follow‐up data collection were analysed (per‐protocol analyses). Effect of adhering: Comment: Patients and personnnel were probably aware of assignment to the intervention during the trial.

High risk of bias

“The reasons for the failure to complete the study were patient was unavailable for follow‐up (26.7%) and death (3.7%).” p. 2020. “However, patients who dropped out of the study had more severe airflow obstruction, higher levels of distress, and lower quality of life, as measured with the SGRQ, compared with the patients who had completed the study (data not shown).” p. 2020. Comment: Only patients who completed the 6‐month intervention and follow‐up data collection were analysed.

Low risk of bias

“To limit interviewer bias, each interviewer who obtained the baseline and 6‐month outcome data was blinded to the patient’s treatment group.” p. 2023. "Although self‐reported health‐care utilization may be subject to recall bias, it has been shown to be an acceptable method of measurement and may actually result in an underestimation of utilization with more frequent use.” p. 2023.

Some concerns

Comment: No protocol available. Measures and analysis intentions were only reported in the method section of the paper.

High risk of bias

Unpredictable direction of the outcome.

Coultas 2005

Low risk of bias

“Patients were randomly assigned to one of three intervention groups (usual care (UC), nurse‐assisted medical management (MM), or nurse‐assisted collaborative management (CM)) using a computer‐generated random list.” p. 2018. “Although the distribution of most of the baseline characteristics of the three groups was similar (Tables 2 and 3), only the distribution of gender and the tangible domain of social support were significantly different among the intervention groups.” p. 2020. Comment: It is unclear how allocation concealment was guaranteed. Observed imbalances are probably compatible with chance. Random sequence generation was adequately performed.

High risk of bias

Effect of assignment: “Of the 217 patients enrolled in the study, 151 (69.6%) completed the 6‐month intervention and follow‐up data collection.” p. 2020. “The reasons for the failure to complete the study were patient was unavailable for follow‐up (26.7%) and death (3.7%) [Fig 1]. The frequency of patients being unavailable for follow‐up was equally distributed among the three intervention groups (Fig 1). Overall, the demographic characteristics of the patients who dropped out of the study were similar to those who completed the study (data not shown). However, patients who dropped out of the study had more severe airflow obstruction, higher levels of distress, and lower quality of life, as measured with the SGRQ, compared with the patients who had completed the study (data not shown).” p. 2020. Comment: Not reported whether deviations arose because of the trial context. Only patients who completed the 6‐month intervention and follow‐up data collection were analysed (per‐protocol analyses). Effect of adhering: Comment: Patients and personnnel were probably aware of assignment to the intervention during the trial.

High risk of bias

“The reasons for the failure to complete the study were patient was unavailable for follow‐up (26.7%) and death (3.7%).” p. 2020. “However, patients who dropped out of the study had more severe airflow obstruction, higher levels of distress, and lower quality of life, as measured with the SGRQ, compared with the patients who had completed the study (data not shown).” p. 2020. Comment: Only patients who completed the 6‐month intervention and follow‐up data collection were analysed.

Low risk of bias

“To limit interviewer bias, each interviewer who obtained the baseline and 6‐month outcome data was blinded to the patient’s treatment group.” p. 2023. "Although self‐reported health‐care utilization may be subject to recall bias, it has been shown to be an acceptable method of measurement and may actually result in an underestimation of utilization with more frequent use.” p. 2023.

Some concerns

Comment: No protocol available. Measures and analysis intentions were only reported in the method section of the paper.

High risk of bias

Unpredictable direction of the outcome.

Rice 2010

Some concerns

"We assigned subjects in equal proportions to each of the two treatment arms by permuted‐block randomisation." Appendix 1, p. 3. "The baseline characteristics of the two treatment groups were generally well matched, with differences in age, lung function, and ED visits (not significant)." p. 891. Comment: Information on the method of allocation concealment was not reported.

High risk of bias

Effect of assignment: "We performed a randomized, adjudicator‐blinded, controlled, 1‐year trial." p. 890. "The status of all 743 patients was determined after 1 year. Electronic medical records were available for all hospitalizations and ED visits that occurred within the VA medical system. The success rate for obtaining discharge summaries and ED records for self‐reported events that occurred outside the VA was 99% in the usual care group and 98% in the disease management group." p. 891. Effect of adhering: Comment: 336 out of 372 (90.3%) patients completed disease management 323 out of 371 (87%) completed usual care.

Low risk of bias

"55% of patients in the usual care group and 60% of patients in the disease management group returned a completed the Saint George’s Respiratory Questionnaire in response to a single mailing at the end of the study. low response rates on SGRQ leading to a high risk of bias. However, data on healthcare utilisation seem complete with no risk of bias." p. 892.

Low risk of bias

"Blinded pulmonologists independently reviewed all discharge summaries and ED reports and assigned a primary cause for each." p. 891.

High risk of bias

Comment: Primary outcome in the paper is reported as a combined outcome measure, whereas it was planned to analyse the two outcomes separately.

High risk of bias

Unpredictable direction of the outcome.

Bucknall 2012

Low risk of bias

“We used a minimisation technique to stratify randomisation of participants by demographic factors (deprivation category of area of residence,11 age and sex, FEV1 per cent predicted at the time of randomisation, smoking status, participation in pulmonary rehabilitation within two years, and number of previous admissions) to control for key aspects of disease severity and predictors of readmission. We constructed a computer generated sequence by using the method of randomised permuted blocks of length four, with two allocations being made at random and two by minimisation. Treatment group allocations were obtained by telephone, after baseline assessments had been made.” p 2. Comment: Random sequence allocation was adequately performed.

High risk of bias

Effect of assignment: Comment: No blinding of patients and personnel. Data for the primary outcome (time to first acute COPD hospital admission or death) were available for all patients and analysed with a Cox proportional hazards model adjusting for the stratification variables. Effect of adhering: Comment: No blinding of patients and personnel. Seventy‐five (42%) of 180 intervention group participants who completed 12 months of follow‐up were judged to have become successful self managers.

Low risk of bias

Comment: Only for secondary outcome measures (e.g. SGRQ, HADS, CSES) not all data were available for all patients randomised (Table 3).

Low risk of bias

‐

Some concerns

‐

High risk of bias

Unpredictable direction of the outcome.

Fan 2012

Low risk of bias

“Randomization lists were generated on the basis of random, permuted blocks of variable size to ensure approximate balance over time.” p. 674. “The CSP Coordinating Center in Boston, Massachusetts, randomly assigned eligible patients in equal numbers to 2 groups, stratifying patients per site to allow for possible regional differences in patient characteristics and clinical practice patterns.” p. 674. Comment: Random sequence generation was adequately performed. The allocation was adequately concealed. Authors didn’t report whether there were any significant differences. Most characteristics are equally balanced, but there is a bit larger difference in the following characteristics (>5%): Currently married Completed high school Long acting inhaled B‐agonist Inhaled corticosteroid Home oxygen use IHD CHF

High risk of bias

Effect of assignment: “The 2 groups differed on the basis of a complex behavioral intervention that made blinding impossible.” p. 674. Comment: No blinding of participants and personnel. "For the primary intention‐to‐treat analysis, the null hypothesis was that the COPD hospitalization hazard rate for intervention did not differ from that for standardized care."p. 675. "Also, with respect to time to first COPD hospitalization, we conducted a futility analysis to assess what might have resulted if the trial had continued." p. 676. "When the study was stopped, the 1‐year cumulative incidence of COPD‐related hospitalization was 27% in the intervention group and 24% in the usual care group (hazard ratio, 1.13 [95% CI, 0.70 to 1.80]; P 0.62)." p. 677. Effect of adhering: "During the entire follow‐up period extending to September 2009, a total of 8 patients in the intervention group and 10 in the usual care group either did not attend scheduled study visits or formally withdrew from the study. Among patients in the intervention group, 87% completed all 4 individual educational visits and 57% completed the group visit. A review of the recorded sessions found that case managers covered 77% to 89% of the educational items in each session. The ranges of average scores across sites that provided audio recordings for each session were 18.3 to 31 for session 1 (13 sites), 12 to 19 for session 2 (12 sites), 12.7 to 23 for session 3 (10 sites), and 6 to 18 for session 4 (10 sites)." p. 677.

High risk of bias

“This multisite, randomized, controlled trial of an educational and acute care management program was stopped early when a safety monitoring board noted more deaths in the intervention group.” p. 674 Comment: There is incomplete outcome data due to early termination of the study.

Low risk of bias

“Telephone‐based ascertainment of study outcomes (COPD hospitalizations and exacerbations) was performed by centralized research staff blinded to assignment. All outcomes were collected by centralized staff blinded to study group, and all hospitalizations were adjudicated by a committee that was also blinded to study group.” p. 674. Comment: Research staff blinded to study group contacted patients every 2 months to determine whether they developed symptoms of a COPD exacerbation, along with details of treatment and health care use. Deaths were ascertained by staff at the study sites, by review of VA electronic medical and administrative records and by a search of the Social Security Death Index. Extensive multivariate analyses failed to identify any factors alone or combined that could plausibly explain the difference in mortality. Two of the authors who were blinded to study group reviewed all available medical information and did not find anything in the deaths that would not be expected in frail patients with COPD.

Some concerns

Comment: The primary and secondary outcomes were reported, only healthcare costs as (secondary objective) were not reported.

High risk of bias

Predicted directon: Favours comparator, because more deaths in the intervention group were observed. There is still a chance that this was because of chance.

Walters 2013

High risk of bias

Randomisation process: "After recruitment, practices were randomised using a code generated by investigators from a random numbers table stratified in blocks of four by Rural, Remote and Metropolitan Areas (RRMA) classifications in Tasmania. Allocation occurred independently using sequentially numbered, opaque and sealed envelopes." p. 2. Timing of identification or recruitment of participants: GPs identified patients with COPD aged over 45 years seen within the previous 12 months through database searches, based on a diagnostic code for COPD or prescription of tiotropium. Comment: Most variables are equally distributed, but more smokers (not statistically significant, p=0.102) and more previous admissions in the intervention group (statistically significant, p=0.028), so potentially more room for improvement. Although, this small number of differences identified as ‘statistically significant’ at the conventional 0.05 threshold should usually be considered to be compatible with chance.

High risk of bias

Effect of assignment: ‐ Effect of adhering: "The limitations of the study were the high rate of misclassification of COPD in general practices, which impacted on enrolment, and higher than expected withdrawal rates, especially in the HM group, which reduced the intended statistical power." p. 6. Comment: Nine patients discontinued and 8 patients withdrew in the intervention group.

Low risk of bias

Comment: Figure 1: 9 intervention patients discontinued, but they attended follow‐up. In Table 3, no patient numbers are mentioned.

Low risk of bias

Comment: Blinding of participants or research officers was not possible given the nature of the study.

Low risk of bias

Comment: Data that produced this result analysed in accordance with a pre‐specified analysis plan.

High risk of bias

Unpredictable direction of the outcome.

Mitchell 2014

Low risk of bias

“Patients were assigned to either usual care or SPACE FOR COPD via a web‐based, concealed allocation programme, using simple randomisation codes prepared by the trial statistician (J. Bankart).” p. 1539. "Randomisation was conducted by the trial investigator responsible for administering the intervention (K.E. Mitchell).” p. 1539. Comment: Random sequence allocation was adequately performed. The method of allocation concealment was not reported, but they say it is concealed. The only significant baseline difference between treatment groups was the CRQ‐SR dyspnoea score; therefore, this was corrected for in subsequent analyses.

High risk of bias

Effect of assignment: “Lack of participant blinding may have increased motivation when receiving the treatment and attempts to satisfy the researchers might have increased the observed treatment effects in the intervention arm. We cannot, therefore, rule out the possible impact of attention.” p. 1546. Comment: No blinding of participants. Analysis was carried out on an intention‐to‐treat basis. Effect of adhering: Comment: Withdrawal around intervention group: 27% and UC: 17%. The adherence of using the manual in by the intervention patients was not reported.

Low risk of bias

Comment: Data were not available for all randomised patients (withdrawal intervention: 27%, usual care: 17%). However, they found no significant differences in demographics or baseline variables between those who completed and those who did not complete the study.

Low risk of bias

“The assessments at week 6 and 6 months were conducted by a member of the research team who was blind to randomisation allocation (V. Johnson‐Warrington)." p. 1540 Comment: The outcome assessment was blinded.

Low risk of bias

Comment: No signs for selective outcome reporting. The primary outcome measure and most of the secondary outcome measures were reported.

High risk of bias

Unpredictable direction of the outcome.

Tabak 2014

Low risk of bias

"Patients were randomized using a computer‐generated randomization list (Blocked Stratified Randomization version 5; Steven Piantadosi), where randomization was applied in blocks of two and four.” p. 936. “Participants were allocated by a data manager in order of inclusion following the randomization list, placed in a sealed envelope.” p. 936. Comment: There was a significant difference in dyspnea (MRC scale) but this seems by chance. All other parameters seem evenly distributed.

Some concerns

Effect of assignment: "To present the outcome measures over time in both groups, a mixed‐model analysis for repeated measures was performed (intention to treat)." p. 938. Comment: Patients and personnel were not blinded. No information was reported regarding deviations from the intervention that arose because of the trial context. Effect of adhering: "Participants in the control group received only usual care. Patients in the control group were allowed to attend regular physiotherapy sessions if this was prescribed as part of usual care." p. 937. “Although some patients in the intervention group quit the physiotherapy modules ( exercising and activity coach) due to weak (n=l, < T l ) or unstable condition (n=l, <T2) and personal circumstances (n=2, <T4), they persisted in using the web portal and triage diary til! T4.” p. 939. Comment: Most outcome measures are reported for 3 months follow‐up, whereas there was a total of 9 months follow‐up. There was a high number of withdrawals for the 9 months follow‐up. However, there were more drop‐outs in the usual care control group than in the intervention group.

High risk of bias

“A large number of patients were not able or willing to continue study participation: 33% in the intervention group and 86% in the control group.” p. 939. Comment: Most outcome measures are reported for 3 months follow‐up, whereas there was a total of 9 months follow‐up. There was a high number of withdrawals for the 9 months follow‐up (more dropouts in the control group). Reasons reported for patients from the control group: too much effort (n=2), exacerbation (n‐1), hospitalisation (n‐2), kidney problems (n‐1), exacerbation, and cancer (n‐1).

Low risk of bias

‐

Low risk of bias

Comment: Not all outcome measures were reported for the 9 months follow‐up. Exacerbations (duration) were not reported. Also, no information or results provided for the use of diaries in the usual care control group.

High risk of bias

Unpredictable direction of the outcome.

Hernández 2015

Low risk of bias

"A computer‐generated list of random numbers with no restrictions and administered by personnel who were not involved in the study ensured blinded randomisation." p. 2. "(…) and administered by personnel who were not involved in the study." p. 2. Comment: Random sequence generation was adequately performed. No blinding of participants or personnel. The patients of the two groups showed similar characteristics, except that influenza and pneumococcal vaccination were more prevalent in the IC group (P<0.01).

High risk of bias

Effect of assignment: Comment: It was not possible to blind the treatment for patients and people delivering the treatment. Intention‐to‐treat analysis performed. Effect of adhering: "As a sensitivity analysis, we repeated all analyses after excluding patients who had a previous contact with any IC programme (n = 15)." p. 2. "25 patients (22%) died and 16 (14%) were lost during the 1‐year follow‐up, and the remaining 114 patients (IC, n = 59 and UC, n = 55) completed the 12‐month assessment visit. Patients lost to follow‐up exhibited a higher affectation of activities of daily living and health‐related quality of life at baseline than those who were followed up, without differences in other sociodemographic, clinical, functional and medical care‐related variables." p. 3. Comment: It is not mentioned in which way these patients were distributed over the two groups, nor are the results of the sensitivity analysis reported. Although the approach was not adequately adopted at the community level and thus the intervention was not continued after the 1 year of follow‐up, this was not a failure of implementing the intervention, since the intervention stopped at 1 year according to the protocol.

Low risk of bias

‐

Low risk of bias

"A blind evaluation of the study group carried out before randomisation and after the 12‐month follow‐up consisted of a patient interview and analysis of medical records, self‐administered questionnaires and lung function testing." p. 2.

Some concerns

"Although the study was officially registered as part of the EU Grant (CIP‐ICT‐PSP‐2007.225025), the RCT was not included in the clinicaltrials.gov registry because at that time it was not compulsory." p. 5. Comment: No protocol available.

High risk of bias

Unpredicted direction of outcome.

Titova 2015

High risk of bias

Randomisation process: “They were randomly allocated to either integrated care (IC) or usual care (UC) based on their address of permanent residence. In order to create two pairs of districts with approximately equal numbers of citizens, a pair‐wise matching of districts was carried out. It was decided by lottery that participants from District Pair 1 were assigned to the UC group, and participants from District Pair 2 were assigned to the IC group.” Titova 2015, p. 2. “There were no significant differences in baseline characteristics between the IC and the UC groups, but when comparing those who died during follow‐up with those who were still alive after two years, those who died had a significantly higher PaCO2 and lower BMI at baseline (Table 1). Those who died during follow‐up also had significantly more HA and HD during the year prior to study start.” Titova 2015, p. 4. Comment: No information provided about the allocation concealment. Timing of identification or recruitment of participants: "They were randomly allocated to either IC or UC group based on their address of permanent residence. Patient data were registered at discharge and 6, 12, and 24 months after discharge from the TUH." Titova 2017, p 3. Comment: If allocation was known before study inclusion, suitable patients could be selected for the intervention.

High risk of bias

Effect of assignment: “The study was a prospective, open, single‐centre intervention study.” Titova 2015, p. 2. "The principal comparisons of changes were performed on an intention‐to treat basis." Titova 2017, p. 4. Comment: No blinding of participants and personnel. Effect of adhering: ‐

High risk of bias

“172 patients were randomly allocated to the IC group (n = 91) or the UC group (n = 81), respectively. After two years of follow‐up, a total of 100 patients (58% of the included patients), 51 patients in the IC group and 49 patients in the UC group, were available for evaluation (Figure 1).” Titova 2017, p. 5. "Those who died had a significantly higher PaCO2 and lower BMI at baseline (Table 2). Those who died during follow‐up also had significantly more HA and HD during the year prior to study start." Titova 2015, p. 4. Comment: No mortality reported; however, Figure 1 shows higher mortality for the IC group, n=35 (38.4%), compared to the UC group, n=21 (25.9%).

High risk of bias

“Information concerning the number and duration of the COPD exacerbations, as well as the time from onset of symptoms until the start of self‐initiated treatment is insufficient due to many incomplete registrations in “My COPD book.” Titova 2015, p. 9. "“Data concerning HA (hospital admissions) and HD (hospital days) were collected from the hospital registry database’s medical charts.” Titova 2015, p. 3. Comment: Unclear who was the outcome assessor.

Some concerns

Trial register reported number of hospitalisations as the primary outcome. Secondary outcomes were quality of life, lung function, use of medication, cost‐effectiveness. However, the article reported not all secondary outcomes (e.g. cost‐effectiveness).

High risk of bias

Unpredictable direction of the outcome.

Sanchez‐Nieto 2016

Some concerns

"Controlled, randomized, parallel‐group, single‐blind study with follow‐up of 1 year. After consenting to take part in the study, simple randomization was carried out separately at each site by means of a list of computer‐generated random num­bers, assigning the patients to two groups." p. 1940. Comment: No information reported regarding concealed allocation sequence.

High risk of bias

Effect of assignment: Comment: Patients and personnel were not blinded. No proper intention‐to‐treat analysis used, as the baseline characteristics of the people that have dropped out haven’t been included in the analyses. Patients lost to follow‐up were evaluated until the time of the last contact with them by the research staff. Effect of adhering: Comment: No information reported regarding adherence.

High risk of bias

"In the end, 85 patients (88.5%) completed the study: 38 patients in the CG (18 (47.4%) in area VI and 20 (52.6%) in area I); and 47 patients in the IG (25 (53.2%) in area VI and 22 (46.8%) in area I; P=0.593). In the CG, seven patients were lost to follow‐up (one protocol violation and six dropouts), and in the IG, four were lost (one protocol violation and three dropouts)." p. 1942. Comment: Data were not available for all randomised patients.

Low risk of bias

"Because double‐blinding was not possible, an independent evalua­tor, who did not know the patients’ group assignments, was responsible for evaluating the outcome variables." p. 1940.

High risk of bias

Comment: No trial register number and study protocol were available.

High risk of bias

Unpredictable direction of the outcome.

Benzo 2016

Low risk of bias

Comment: Patients were randomly assigned, using an online, computer‐generated, simple binomial randomization program to one of the two groups, stratified by center.

High risk of bias

Effect of assignment: "85% of the individuals in the intervention group received the intervention as intended, defined as 15 (.70%) of 21 calls completed. Reasons for not completing the intervention were death during the study period (n = 3), unable to contact (n = 6), and refusal to complete the scheduled calls (n = 7)." p. 674. Comment: It was possible to blind the treatment for patients and people delivering the treatment. Intention‐to‐treat analysis was used. Effect of adhering: "Attendance at pulmonary rehabilitation visits anytime in the first 12 months after discharge tended to be higher in the intervention group (53% vs. 43%; P = 0.056)." p. 675. "In the intervention group, 85% of the individuals received a complete intervention, defined as 15 (.70%) of 21 calls completed." p. 674.

Low risk of bias

‐

Low risk of bias

Comment: Outcome assessors were blinded.

High risk of bias

Comment: No analysis plan available in the trial register. Furthermore, the trial register mentions that the outcomes will be assessed in the 12 months time frame, but not mentioned how and when they were assessed. Although 12 months analyses show no differences, the additional time points that were reported (not prespecified) are in favour of experimental.

High risk of bias

Predicted direction: Favours experimental.

Johnson‐Warrington 2016

Low risk of bias

"Participants were randomized to receive usual care or SPACE for COPD via a web‐based, concealed allocation program (www.sealedenvelope.com) using simple random permuted block 1:1 randomization by VJ‐W. Randomization was performed after the participants completed the baseline assessment, with treatment allocation prior to hospital discharge." p. 1162.

Some concerns

Effect of assignment: "Analysis was conducted on an intention‐to‐treat basis." p. 1163. Effect of adhering: Patients and personnel were not blinded. No information regarding adherence reported.

High risk of bias

Some patients were missing: n= 78 were randomized (n= 39 to each arm). Included in the analysis: Intervention: n= 35 and usual care: n= 36.

Low risk of bias

"All outcomes were measured at baseline (during admission but as close to discharge as possible) and 3 months after randomization, by a clinician blinded to treatment allocation." p. 1163.

Some concerns

No protocol including a pre‐specified analysis plan was available.

High risk of bias

Unpredictable direction of the outcome.

Lenferink 2019

Low risk of bias

Comment: Allocation was stratified on potential confounders. Only a baseline difference in 6MWD was reported.

High risk of bias

Effect of assignment: Comment: It was not possible to blind the treatment for patients and people delivering the treatment. Intention‐to‐treat analyses were performed. Effect of adhering: "The 12‐month follow‐up was completed by 83.3% of the individuals in the self‐management group and 84.8% in the UC group (figure 1). Patients completed 81.3% (self‐management: 82.3%; UC: 80.2%) of the diary data." p. 6. Comment: The effect of adherence to the intervention was not analysed. However, it was observed that a high number of patients in completed follow‐up and their symptom diaries. Adherence to action plans was measured, but not yet reported in the current publication.

Low risk of bias

Comment: Relatively low number of drop‐outs. The imputated data analyses showed similar results as the observed data analyses.

Low risk of bias

"Wherever possible, though, assessors of outcomes were blinded to the treatment group." p. 3.

Low risk of bias

Comment: Some small changes in final analyses compared to analysis plan that was published with the study protocol.

High risk of bias

Additional bias: "Based on the exacerbation rates/patient/day from the COPE‐II study (intervention: 0.116; control: 0.176, both with an estimated standard deviation of 0.17) and allowing for overdispersion and an attrition rate of 10%, 105 patients per group provided 80% power to detect an effect of this size, with a two‐tailed p<0.05." p. 4. Comment: The study included less patients (intervention: n= 102; usual care: n= 99) than aimed by power calculation (both groups n= 105). Unpredicted direction of the outcome.

Risk of bias for analysis 1.7 Healthcare utilisation: respiratory‐related hospital admissions (mean number per participant) (primary analysis).

Study

Bias

Randomisation process

Deviations from intended interventions

Missing outcome data

Measurement of the outcome

Selection of the reported results

Overall

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Bösch 2007

Some concerns

Information from the author regarding allocation concealment: “Pick of envelope. Enrolment and selection were right before the start of the study‐ a selection bias cannot be fully excluded". Comment: The method used to generate the random sequence generation was not clearly reported. The information received from the author regarding allocation concealment is too concise to assess risk of bias.

High risk of bias

Effect of assignment: Comment: Blinding of participants and personnel was not reported. No information reported whether patients were analysed in the wrong intervention group or were excluded from the analysis. No intention‐to‐treat analyses were performed. Effect of adhering: Comment: Eight (21%) participants in the intervention group and one (8%) participant in the control group dropped out, without clearly reported reasons.

Some concerns

Comment: Only the patients who completed the study follow‐up were included in the baseline characteristics and analysis (per protocol analysis). Eight (21%) participants in the intervention group and one (8%) participant in the control group dropped out. Reasons for drop‐out were not clearly reported. Furthermore, the trial report provides no information about the extent of missing outcome data.

Some concerns

Comment: There was no protocol available. Some of the outcomes were self‐reported outcomes (e.g. frequency of avoidance of physical exertion, frequency of sport activities, frequency of impairment of lifestyle), which could have been influenced by knowledge of the received intervention. Blinding of outcome assessment was not reported.

Some concerns

Comment: No signs for selective outcome reporting, results were reported extensively; however, no protocol was available.

High risk of bias

Unpredictable direction of the outcome.

Bucknall 2012

Low risk of bias

“We used a minimisation technique to stratify randomisation of participants by demographic factors (deprivation category of area of residence,11 age and sex, FEV1 per cent predicted at the time of randomisation, smoking status, participation in pulmonary rehabilitation within two years, and number of previous admissions) to control for key aspects of disease severity and predictors of readmission. We constructed a computer generated sequence by using the method of randomised permuted blocks of length four, with two allocations being made at random and two by minimisation. Treatment group allocations were obtained by telephone, after baseline assessments had been made.” p 2. Comment: Random sequence allocation was adequately performed.

High risk of bias

Effect of assignment: Comment: No blinding of patients and personnel. Data for the primary outcome (time to first acute COPD hospital admission or death) were available for all patients and analysed with a Cox proportional hazards model adjusting for the stratification variables. Effect of adhering: Comment: No blinding of patients and personnel. Seventy‐five (42%) of 180 intervention group participants who completed 12 months of follow‐up were judged to have become successful self managers.

Low risk of bias

Comment: Only for secondary outcome measures (e.g. SGRQ, HADS, CSES) not all data were available for all patients randomised (Table 3).

Low risk of bias

‐

Some concerns

‐

High risk of bias

Unpredictable direction of the outcome.

Tabak 2014

Low risk of bias

"Patients were randomized using a computer‐generated randomization list (Blocked Stratified Randomization version 5; Steven Piantadosi), where randomization was applied in blocks of two and four.” p. 936. “Participants were allocated by a data manager in order of inclusion following the randomization list, placed in a sealed envelope.” p. 936. Comment: There was a significant difference in dyspnea (MRC scale) but this seems by chance. All other parameters seem evenly distributed.

Some concerns

Effect of assignment: "To present the outcome measures over time in both groups, a mixed‐model analysis for repeated measures was performed (intention to treat)." p. 938. Comment: Patients and personnel were not blinded. No information was reported regarding deviations from the intervention that arose because of the trial context. Effect of adhering: "Participants in the control group received only usual care. Patients in the control group were allowed to attend regular physiotherapy sessions if this was prescribed as part of usual care." p. 937. “Although some patients in the intervention group quit the physiotherapy modules ( exercising and activity coach) due to weak (n=l, < T l ) or unstable condition (n=l, <T2) and personal circumstances (n=2, <T4), they persisted in using the web portal and triage diary til! T4.” p. 939. Comment: Most outcome measures are reported for 3 months follow‐up, whereas there was a total of 9 months follow‐up. There was a high number of withdrawals for the 9 months follow‐up. However, there were more drop‐outs in the usual care control group than in the intervention group.

High risk of bias

“A large number of patients were not able or willing to continue study participation: 33% in the intervention group and 86% in the control group.” p. 939. Comment: Most outcome measures are reported for 3 months follow‐up, whereas there was a total of 9 months follow‐up. There was a high number of withdrawals for the 9 months follow‐up (more dropouts in the control group). Reasons reported for patients from the control group: too much effort (n=2), exacerbation (n‐1), hospitalisation (n‐2), kidney problems (n‐1), exacerbation, and cancer (n‐1).

Low risk of bias

‐

Low risk of bias

Comment: Not all outcome measures were reported for the 9 months follow‐up. Exacerbations (duration) was not reported. Also, no information or results provided for the use of diaries in the usual care control group.

High risk of bias

Unpredictable direction of the outcome.

Titova 2015

High risk of bias

Randomisation process: “They were randomly allocated to either integrated care (IC) or usual care (UC) based on their address of permanent residence. In order to create two pairs of districts with approximately equal numbers of citizens, a pair‐wise matching of districts was carried out. It was decided by lottery that participants from District Pair 1 were assigned to the UC group, and participants from District Pair 2 were assigned to the IC group.” Titova 2015, p. 2. “There were no significant differences in baseline characteristics between the IC and the UC groups, but when comparing those who died during follow‐up with those who were still alive after two years, those who died had a significantly higher PaCO2 and lower BMI at baseline (Table 1). Those who died during follow‐up also had significantly more HA and HD during the year prior to study start.” Titova 2015, p. 4. Comment: No information provided about the allocation concealment. Timing of identification or recruitment of participants: "They were randomly allocated to either IC or UC group based on their address of permanent residence. Patient data were registered at discharge and 6, 12, and 24 months after discharge from the TUH." Titova 2017, p 3. Comment: If allocation was known before study inclusion, suitable patients could be selected for the intervention.

High risk of bias

Effect of assignment: “The study was a prospective, open, single‐centre intervention study.” Titova 2015, p. 2. "The principal comparisons of changes were performed on an intention‐to treat basis." Titova 2017, p. 4. Comment: No blinding of participants and personnel. Effect of adhering: ‐

High risk of bias

“172 patients were randomly allocated to the IC group (n = 91) or the UC group (n = 81), respectively. After two years of follow‐up, a total of 100 patients (58% of the included patients), 51 patients in the IC group and 49 patients in the UC group, were available for evaluation (Figure 1).” Titova 2017, p. 5. "Those who died had a significantly higher PaCO2 and lower BMI at baseline (Table 2). Those who died during follow‐up also had significantly more HA and HD during the year prior to study start." Titova 2015, p. 4. Comment: No mortality reported; however, Figure 1 shows higher mortality for the IC group, n=35 (38.4%), compared to the UC group, n=21 (25.9%).

High risk of bias

“Information concerning the number and duration of the COPD exacerbations, as well as the time from onset of symptoms until the start of self‐initiated treatment is insufficient due to many incomplete registrations in “My COPD book.” Titova 2015, p. 9. "“Data concerning HA (hospital admissions) and HD (hospital days) were collected from the hospital registry database’s medical charts.” Titova 2015, p. 3. Comment: Unclear who was the outcome assessor.

Some concerns

Trial register reported number of hospitalisations as the primary outcome. Secondary outcomes were quality of life, lung function, use of medication, cost‐effectiveness. However, the article reported not all secondary outcomes (e.g. cost‐effectiveness).

High risk of bias

Unpredictable direction of the outcome.

Jolly 2018

High risk of bias

"When a patient is identified as eligible for the study, and has given written, informed consent to take part, the researcher randomises the patient using a web‐based programme hosted by the Primary Care Clinical Trials Unit (PC‐CTU), University of Birmingham. Centre specific randomisation lists were produced by a statistician at the trials unit. Participants are individually randomised, stratified by centre. The four recruitment centres are Birmingham and West‐Midlands South; Greater Manchester; North West Midlands and Oxfordshire/Gloucestershire. Only the PC‐CTU have access to the allocation sequence." Sidhu 2015 (protocol paper), p. 4. "The study groups were generally well balanced in terms of patient characteristics, although there was a higher proportion of current smokers in the telephone health coaching group. The usual care group reported a higher level of self reported moderate and vigorous physical activity, but this was not reflected in the accelerometry data at baseline." Jolly 2018 (results paper), p. 5.

High risk of bias

Effect of assignment: Comment: Patients were informed of their allocation at the end of the recruitment appointment; they were not masked to treatment allocation. All data were analysed by intention to treat. Effect of adhering: "We achieved good fidelity of delivery of the intervention, with 75% of intervention participants receiving all four calls and only four patients receiving none." p. 11. "The dose and coverage of intervention delivery were high: 86.4% (999/1156) of the scheduled calls were delivered and 75.4% (218/289) of participants received all four calls." No information regarding pulmonary rehabilitation, oxygen use.

High risk of bias

"There was imbalance in the follow‐up rates between telephone health coaching (82.7%; 37 withdrawals) and usual care (96.2%; 7 withdrawals)." p. 4. "Participants who did not provide data at 12 months were more likely to be in GOLD stage 3, to be smokers, had lower levels of self‐reported physical activity, and to live alone than responders. " p. 5.

Some concerns

"Patients were informed of their allocation at the end of the recruitment appointment; they were not masked to treatment allocation. Data were entered into the study database by researchers at the University of Birmingham who were masked to the treatment allocation." p. 3. Comment: No information about outcome assessors.

Low risk of bias

‐

High risk of bias

Unpredictable direction of the outcome.

Lenferink 2019

Low risk of bias

Comment: Allocation was stratified on potential confounders. Only a baseline difference in 6MWD was reported.

High risk of bias

Effect of assignment: Comment: It was not possible to blind the treatment for patients and people delivering the treatment. Intention‐to‐treat analyses were performed. Effect of adhering: "The 12‐month follow‐up was completed by 83.3% of the individuals in the self‐management group and 84.8% in the UC group (figure 1). Patients completed 81.3% (self‐management: 82.3%; UC: 80.2%) of the diary data." p. 6. Comment: The effect of adherence to the intervention was not analysed. However, it was observed that a high number of patients in completed follow‐up and their symptom diaries. Adherence to action plans was measured, but not yet reported in the current publication.

Low risk of bias

Comment: Relatively low number of drop‐outs. The imputated data analyses showed similar results as the observed data analyses.

Low risk of bias

"Wherever possible, though, assessors of outcomes were blinded to the treatment group." p. 3.

Low risk of bias

Comment: Some small changes in final analyses compared to analysis plan that was published with the study protocol.

High risk of bias

Additional bias: "Based on the exacerbation rates/patient/day from the COPE‐II study (intervention: 0.116; control: 0.176, both with an estimated standard deviation of 0.17) and allowing for overdispersion and an attrition rate of 10%, 105 patients per group provided 80% power to detect an effect of this size, with a two‐tailed p<0.05." p. 4. Comment: The study included less patients (intervention: n= 102; usual care: n= 99) than aimed by power calculation (both groups n= 105). Unpredicted direction of the outcome.

Wang 2019

Low risk of bias

"The allocation sequence was generated and released to the interventionist on a case‐by‐case basis by an independent department that specialised in supplying randomly generated sequences for research. The randomisation sequence was recorded." p. 151. "Participants were randomly assigned to groups after consent was obtained and baseline data were collected." p. 151. "There were no significant differences in demographic and clinical data between the two groups at baseline (Table 1)." p. 151.

High risk of bias

Effect of assignment: Comment: It is not possible to blind the treatment for patients and people delivering the treatment. Intention‐to‐treat analysis was used. Effect of adhering: "154 participants were enrolled, among whom, 143 completed the 12‐month study program. The attrition rate was 7.14%." p. 151. "...adherence to the self‐management intervention was reported by the participants but not monitored by the nurses." p. 155. Comment: No info on reasons of adherence by patients reported in the current publication.

Low risk of bias

Comment: Relatively low attrition rate.

Low risk of bias

"To minimize researcher bias, the interventionists who provided care were blinded to the participants’ baseline and allocation sequence. The statistician was blinded to the participants’ results during the trial." p. 151. Comment: Not possible to blind the treatment for patients and people delivering the treatment.

Some concerns

Comment: No pre‐specified analysis plan published which can be checked. However they mention: The statistician was blinded to the participants’ results during the trial.

High risk of bias

Unpredictable direction of the outcome.

Risk of bias for analysis 1.9 Mortality: respiratory‐related mortality (primary analysis).

Study

Bias

Randomisation process

Deviations from intended interventions

Missing outcome data

Measurement of the outcome

Selection of the reported results

Overall

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Gallefoss 1999

Low risk of bias

Comment: At inclusion they signed a written consent and were then randomised to an intervention group or a control group.

High risk of bias

Effect of assignment: Comment: No information reported about blinding, but it is almost impossible to blind these patients because of the nature of the intervention. Furthermore, no information was reported whether deviations arose from the intended intervention because of the trial context. Effect of adhering: "Forty‐six percent of those educated (12 of 26) received a standard treatment plan. Non‐standard treatment plans, incorporating the use of oral steroids as the first line of action in the yellow zone, were followed in some cases, for example, if the patient already used high dosages of inhaled steroids as maintenance therapy, or could report that a double or triple increase in inhaled steroids previously had shown marginal effect on the course of attacks/exacerbations. Among those 14 patients receiving non‐standard treatment plans, eight patients did not want to or were not able to use peak flow monitoring as a basis for change in medication. For those patients, symptom‐only based treatment plans were issued." Gallefoss 2002, p. 425. "In the intervention group, four patients failed to complete the educational program (social problems (n = 1), unannounced emigration (n = 1), failure to meet at educational group sessions for unknown reasons (n = 1) and serious myocardial infarction (n = 1))." Gallefoss 2002, p. 427. Comment: No information reported about adherence to the programme of the other patients (who completed the programme).

High risk of bias

“In the intervention group, four patients failed to complete the educational programme (social problems (n = 1), unannounced emigration (n = 1), failure to meet at educational group sessions for unknown reasons (n = 1) and serious myocardial infarction (n = 1)). Another patient was withdrawn from the study during the follow‐up due to lymphoma (n = 1). This left us with 26 patients (81%) for a 1‐year follow‐up. The patients who were withdrawn from the intervention group did not, to our knowledge, have any serious deterioration in their obstructive lung disease, and none were hospitalised. In the control group four patients were withdrawn (lack of co‐operation (n = 2), diagnosis of rectal cancer (n = 1) and emigration (n = 1)). Two of the withdrawn control group patients were hospitalised for exacerbations of their COPD. This left us with 27 patients (84%) for the 1‐year follow‐up” Gallefoss 2002, p. 427. Comment: The number of dropouts was relatively low, and reasons for dropout were comparable over groups

High risk of bias

"Prebronchodilator spirometry was performed before randomization and at 12 mo follow‐up by standard methods using a Jaeger MasterLab Body Box (Würzburg, Germany). The technical staff did not know whether the patients belonged to the control or intervention groups." Gallefoss 1999, p. 2003. Comment: No information reported regarding the assessors of the primary outcome. Number of prednisolone courses was retrospectively self‐reported at 12 months of follow‐up.

Some concerns

"We defined a priori the patient as compliant when dispensed regular medication was greater than 75% of prescribed regular medication during the study period." Gallefoss 1999, p. 2002.

High risk of bias

Unpredictable direction of the outcome.

Fan 2012

Low risk of bias

“Randomization lists were generated on the basis of random, permuted blocks of variable size to ensure approximate balance over time.” p. 674. “The CSP Coordinating Center in Boston, Massachusetts, randomly assigned eligible patients in equal numbers to 2 groups, stratifying patients per site to allow for possible regional differences in patient characteristics and clinical practice patterns.” p. 674. Comment: Random sequence generation was adequately performed. The allocation was adequately concealed. Authors didn’t report whether there were any significant differences. Most characteristics are equally balanced, but there is a bit larger difference in the following characteristics (>5%): Currently married Completed high school Long acting inhaled B‐agonist Inhaled corticosteroid Home oxygen use IHD CHF

High risk of bias

Effect of assignment: “The 2 groups differed on the basis of a complex behavioral intervention that made blinding impossible.” p. 674. Comment: No blinding of participants and personnel. "For the primary intention‐to‐treat analysis, the null hypothesis was that the COPD hospitalization hazard rate for intervention did not differ from that for standardized care."p. 675. "Also, with respect to time to first COPD hospitalization, we conducted a futility analysis to assess what might have resulted if the trial had continued." p. 676. "When the study was stopped, the 1‐year cumulative incidence of COPD‐related hospitalization was 27% in the intervention group and 24% in the usual care group (hazard ratio, 1.13 [95% CI, 0.70 to 1.80]; P 0.62)." p. 677. Effect of adhering: "During the entire follow‐up period extending to September 2009, a total of 8 patients in the intervention group and 10 in the usual care group either did not attend scheduled study visits or formally withdrew from the study. Among patients in the intervention group, 87% completed all 4 individual educational visits and 57% completed the group visit. A review of the recorded sessions found that case managers covered 77% to 89% of the educational items in each session. The ranges of average scores across sites that provided audio recordings for each session were 18.3 to 31 for session 1 (13 sites), 12 to 19 for session 2 (12 sites), 12.7 to 23 for session 3 (10 sites), and 6 to 18 for session 4 (10 sites)." p. 677.

High risk of bias

“This multisite, randomized, controlled trial of an educational and acute care management program was stopped early when a safety monitoring board noted more deaths in the intervention group.” p. 674 Comment: There is incomplete outcome data due to early termination of the study.

Low risk of bias

“Telephone‐based ascertainment of study outcomes (COPD hospitalizations and exacerbations) was performed by centralized research staff blinded to assignment. All outcomes were collected by centralized staff blinded to study group, and all hospitalizations were adjudicated by a committee that was also blinded to study group.” p. 674. Comment: Research staff blinded to study group contacted patients every 2 months to determine whether they developed symptoms of a COPD exacerbation, along with details of treatment and health care use. Deaths were ascertained by staff at the study sites, by review of VA electronic medical and administrative records and by a search of the Social Security Death Index. Extensive multivariate analyses failed to identify any factors alone or combined that could plausibly explain the difference in mortality. Two of the authors who were blinded to study group reviewed all available medical information and did not find anything in the deaths that would not be expected in frail patients with COPD.

Some concerns

Comment: The primary and secondary outcomes were reported, only healthcare costs as (secondary objective) were not reported.

High risk of bias

Predicted directon: Favours comparator, because more deaths in the intervention group were observed. There is still a chance that this was because of chance.

Bucknall 2012

Low risk of bias

“We used a minimisation technique to stratify randomisation of participants by demographic factors (deprivation category of area of residence,11 age and sex, FEV1 per cent predicted at the time of randomisation, smoking status, participation in pulmonary rehabilitation within two years, and number of previous admissions) to control for key aspects of disease severity and predictors of readmission. We constructed a computer generated sequence by using the method of randomised permuted blocks of length four, with two allocations being made at random and two by minimisation. Treatment group allocations were obtained by telephone, after baseline assessments had been made.” p 2. Comment: Random sequence allocation was adequately performed.

High risk of bias

Effect of assignment: Comment: No blinding of patients and personnel. Data for the primary outcome (time to first acute COPD hospital admission or death) were available for all patients and analysed with a Cox proportional hazards model adjusting for the stratification variables. Effect of adhering: Comment: No blinding of patients and personnel. Seventy‐five (42%) of 180 intervention group participants who completed 12 months of follow‐up were judged to have become successful self managers.

Low risk of bias

Comment: Only for secondary outcome measures (e.g. SGRQ, HADS, CSES) not all data were available for all patients randomised (Table 3).

Low risk of bias

‐

Some concerns

‐

High risk of bias

Unpredictable direction of the outcome.

Tabak 2014

Low risk of bias

"Patients were randomized using a computer‐generated randomization list (Blocked Stratified Randomization version 5; Steven Piantadosi), where randomization was applied in blocks of two and four.” p. 936. “Participants were allocated by a data manager in order of inclusion following the randomization list, placed in a sealed envelope.” p. 936. Comment: There was a significant difference in dyspnea (MRC scale) but this seems by chance. All other parameters seem evenly distributed.

Some concerns

Effect of assignment: "To present the outcome measures over time in both groups, a mixed‐model analysis for repeated measures was performed (intention to treat)." p. 938. Comment: Patients and personnel were not blinded. No information was reported regarding deviations from the intervention that arose because of the trial context. Effect of adhering: "Participants in the control group received only usual care. Patients in the control group were allowed to attend regular physiotherapy sessions if this was prescribed as part of usual care." p. 937. “Although some patients in the intervention group quit the physiotherapy modules ( exercising and activity coach) due to weak (n=l, < T l ) or unstable condition (n=l, <T2) and personal circumstances (n=2, <T4), they persisted in using the web portal and triage diary til! T4.” p. 939. Comment: Most outcome measures are reported for 3 months follow‐up, whereas there was a total of 9 months follow‐up. There was a high number of withdrawals for the 9 months follow‐up. However, there were more drop‐outs in the usual care control group than in the intervention group.

High risk of bias

“A large number of patients were not able or willing to continue study participation: 33% in the intervention group and 86% in the control group.” p. 939. Comment: Most outcome measures are reported for 3 months follow‐up, whereas there was a total of 9 months follow‐up. There was a high number of withdrawals for the 9 months follow‐up (more dropouts in the control group). Reasons reported for patients from the control group: too much effort (n=2), exacerbation (n‐1), hospitalisation (n‐2), kidney problems (n‐1), exacerbation, and cancer (n‐1).

Low risk of bias

‐

Low risk of bias

Comment: Not all outcome measures were reported for the 9 months follow‐up. Exacerbations (duration) was not reported. Also, no information or results provided for the use of diaries in the usual care control group.

High risk of bias

Unpredictable direction of the outcome.

Titova 2015

High risk of bias

Randomisation process: “They were randomly allocated to either integrated care (IC) or usual care (UC) based on their address of permanent residence. In order to create two pairs of districts with approximately equal numbers of citizens, a pair‐wise matching of districts was carried out. It was decided by lottery that participants from District Pair 1 were assigned to the UC group, and participants from District Pair 2 were assigned to the IC group.” Titova 2015, p. 2. “There were no significant differences in baseline characteristics between the IC and the UC groups, but when comparing those who died during follow‐up with those who were still alive after two years, those who died had a significantly higher PaCO2 and lower BMI at baseline (Table 1). Those who died during follow‐up also had significantly more HA and HD during the year prior to study start.” Titova 2015, p. 4. Comment: No information provided about the allocation concealment. Timing of identification or recruitment of participants: "They were randomly allocated to either IC or UC group based on their address of permanent residence. Patient data were registered at discharge and 6, 12, and 24 months after discharge from the TUH." Titova 2017, p 3. Comment: If allocation was known before study inclusion, suitable patients could be selected for the intervention.

High risk of bias

Effect of assignment: “The study was a prospective, open, single‐centre intervention study.” Titova 2015, p. 2. "The principal comparisons of changes were performed on an intention‐to treat basis." Titova 2017, p. 4. Comment: No blinding of participants and personnel. Effect of adhering: ‐

High risk of bias

“172 patients were randomly allocated to the IC group (n = 91) or the UC group (n = 81), respectively. After two years of follow‐up, a total of 100 patients (58% of the included patients), 51 patients in the IC group and 49 patients in the UC group, were available for evaluation (Figure 1).” Titova 2017, p. 5. "Those who died had a significantly higher PaCO2 and lower BMI at baseline (Table 2). Those who died during follow‐up also had significantly more HA and HD during the year prior to study start." Titova 2015, p. 4. Comment: No mortality reported; however, Figure 1 shows higher mortality for the IC group, n=35 (38.4%), compared to the UC group, n=21 (25.9%).

High risk of bias

“Information concerning the number and duration of the COPD exacerbations, as well as the time from onset of symptoms until the start of self‐initiated treatment is insufficient due to many incomplete registrations in “My COPD book.” Titova 2015, p. 9. "“Data concerning HA (hospital admissions) and HD (hospital days) were collected from the hospital registry database’s medical charts.” Titova 2015, p. 3. Comment: Unclear who was the outcome assessor.

Some concerns

Trial register reported number of hospitalisations as the primary outcome. Secondary outcomes were quality of life, lung function, use of medication, cost‐effectiveness. However, the article reported not all secondary outcomes (e.g. cost‐effectiveness).

High risk of bias

Unpredictable direction of the outcome.

Johnson‐Warrington 2016

Low risk of bias

"Participants were randomized to receive usual care or SPACE for COPD via a web‐based, concealed allocation program (www.sealedenvelope.com) using simple random permuted block 1:1 randomization by VJ‐W. Randomization was performed after the participants completed the baseline assessment, with treatment allocation prior to hospital discharge." p. 1162.

Some concerns

Effect of assignment: "Analysis was conducted on an intention‐to‐treat basis." p. 1163. Effect of adhering: Patients and personnel were not blinded. No information regarding adherence reported.

High risk of bias

Some patients were missing: n= 78 were randomized (n= 39 to each arm). Included in the analysis: Intervention: n= 35 and usual care: n= 36.

Low risk of bias

"All outcomes were measured at baseline (during admission but as close to discharge as possible) and 3 months after randomization, by a clinician blinded to treatment allocation." p. 1163.

Some concerns

No protocol including a pre‐specified analysis plan was available.

High risk of bias

Unpredictable direction of the outcome.

Lenferink 2019

Low risk of bias

Comment: Allocation was stratified on potential confounders. Only a baseline difference in 6MWD was reported.

High risk of bias

Effect of assignment: Comment: It was not possible to blind the treatment for patients and people delivering the treatment. Intention‐to‐treat analyses were performed. Effect of adhering: "The 12‐month follow‐up was completed by 83.3% of the individuals in the self‐management group and 84.8% in the UC group (figure 1). Patients completed 81.3% (self‐management: 82.3%; UC: 80.2%) of the diary data." p. 6. Comment: The effect of adherence to the intervention was not analysed. However, it was observed that a high number of patients in completed follow‐up and their symptom diaries. Adherence to action plans was measured, but not yet reported in the current publication.

Low risk of bias

Comment: Relatively low number of drop‐outs. The imputated data analyses showed similar results as the observed data analyses.

Low risk of bias

"Wherever possible, though, assessors of outcomes were blinded to the treatment group." p. 3.

Low risk of bias

Comment: Some small changes in final analyses compared to analysis plan that was published with the study protocol.

High risk of bias

Additional bias: "Based on the exacerbation rates/patient/day from the COPE‐II study (intervention: 0.116; control: 0.176, both with an estimated standard deviation of 0.17) and allowing for overdispersion and an attrition rate of 10%, 105 patients per group provided 80% power to detect an effect of this size, with a two‐tailed p<0.05." p. 4. Comment: The study included less patients (intervention: n= 102; usual care: n= 99) than aimed by power calculation (both groups n= 105). Unpredicted direction of the outcome.

Wang 2019

Low risk of bias

"The allocation sequence was generated and released to the interventionist on a case‐by‐case basis by an independent department that specialised in supplying randomly generated sequences for research. The randomisation sequence was recorded." p. 151. "Participants were randomly assigned to groups after consent was obtained and baseline data were collected." p. 151. "There were no significant differences in demographic and clinical data between the two groups at baseline (Table 1)." p. 151.

Some concerns

Effect of assignment: Comment: It is not possible to blind the treatment for patients and people delivering the treatment. Intention‐to‐treat analysis was used. Effect of adhering: "154 participants were enrolled, among whom, 143 completed the 12‐month study program. The attrition rate was 7.14%." p. 151. "...adherence to the self‐management intervention was reported by the participants but not monitored by the nurses." p. 155. Comment: No info on reasons of adherence by patients reported in the current publication.

Low risk of bias

Comment: Relatively low attrition rate.

Low risk of bias

"To minimize researcher bias, the interventionists who provided care were blinded to the participants’ baseline and allocation sequence. The statistician was blinded to the participants’ results during the trial." p. 151. Comment: Not possible to blind the treatment for patients and people delivering the treatment.

Some concerns

Comment: No pre‐specified analysis plan published which can be checked. However they mention: The statistician was blinded to the participants’ results during the trial.

Some concerns

Unpredictable direction of the outcome.

Risk of bias for analysis 1.11 Mortality: all‐cause mortality (primary analysis).

Study

Bias

Randomisation process

Deviations from intended interventions

Missing outcome data

Measurement of the outcome

Selection of the reported results

Overall

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Gallefoss 1999

Low risk of bias

Comment: At inclusion they signed a written consent and were then randomised to an intervention group or a control group.

High risk of bias

Effect of assignment: Comment: No information reported about blinding, but it is almost impossible to blind these patients because of the nature of the intervention. Furthermore, no information was reported whether deviations arose from the intended intervention because of the trial context. Effect of adhering: "Forty‐six percent of those educated (12 of 26) received a standard treatment plan. Non‐standard treatment plans, incorporating the use of oral steroids as the first line of action in the yellow zone, were followed in some cases, for example, if the patient already used high dosages of inhaled steroids as maintenance therapy, or could report that a double or triple increase in inhaled steroids previously had shown marginal effect on the course of attacks/exacerbations. Among those 14 patients receiving non‐standard treatment plans, eight patients did not want to or were not able to use peak flow monitoring as a basis for change in medication. For those patients, symptom‐only based treatment plans were issued." Gallefoss 2002, p. 425. "In the intervention group, four patients failed to complete the educational program (social problems (n = 1), unannounced emigration (n = 1), failure to meet at educational group sessions for unknown reasons (n = 1) and serious myocardial infarction (n = 1))." Gallefoss 2002, p. 427. Comment: No information reported about adherence to the programme of the other patients (who completed the programme).

High risk of bias

“In the intervention group, four patients failed to complete the educational programme (social problems (n = 1), unannounced emigration (n = 1), failure to meet at educational group sessions for unknown reasons (n = 1) and serious myocardial infarction (n = 1)). Another patient was withdrawn from the study during the follow‐up due to lymphoma (n = 1). This left us with 26 patients (81%) for a 1‐year follow‐up. The patients who were withdrawn from the intervention group did not, to our knowledge, have any serious deterioration in their obstructive lung disease, and none were hospitalised. In the control group four patients were withdrawn (lack of co‐operation (n = 2), diagnosis of rectal cancer (n = 1) and emigration (n = 1)). Two of the withdrawn control group patients were hospitalised for exacerbations of their COPD. This left us with 27 patients (84%) for the 1‐year follow‐up” Gallefoss 2002, p. 427. Comment: The number of dropouts was relatively low, and reasons for dropout were comparable over groups

High risk of bias

"Prebronchodilator spirometry was performed before randomization and at 12 mo follow‐up by standard methods using a Jaeger MasterLab Body Box (Würzburg, Germany). The technical staff did not know whether the patients belonged to the control or intervention groups." Gallefoss 1999, p. 2003. Comment: No information reported regarding the assessors of the primary outcome. Number of prednisolone courses was retrospectively self‐reported at 12 months of follow‐up.

Some concerns

"We defined a priori the patient as compliant when dispensed regular medication was greater than 75% of prescribed regular medication during the study period." Gallefoss 1999, p. 2002.

High risk of bias

Unpredictable direction of the outcome.

Bourbeau 2003

Low risk of bias

“… central computergenerated list of random numbers. Randomization was stratified per center and in blocks of 6, and patients were assigned to the self‐management program (intervention group) or to usual care.” p. 586. “The blocking factor was not known by the investigators or their staff in each participating center." p. 586. “The use of respiratory medications was similar between study groups, except that oral steroids were used less commonly in the intervention group (7%) than in the usual care group (13%). None of the disease severity characteristics were otherwise different between the study groups.” p. 586. Comment: Random sequence allocation was adequately performed.

High risk of bias

Effect of assignment: "Since a double‐blind design was impossible..." p. 586. Comment: Participants and personnel were not blinded. ITT analysis was performed. Effect of adhering: "Since a double‐blind design was impossible..." p. 586. Comment: Participants and personnel were not blinded. Comment: There might be some imbalance in non‐protocol interventions such as exercise evaluation (not mandatory). However, no information available.

Low risk of bias

‐

Low risk of bias

"...an independent evaluator unaware of the patient assignment was responsible for the evaluation process in each center. The evaluator was cautioned not to ask about the workbook modules and types of contact.” p. 586. Comment: Outcome assessment was blinded.

Some concerns

Comment: No protocol or prespecified analysis plan available.

High risk of bias

Unpredictable direction of the outcome.

Martin 2004

High risk of bias

"Patients were randomly assigned to the intervention (care plan) or control (usual care) groups." p. 192. Comment: The method of random sequence allocation was not reported. The method of allocation concealment was not reported. Significant differences in gender and cigarette consumption (the latter can be a predictor of worse outcomes). No information reported about smokers – no‐smokers.

High risk of bias

Effect of assignment: Comment: Blinding of patients and personnel was not reported. The intervention group patients did however receive a care plan with specific instructions for the patient and/or carer. No intention‐to‐treat analysis was used. Effect of adhering: Comment: No information regarding adherence was reported.

High risk of bias

"Three subsequently withdrew for personal reasons. A further 13 died during the follow‐up period (...) (nine in the intervention group and four in the control group (NS)" p. 192. Comments: The number of withdrawals was higher in the intervention group compared to the usual care control group. However, no information was provided regarding the differences in drop‐out rates.

Some concerns

“Quality of life was measured by the St George’s Respiratory Questionnaire (SGRQ). The questionnaire was administered by the research nurse (DMcN) at each visit.” p. 192. Comment: The blinding of outcome assessment was not reported. However, it seems that the research nurse was aware of the intervention received.

Some concerns

Comment: No signs of selective reporting, however, no protocol available.

High risk of bias

Unpredictable direction of the outcome.

Coultas 2005

Low risk of bias

“Patients were randomly assigned to one of three intervention groups (usual care (UC), nurse‐assisted medical management (MM), or nurse‐assisted collaborative management (CM)) using a computer‐generated random list.” p. 2018. “Although the distribution of most of the baseline characteristics of the three groups was similar (Tables 2 and 3), only the distribution of gender and the tangible domain of social support were significantly different among the intervention groups.” p. 2020. Comment: It is unclear how allocation concealment was guaranteed. Observed imbalances are probably compatible with chance. Random sequence generation was adequately performed.

High risk of bias

Effect of assignment: “Of the 217 patients enrolled in the study, 151 (69.6%) completed the 6‐month intervention and follow‐up data collection.” p. 2020. “The reasons for the failure to complete the study were patient was unavailable for follow‐up (26.7%) and death (3.7%) [Fig 1]. The frequency of patients being unavailable for follow‐up was equally distributed among the three intervention groups (Fig 1). Overall, the demographic characteristics of the patients who dropped out of the study were similar to those who completed the study (data not shown). However, patients who dropped out of the study had more severe airflow obstruction, higher levels of distress, and lower quality of life, as measured with the SGRQ, compared with the patients who had completed the study (data not shown).” p. 2020. Comment: Not reported whether deviations arose because of the trial context. Only patients who completed the 6‐month intervention and follow‐up data collection were analysed (per‐protocol analyses). Effect of adhering: Comment: Patients and personnnel were probably aware of assignment to the intervention during the trial.

High risk of bias

“The reasons for the failure to complete the study were patient was unavailable for follow‐up (26.7%) and death (3.7%).” p. 2020. “However, patients who dropped out of the study had more severe airflow obstruction, higher levels of distress, and lower quality of life, as measured with the SGRQ, compared with the patients who had completed the study (data not shown).” p. 2020. Comment: Only patients who completed the 6‐month intervention and follow‐up data collection were analysed.

Low risk of bias

“To limit interviewer bias, each interviewer who obtained the baseline and 6‐month outcome data was blinded to the patient’s treatment group.” p. 2023. "Although self‐reported health‐care utilization may be subject to recall bias, it has been shown to be an acceptable method of measurement and may actually result in an underestimation of utilization with more frequent use.” p. 2023.

Some concerns

Comment: No protocol available. Measures and analysis intentions were only reported in the method section of the paper.

High risk of bias

Unpredictable direction of the outcome.

Coultas 2005

Low risk of bias

“Patients were randomly assigned to one of three intervention groups (usual care (UC), nurse‐assisted medical management (MM), or nurse‐assisted collaborative management (CM)) using a computer‐generated random list.” p. 2018. “Although the distribution of most of the baseline characteristics of the three groups was similar (Tables 2 and 3), only the distribution of gender and the tangible domain of social support were significantly different among the intervention groups.” p. 2020. Comment: It is unclear how allocation concealment was guaranteed. Observed imbalances are probably compatible with chance. Random sequence generation was adequately performed.

High risk of bias

Effect of assignment: “Of the 217 patients enrolled in the study, 151 (69.6%) completed the 6‐month intervention and follow‐up data collection.” p. 2020. “The reasons for the failure to complete the study were patient was unavailable for follow‐up (26.7%) and death (3.7%) [Fig 1]. The frequency of patients being unavailable for follow‐up was equally distributed among the three intervention groups (Fig 1). Overall, the demographic characteristics of the patients who dropped out of the study were similar to those who completed the study (data not shown). However, patients who dropped out of the study had more severe airflow obstruction, higher levels of distress, and lower quality of life, as measured with the SGRQ, compared with the patients who had completed the study (data not shown).” p. 2020. Comment: Not reported whether deviations arose because of the trial context. Only patients who completed the 6‐month intervention and follow‐up data collection were analysed (per‐protocol analyses). Effect of adhering: Comment: Patients and personnnel were probably aware of assignment to the intervention during the trial.

High risk of bias

“The reasons for the failure to complete the study were patient was unavailable for follow‐up (26.7%) and death (3.7%).” p. 2020. “However, patients who dropped out of the study had more severe airflow obstruction, higher levels of distress, and lower quality of life, as measured with the SGRQ, compared with the patients who had completed the study (data not shown).” p. 2020. Comment: Only patients who completed the 6‐month intervention and follow‐up data collection were analysed.

Low risk of bias

“To limit interviewer bias, each interviewer who obtained the baseline and 6‐month outcome data was blinded to the patient’s treatment group.” p. 2023. "Although self‐reported health‐care utilization may be subject to recall bias, it has been shown to be an acceptable method of measurement and may actually result in an underestimation of utilization with more frequent use.” p. 2023.

Some concerns

Comment: No protocol available. Measures and analysis intentions were only reported in the method section of the paper.

High risk of bias

Unpredictable direction of the outcome.

Rice 2010

Some concerns

"We assigned subjects in equal proportions to each of the two treatment arms by permuted‐block randomisation." Appendix 1, p. 3. "The baseline characteristics of the two treatment groups were generally well matched, with differences in age, lung function, and ED visits (not significant)." p. 891. Comment: Information on the method of allocation concealment was not reported.

High risk of bias

Effect of assignment: "We performed a randomized, adjudicator‐blinded, controlled, 1‐year trial." p. 890. "The status of all 743 patients was determined after 1 year. Electronic medical records were available for all hospitalizations and ED visits that occurred within the VA medical system. The success rate for obtaining discharge summaries and ED records for self‐reported events that occurred outside the VA was 99% in the usual care group and 98% in the disease management group." p. 891. Effect of adhering: Comment: 336 out of 372 (90.3%) patients completed disease management 323 out of 371 (87%) completed usual care.

Low risk of bias

"55% of patients in the usual care group and 60% of patients in the disease management group returned a completed the Saint George’s Respiratory Questionnaire in response to a single mailing at the end of the study. low response rates on SGRQ leading to a high risk of bias. However, data on healthcare utilisation seem complete with no risk of bias." p. 892.

Low risk of bias

"Blinded pulmonologists independently reviewed all discharge summaries and ED reports and assigned a primary cause for each." p. 891.

High risk of bias

Comment: Primary outcome in the paper is reported as a combined outcome measure, whereas it was planned to analyse the two outcomes separately.

High risk of bias

Unpredictable direction of the outcome.

Bucknall 2012

Low risk of bias

“We used a minimisation technique to stratify randomisation of participants by demographic factors (deprivation category of area of residence,11 age and sex, FEV1 per cent predicted at the time of randomisation, smoking status, participation in pulmonary rehabilitation within two years, and number of previous admissions) to control for key aspects of disease severity and predictors of readmission. We constructed a computer generated sequence by using the method of randomised permuted blocks of length four, with two allocations being made at random and two by minimisation. Treatment group allocations were obtained by telephone, after baseline assessments had been made.” p 2. Comment: Random sequence allocation was adequately performed.

High risk of bias

Effect of assignment: Comment: No blinding of patients and personnel. Data for the primary outcome (time to first acute COPD hospital admission or death) were available for all patients and analysed with a Cox proportional hazards model adjusting for the stratification variables. Effect of adhering: Comment: No blinding of patients and personnel. Seventy‐five (42%) of 180 intervention group participants who completed 12 months of follow‐up were judged to have become successful self managers.

Low risk of bias

Comment: Only for secondary outcome measures (e.g. SGRQ, HADS, CSES) not all data were available for all patients randomised (Table 2).

Low risk of bias

‐

Some concerns

‐

High risk of bias

Unpredictable direction of the outcome.

Fan 2012

Low risk of bias

“Randomization lists were generated on the basis of random, permuted blocks of variable size to ensure approximate balance over time.” p. 674. “The CSP Coordinating Center in Boston, Massachusetts, randomly assigned eligible patients in equal numbers to 2 groups, stratifying patients per site to allow for possible regional differences in patient characteristics and clinical practice patterns.” p. 674. Comment: Random sequence generation was adequately performed. The allocation was adequately concealed. Authors didn’t report whether there were any significant differences. Most characteristics are equally balanced, but there is a bit larger difference in the following characteristics (>5%): Currently married Completed high school Long acting inhaled B‐agonist Inhaled corticosteroid Home oxygen use IHD CHF

High risk of bias

Effect of assignment: “The 2 groups differed on the basis of a complex behavioral intervention that made blinding impossible.” p. 674. Comment: No blinding of participants and personnel. "For the primary intention‐to‐treat analysis, the null hypothesis was that the COPD hospitalization hazard rate for intervention did not differ from that for standardized care."p. 675. "Also, with respect to time to first COPD hospitalization, we conducted a futility analysis to assess what might have resulted if the trial had continued." p. 676. "When the study was stopped, the 1‐year cumulative incidence of COPD‐related hospitalization was 27% in the intervention group and 24% in the usual care group (hazard ratio, 1.13 [95% CI, 0.70 to 1.80]; P 0.62)." p. 677. Effect of adhering: "During the entire follow‐up period extending to September 2009, a total of 8 patients in the intervention group and 10 in the usual care group either did not attend scheduled study visits or formally withdrew from the study. Among patients in the intervention group, 87% completed all 4 individual educational visits and 57% completed the group visit. A review of the recorded sessions found that case managers covered 77% to 89% of the educational items in each session. The ranges of average scores across sites that provided audio recordings for each session were 18.3 to 31 for session 1 (13 sites), 12 to 19 for session 2 (12 sites), 12.7 to 23 for session 3 (10 sites), and 6 to 18 for session 4 (10 sites)." p. 677.

High risk of bias

“This multisite, randomized, controlled trial of an educational and acute care management program was stopped early when a safety monitoring board noted more deaths in the intervention group.” p. 674 Comment: There is incomplete outcome data due to early termination of the study.

Low risk of bias

“Telephone‐based ascertainment of study outcomes (COPD hospitalizations and exacerbations) was performed by centralized research staff blinded to assignment. All outcomes were collected by centralized staff blinded to study group, and all hospitalizations were adjudicated by a committee that was also blinded to study group.” p. 674. Comment: Research staff blinded to study group contacted patients every 2 months to determine whether they developed symptoms of a COPD exacerbation, along with details of treatment and health care use. Deaths were ascertained by staff at the study sites, by review of VA electronic medical and administrative records and by a search of the Social Security Death Index. Extensive multivariate analyses failed to identify any factors alone or combined that could plausibly explain the difference in mortality. Two of the authors who were blinded to study group reviewed all available medical information and did not find anything in the deaths that would not be expected in frail patients with COPD.

Some concerns

Comment: The primary and secondary outcomes were reported, only healthcare costs as (secondary objective) were not reported.

High risk of bias

Predicted directon: Favours comparator, because more deaths in the intervention group were observed. There is still a chance that this was because of chance.

Walters 2013

High risk of bias

Randomisation process: "After recruitment, practices were randomised using a code generated by investigators from a random numbers table stratified in blocks of four by Rural, Remote and Metropolitan Areas (RRMA) classifications in Tasmania. Allocation occurred independently using sequentially numbered, opaque and sealed envelopes." p. 2. Timing of identification or recruitment of participants: GPs identified patients with COPD aged over 45 years seen within the previous 12 months through database searches, based on a diagnostic code for COPD or prescription of tiotropium. Comment: Most variables are equally distributed, but more smokers (not statistically significant, p=0.102) and more previous admissions in the intervention group (statistically significant, p=0.028), so potentially more room for improvement. Although, this small number of differences identified as ‘statistically significant’ at the conventional 0.05 threshold should usually be considered to be compatible with chance.

High risk of bias

Effect of assignment: ‐ Effect of adhering: "The limitations of the study were the high rate of misclassification of COPD in general practices, which impacted on enrolment, and higher than expected withdrawal rates, especially in the HM group, which reduced the intended statistical power." p. 6. Comment: Nine patients discontinued and 8 patients withdrew in the intervention group.

Low risk of bias

Comment: Figure 1: 9 intervention patients discontinued, but they attended follow‐up. In Table 3, no patient numbers are mentioned.

Low risk of bias

Comment: Blinding of participants or research officers was not possible given the nature of the study.

Low risk of bias

Comment: Data that produced this result analysed in accordance with a pre‐specified analysis plan.

High risk of bias

Unpredictable direction of the outcome.

Mitchell 2014

Low risk of bias

“Patients were assigned to either usual care or SPACE FOR COPD via a web‐based, concealed allocation programme, using simple randomisation codes prepared by the trial statistician (J. Bankart).” p. 1539. "Randomisation was conducted by the trial investigator responsible for administering the intervention (K.E. Mitchell).” p. 1539. Comment: Random sequence allocation was adequately performed. The method of allocation concealment was not reported, but they say it is concealed. The only significant baseline difference between treatment groups was the CRQ‐SR dyspnoea score; therefore, this was corrected for in subsequent analyses.

High risk of bias

Effect of assignment: “Lack of participant blinding may have increased motivation when receiving the treatment and attempts to satisfy the researchers might have increased the observed treatment effects in the intervention arm. We cannot, therefore, rule out the possible impact of attention.” p. 1546. Comment: No blinding of participants. Analysis was carried out on an intention‐to‐treat basis. Effect of adhering: Comment: Withdrawal around intervention group: 27% and UC: 17%. The adherence of using the manual in by the intervention patients was not reported.

Low risk of bias

Comment: Data were not available for all randomised patients (withdrawal intervention: 27%, usual care: 17%). However, they found no significant differences in demographics or baseline variables between those who completed and those who did not complete the study.

Low risk of bias

“The assessments at week 6 and 6 months were conducted by a member of the research team who was blind to randomisation allocation (V. Johnson‐Warrington)." p. 1540 Comment: The outcome assessment was blinded.

Low risk of bias

Comment: No signs for selective outcome reporting. The primary outcome measure and most of the secondary outcome measures were reported.

High risk of bias

Unpredictable direction of the outcome.

Tabak 2014

Low risk of bias

"Patients were randomized using a computer‐generated randomization list (Blocked Stratified Randomization version 5; Steven Piantadosi), where randomization was applied in blocks of two and four.” p. 936. “Participants were allocated by a data manager in order of inclusion following the randomization list, placed in a sealed envelope.” p. 936. Comment: There was a significant difference in dyspnea (MRC scale) but this seems by chance. All other parameters seem evenly distributed.

Some concerns

Effect of assignment: "To present the outcome measures over time in both groups, a mixed‐model analysis for repeated measures was performed (intention to treat)." p. 938. Comment: Patients and personnel were not blinded. No information was reported regarding deviations from the intervention that arose because of the trial context. Effect of adhering: "Participants in the control group received only usual care. Patients in the control group were allowed to attend regular physiotherapy sessions if this was prescribed as part of usual care." p. 937. “Although some patients in the intervention group quit the physiotherapy modules ( exercising and activity coach) due to weak (n=l, < T l ) or unstable condition (n=l, <T2) and personal circumstances (n=2, <T4), they persisted in using the web portal and triage diary til! T4.” p. 939. Comment: Most outcome measures are reported for 3 months follow‐up, whereas there was a total of 9 months follow‐up. There was a high number of withdrawals for the 9 months follow‐up. However, there were more drop‐outs in the usual care control group than in the intervention group.

High risk of bias

“A large number of patients were not able or willing to continue study participation: 33% in the intervention group and 86% in the control group.” p. 939. Comment: Most outcome measures are reported for 3 months follow‐up, whereas there was a total of 9 months follow‐up. There was a high number of withdrawals for the 9 months follow‐up (more dropouts in the control group). Reasons reported for patients from the control group: too much effort (n=2), exacerbation (n‐1), hospitalisation (n‐2), kidney problems (n‐1), exacerbation, and cancer (n‐1).

Low risk of bias

‐

Low risk of bias

Comment: Not all outcome measures were reported for the 9 months follow‐up. Exacerbations (duration) was not reported. Also, no information or results provided for the use of diaries in the usual care control group.

High risk of bias

Unpredictable direction of the outcome.

Hernández 2015

Low risk of bias

"A computer‐generated list of random numbers with no restrictions and administered by personnel who were not involved in the study ensured blinded randomisation." p. 2. "(…) and administered by personnel who were not involved in the study." p. 2. Comment: Random sequence generation was adequately performed. No blinding of participants or personnel. The patients of the two groups showed similar characteristics, except that influenza and pneumococcal vaccination were more prevalent in the IC group (P<0.01).

High risk of bias

Effect of assignment: Comment: It was not possible to blind the treatment for patients and people delivering the treatment. Intention‐to‐treat analysis performed. Effect of adhering: "As a sensitivity analysis, we repeated all analyses after excluding patients who had a previous contact with any IC programme (n = 15)." p. 2. "25 patients (22%) died and 16 (14%) were lost during the 1‐year follow‐up, and the remaining 114 patients (IC, n = 59 and UC, n = 55) completed the 12‐month assessment visit. Patients lost to follow‐up exhibited a higher affectation of activities of daily living and health‐related quality of life at baseline than those who were followed up, without differences in other sociodemographic, clinical, functional and medical care‐related variables." p. 3. Comment: It is not mentioned in which way these patients were distributed over the two groups, nor are the results of the sensitivity analysis reported. Although the approach was not adequately adopted at the community level and thus the intervention was not continued after the 1 year of follow‐up, this was not a failure of implementing the intervention, since the intervention stopped at 1 year according to the protocol.

Low risk of bias

‐

Low risk of bias

"A blind evaluation of the study group carried out before randomisation and after the 12‐month follow‐up consisted of a patient interview and analysis of medical records, self‐administered questionnaires and lung function testing." p. 2.

Some concerns

"Although the study was officially registered as part of the EU Grant (CIP‐ICT‐PSP‐2007.225025), the RCT was not included in the clinicaltrials.gov registry because at that time it was not compulsory." p. 5. Comment: No protocol available.

High risk of bias

Unpredictable direction of outcome.

Jonsdottir 2015

Low risk of bias

"After accepting the invitation, prior to entering the research centre, the research nurses randomized the patients to experimental and control groups in accordance with CONSORT requirements using a computer‐generated list of random numbers." p. 2638.

High risk of bias

Effect of assignment: Comment: Not possible to blind the treatment for patients and people delivering the treatment. Intention‐to‐treat analysis was not carried out as stated in the article, but in table 3 all patients that completed the follow‐up seem to be included. Effect of adhering: "To minimize treatment bias the data were concealed from the nurses providing the treatment and the treatment notes were concealed from the nurses conducting data collection." p. 2640. "Both groups followed standardized protocols and had regular meetings to maintain consistency. Few family members participated. This actually changed the nature of the study from being a family‐centered intervention as planned, into becoming focused on patients, with family involvement when possible." p. 2640. "Attendance at treatment sessions in the experimental group was more than 80% in the patient/family conversations (43/48 attended three or four out of the four possible meetings) and the group meeting (40/48). Half of the patients in the experimental group smoked (24/48). Of those, 58% (14/24) participated in the smoking cessation treatment." p. 2641.

High risk of bias

"Three patients died in the experimental group... The dropout rate was 20% (n = 12) in the experimental group and 12% (n = 7) in the control group. Also additional missing data on outcome variables as displayed in table 3. The dropout rate for family members was 31% (n = 5) and 50% (n = 7) in the experimental and control groups respectively." p. 2640. Comment: No info on dropout reasons or characteristics of dropouts.

Low risk of bias

"To minimize treatment bias the data were concealed from the nurses providing the treatment and the treatment notes were concealed from the nurses conducting data collection." p. 2640.

High risk of bias

No information, not registered in a trial register.

High risk of bias

Unpredictable direction of the outcome.

Titova 2015

High risk of bias

Randomisation process: “They were randomly allocated to either integrated care (IC) or usual care (UC) based on their address of permanent residence. In order to create two pairs of districts with approximately equal numbers of citizens, a pair‐wise matching of districts was carried out. It was decided by lottery that participants from District Pair 1 were assigned to the UC group, and participants from District Pair 2 were assigned to the IC group.” Titova 2015, p. 2. “There were no significant differences in baseline characteristics between the IC and the UC groups, but when comparing those who died during follow‐up with those who were still alive after two years, those who died had a significantly higher PaCO2 and lower BMI at baseline (Table 1). Those who died during follow‐up also had significantly more HA and HD during the year prior to study start.” Titova 2015, p. 4. Comment: No information provided about the allocation concealment. Timing of identification or recruitment of participants: "They were randomly allocated to either IC or UC group based on their address of permanent residence. Patient data were registered at discharge and 6, 12, and 24 months after discharge from the TUH." Titova 2017, p 3. Comment: If allocation was known before study inclusion, suitable patients could be selected for the intervention.

High risk of bias

Effect of assignment: “The study was a prospective, open, single‐centre intervention study.” Titova 2015, p. 2. "The principal comparisons of changes were performed on an intention‐to treat basis." Titova 2017, p. 4. Comment: No blinding of participants and personnel. Effect of adhering: ‐

High risk of bias

“172 patients were randomly allocated to the IC group (n = 91) or the UC group (n = 81), respectively. After two years of follow‐up, a total of 100 patients (58% of the included patients), 51 patients in the IC group and 49 patients in the UC group, were available for evaluation (Figure 1).” Titova 2017, p. 5. "Those who died had a significantly higher PaCO2 and lower BMI at baseline (Table 1). Those who died during follow‐up also had significantly more HA and HD during the year prior to study start." Titova 2015, p. 4. Comment: No mortality reported; however, Figure 1 shows higher mortality for the IC group, n=35 (38.4%), compared to the UC group, n=21 (25.9%).

High risk of bias

“Information concerning the number and duration of the COPD exacerbations, as well as the time from onset of symptoms until the start of self‐initiated treatment is insufficient due to many incomplete registrations in “My COPD book.” Titova 2015, p. 9. "“Data concerning HA (hospital admissions) and HD (hospital days) were collected from the hospital registry database’s medical charts.” Titova 2015, p. 3. Comment: Unclear who was the outcome assessor.

Some concerns

Trial register reported number of hospitalisations as the primary outcome. Secondary outcomes were quality of life, lung function, use of medication, cost‐effectiveness. However, the article reported not all secondary outcomes (e.g. cost‐effectiveness).

High risk of bias

Unpredictable direction of the outcome.

Sanchez‐Nieto 2016

Some concerns

"Controlled, randomized, parallel‐group, single‐blind study with follow‐up of 1 year. After consenting to take part in the study, simple randomization was carried out separately at each site by means of a list of computer‐generated random num­bers, assigning the patients to two groups." p. 1940. Comment: No information reported regarding concealed allocation sequence.

High risk of bias

Effect of assignment: Comment: Patients and personnel were not blinded. No proper intention‐to‐treat analysis used, as the baseline characteristics of the people that have dropped out haven’t been included in the analyses. Patients lost to follow‐up were evaluated until the time of the last contact with them by the research staff. Effect of adhering: Comment: No information reported regarding adherence.

High risk of bias

"In the end, 85 patients (88.5%) completed the study: 38 patients in the CG (18 (47.4%) in area VI and 20 (52.6%) in area I); and 47 patients in the IG (25 (53.2%) in area VI and 22 (46.8%) in area I; P=0.593). In the CG, seven patients were lost to follow‐up (one protocol violation and six dropouts), and in the IG, four were lost (one protocol violation and three dropouts)." p. 1942. Comment: Data were not available for all randomised patients.

Low risk of bias

"Because double‐blinding was not possible, an independent evalua­tor, who did not know the patients’ group assignments, was responsible for evaluating the outcome variables." p. 1940.

High risk of bias

Comment: No trial register number and study protocol were available.

High risk of bias

Unpredictable direction of the outcome.

Benzo 2016

Low risk of bias

Comment: Patients were randomly assigned, using an online, computer‐generated, simple binomial randomization program to one of the two groups, stratified by center.

High risk of bias

Effect of assignment: "85% of the individuals in the intervention group received the intervention as intended, defined as 15 (.70%) of 21 calls completed. Reasons for not completing the intervention were death during the study period (n = 3), unable to contact (n = 6), and refusal to complete the scheduled calls (n = 7)." p. 674. Comment: It was possible to blind the treatment for patients and people delivering the treatment. Intention‐to‐treat analysis was used. Effect of adhering: "Attendance at pulmonary rehabilitation visits anytime in the first 12 months after discharge tended to be higher in the intervention group (53% vs. 43%; P = 0.056)." p. 675. "In the intervention group, 85% of the individuals received a complete intervention, defined as 15 (.70%) of 21 calls completed." p. 674.

Low risk of bias

‐

Low risk of bias

Comment: Outcome assessors were blinded.

High risk of bias

Comment: No analysis plan available in the trial register. Furthermore, the trial register mentions that the outcomes will be assessed in the 12 months time frame, but not mentioned how and when they were assessed. Although 12 months analyses show no differences, the additional time points that were reported (not prespecified) are in favour of experimental.

High risk of bias

Predicted direction: Favours experimental.

Johnson‐Warrington 2016

Low risk of bias

"Participants were randomized to receive usual care or SPACE for COPD via a web‐based, concealed allocation program (www.sealedenvelope.com) using simple random permuted block 1:1 randomization by VJ‐W. Randomization was performed after the participants completed the baseline assessment, with treatment allocation prior to hospital discharge." p. 1162.

Some concerns

Effect of assignment: "Analysis was conducted on an intention‐to‐treat basis." p. 1163. Effect of adhering: Patients and personnel were not blinded. No information regarding adherence reported.

High risk of bias

Some patients were missing: n= 78 were randomized (n= 39 to each arm). Included in the analysis: Intervention: n= 35 and usual care: n= 36.

Low risk of bias

"All outcomes were measured at baseline (during admission but as close to discharge as possible) and 3 months after randomization, by a clinician blinded to treatment allocation." p. 1163.

Some concerns

No protocol including a pre‐specified analysis plan was available.

High risk of bias

Unpredictable direction of the outcome.

Rose 2018

Low risk of bias

"Randomisation was performed according to a centralised, computer generated 1:1 randomisation schedule stratified by study site. Because of the nature of the intervention and co‐location of research staff within the respiratory clinics, healthcare providers, patients and outcome assessors were not blinded, though treating respirologists were not informed of study allocation." p. 3. Comment: Some differences between intervention and control group were reported in home‐oxygen therapy (around 6% between groups).

High risk of bias

Effect of assignment: Comment: Due to the nature of the intervention it was not possible to blind patients and personnel. Treating respirologists were not informed of study allocation. Effect of adhering: "Of the 221 participants with case manager contact for all 12 weekly calls, 29% were 100% compliant, 62% 50–99% compliant, and 22% <50% compliant i.e., completed call activities. Of the 203 participants with case manager contact for all nine monthly calls, 31% were 100% compliant, 47% 50–99% compliant, and 22% <50% compliant. Most (65%) participants did not refuse any calls; 20% refused two, 5% refused three, and 1% refused ⩾4 calls. Most common reason for refusal was being too busy (69% of refused calls), other reasons were too fatigued or ill (20%), or not interested in talking when called (6%)." p. 5.

Low risk of bias

Comment: Only a lot of missing data for questionnaire data. Even a lot of baseline data are missing.

Low risk of bias

Comment: Intention‐to‐treat analysis was used.

Low risk of bias

No signs for selective outcome reporting.

High risk of bias

Unpredictable direction of the outcome.

Bringsvor 2018

Low risk of bias

“The allocation ratio was 1:1. Included participants were given a running number (based on the alphabetical order of last names), and then the numbers were randomized by a statistician (using MATLAB) with no access to the participant list.” p. 3678. “(...) an attrition of 20% was expected, this indicates that at a minimum of six participants were to be allocated to each group. In municipalities with fewer than 12 registered participants, it was therefore not possible to establish two groups, and we decided to allocate one group to either intervention or control. (...) participants from municipalities with only one group were randomized to either intervention or control at the municipality level.” p. 3678‐3679.

High risk of bias

Effect of assignment: “All models were estimated using intention‐to‐treat (ITT) analysis and per‐protocol analysis (PPA). The ITT analysis was conducted for all available data, while inclusion in the PPA required participants to follow the protocol sufficiently; ie, we included in the PPA all participants who had participated in more than half of the group conversations.” p. 3680. Comment: Patients and personnel were not blinded. Effect of adhering: “The attrition rate was 22.2% in the control group and 31.5% in the intervention group (ITT).” p. 3681. "37 patients did not receive allocated intervention (participated <6 sessions).” p. 3681. Comment: Not reported whether there were used analyses with regard to the effect of adhering to intervention.

High risk of bias

“One study limitation is that the follow‐up measurements were limited because of a clerical error, where 22 intervention group participants did not receive follow‐up questionnaires (Figure 1) after leaving the groups. This introduced unnecessary missing data in the ITT analysis. Unfortunately, we lack sufficient information to determine whether this affected the results, although there were no significant differences between these 22 and other participants at baseline on any of the self‐management‐related domains.” p. 3685.

Low risk of bias

“Data were blinded during the analyses, ie, the groups were labeled with non‐identifying terms.”p. 3680. Comment: Not reported whether outcome assessors were aware of the intervention during the assessments.

Some concerns

Comment: In the Trial register the following pre‐defined primary outcomes were reported: Change in Health promoting competence, HeiQ (Time Frame: Before, after, and 3,6 and 12 months after comlepting the intervention (self‐management program)). However, measurements were only performed at baseline and after 12 months follow‐up. No detailed protocol was available.

High risk of bias

Unpredictable direction of the outcome.

Jolly 2018

High risk of bias

"When a patient is identified as eligible for the study, and has given written, informed consent to take part, the researcher randomises the patient using a web‐based programme hosted by the Primary Care Clinical Trials Unit (PC‐CTU), University of Birmingham. Centre specific randomisation lists were produced by a statistician at the trials unit. Participants are individually randomised, stratified by centre. The four recruitment centres are Birmingham and West‐Midlands South; Greater Manchester; North West Midlands and Oxfordshire/Gloucestershire. Only the PC‐CTU have access to the allocation sequence." Sidhu 2015 (protocol paper), p. 4. "The study groups were generally well balanced in terms of patient characteristics, although there was a higher proportion of current smokers in the telephone health coaching group. The usual care group reported a higher level of self reported moderate and vigorous physical activity, but this was not reflected in the accelerometry data at baseline." Jolly 2018 (results paper), p. 5.

High risk of bias

Effect of assignment: Comment: Patients were informed of their allocation at the end of the recruitment appointment; they were not masked to treatment allocation. All data were analysed by intention to treat. Effect of adhering: "We achieved good fidelity of delivery of the intervention, with 75% of intervention participants receiving all four calls and only four patients receiving none." p. 11. "The dose and coverage of intervention delivery were high: 86.4% (999/1156) of the scheduled calls were delivered and 75.4% (218/289) of participants received all four calls." No information regarding pulmonary rehabilitation, oxygen use.

High risk of bias

"There was imbalance in the follow‐up rates between telephone health coaching (82.7%; 37 withdrawals) and usual care (96.2%; 7 withdrawals)." p. 4. "Participants who did not provide data at 12 months were more likely to be in GOLD stage 3, to be smokers, had lower levels of self‐reported physical activity, and to live alone than responders. " p. 5.

Some concerns

"Patients were informed of their allocation at the end of the recruitment appointment; they were not masked to treatment allocation. Data were entered into the study database by researchers at the University of Birmingham who were masked to the treatment allocation." p. 3. Comment: No information about outcome assessors.

Low risk of bias

‐

High risk of bias

Unpredictable direction of the outcome.

Kessler 2018

Low risk of bias

"Patients were allocated to groups in a 1:1 fashion according to a pre‐specified randomisation list generated before the study by a partial‐minimisation computer algorithm under supervision of the study sponsor. Patients were assigned a randomisation number by study staff at each centre in sequential numerical order through a telephone‐based interactive voice response system. Randomisation was stratified by smoking status (current or former), need for respiratory assistance (none, or on LTOT and/or HMV), and centre." p. 3.

High risk of bias

Effect of assignment: "For practical reasons, the study was open; neither the patients nor the investigators were blinded to the COPD management strategy." p. 3. "The primary analysis of the primary outcome was performed in the intention‐to‐treat (ITT) population (all patients who entered the follow‐up period). The number of hospitalisation days fitting the established criteria was normalised on a yearly basis for each patient and compared between both groups using the non‐parametric Wilcoxon rank‐sum test." p. 4. "The open design of the study, though necessary owing to the extensive differences between the two disease management programmes, may have led to some bias because doctors could have had a tendency to discharge patients in the disease management arm earlier than those in the usual management arm." p. 12. Effect of adhering: Comment: 15 patients (<10%) received <25% self‐management training.

Low risk of bias

"…leaving 157 (91%) and 162 (94%) patients in the ITT population, 137 (80%) and 128 (74%) of whom completed at least 12 months of follow‐up." p. 5.

Low risk of bias

"Because hospitalisation decisions might have been subject to local practices, social considerations, bed availability, and so on, all hospitalisations were reviewed by an independent, three‐member end‐point validation committee (EVC) to assess the reason and appropriateness of each hospitalisation, determine whether the hospitalisation fulfilled the protocol definition, and determine the extent to which it was to be considered in the efficacy assessments (fully, partially, not at all)." p. 3.

Low risk of bias

"During the study, the protocol was amended to accelerate recruitment and allow study completion (supplementary material)." p. 3. Comment: Inclusion criteria were expanded from COPD patients receiving LTOT to patients with severe COPD, with or without LTOT; the minimum number of patients was reduced from 520 to 306 by retaining the initial assumptions for sample size and removing a multiple comparisons adjustment; the follow‐up period was reduced from 2 years to 1 year; and the contact frequency between DM patients and CMs was increased to improve patient training (either group or phone session, once per month). Prior to these changes, 72 patients on LTOT were enrolled in the study.

High risk of bias

Unpredictable direction of the outcome.

Wang 2019

Low risk of bias

"The allocation sequence was generated and released to the interventionist on a case‐by‐case basis by an independent department that specialised in supplying randomly generated sequences for research. The randomisation sequence was recorded." p. 151. "Participants were randomly assigned to groups after consent was obtained and baseline data were collected." p. 151. "There were no significant differences in demographic and clinical data between the two groups at baseline (Table 1)." p. 151.

High risk of bias

Effect of assignment: Comment: It is not possible to blind the treatment for patients and people delivering the treatment. Intention‐to‐treat analysis was used. Effect of adhering: "154 participants were enrolled, among whom, 143 completed the 12‐month study program. The attrition rate was 7.14%." p. 151. "...adherence to the self‐management intervention was reported by the participants but not monitored by the nurses." p. 155. Comment: No information on reasons of adherence by patients reported in the current publication.

Low risk of bias

Comment: Relatively low attrition rate.

Low risk of bias

"To minimize researcher bias, the interventionists who provided care were blinded to the participants’ baseline and allocation sequence. The statistician was blinded to the participants’ results during the trial." p. 151. Comment: Not possible to blind the treatment for patients and people delivering the treatment.

Some concerns

Comment: No pre‐specified analysis plan published which can be checked. However they mention: The statistician was blinded to the participants’ results during the trial.

High risk of bias

Unpredictable direction of the outcome.

Ferrone 2019

Some concerns

"After obtaining informed consent, subjects were randomly assigned by opaque sealed envelopes in blocks of four, stratified by site, in a ratio of 1:1 to IDM or usual care. The randomization sequence was generated by our biostatistician. Study personnel were blinded to the sequence and blocking factor." p. 8. "In the IDM group, there were numerically more women and more GOLD D patients. The usual care arm had a numerically lower baseline CAT score (better QoL) and more smokers." p. 2.

High risk of bias

Effect of assignment: "Data were analyzed using a complete case analysis approach." p. 8. Comment: Patients and personnel were not blinded. No intention‐to‐treat analysis was uses. 12 patients that withdrew directly after randomisation have not been included in the analyses. Effect of adhering: Comment: No information regarding adherence reported.

Low risk of bias

"A constrained longitudinal data analysis was conducted on the CAT score as a sensitivity analysis to evaluate the impact of missing data from patients lost to follow up." p. 7. "When, as a sensitivity analysis, a constrained longitudinal model was used to analyze CAT scores, the between‐group difference was found to be significant (p < 0.001)." p. 2. Comment: 87% of the randomised patients were analysed. However, 12 patients did not show up after randomisation.

Low risk of bias

‐

Low risk of bias

No signs for selective outcome reporting; however, no protocol available.

High risk of bias

Unpredictable direction of the outcome.

Lenferink 2019

Low risk of bias

Comment: Allocation was stratified on potential confounders. Only a baseline difference in 6MWD was reported.

High risk of bias

Effect of assignment: Comment: It was not possible to blind the treatment for patients and people delivering the treatment. Intention‐to‐treat analyses were performed. Effect of adhering: "The 12‐month follow‐up was completed by 83.3% of the individuals in the self‐management group and 84.8% in the UC group (figure 1). Patients completed 81.3% (self‐management: 82.3%; UC: 80.2%) of the diary data." p. 6. Comment: The effect of adherence to the intervention was not analysed. However, it was observed that a high number of patients in completed follow‐up and their symptom diaries. Adherence to action plans was measured, but not yet reported in the current publication.

Low risk of bias

Comment: Relatively low number of drop‐outs. The imputated data analyses showed similar results as the observed data analyses.

Low risk of bias

"Wherever possible, though, assessors of outcomes were blinded to the treatment group." p. 3.

Low risk of bias

Comment: Some small changes in final analyses compared to analysis plan that was published with the study protocol.

High risk of bias

Additional bias: "Based on the exacerbation rates/patient/day from the COPE‐II study (intervention: 0.116; control: 0.176, both with an estimated standard deviation of 0.17) and allowing for overdispersion and an attrition rate of 10%, 105 patients per group provided 80% power to detect an effect of this size, with a two‐tailed p<0.05." p. 4. Comment: The study included less patients (intervention: n= 102; usual care: n= 99) than aimed by power calculation (both groups n= 105). Unpredicted direction of the outcome.

Liang 2019

High risk of bias

"Group or solo GP clinics with ≥1000 patients on their databases were approached directly or with assistance from primary care organisations. After obtaining signed agreement, clinics were block randomised (block sizes of four and six) to usual care or intervention arms using an externally managed web‐based randomisation programme. Clinics were notified of their allocation." p. 3.

High risk of bias

Effect of assignment: ‐ Effect of adhering: "67 participants (43%) did not receive any part of the intervention." p. 7. "A key limitation is the low uptake of the intervention components." p. 10.

High risk of bias

Comment: Data were not available for all patients randomised.

Low risk of bias

"The effective analysis was according to an intention‐to‐treat principle." p. 4.

Low risk of bias

Results were analysed according to the published design paper.

High risk of bias

Predicted direction: Favours comparator, because general practices could exclude patients from selection to participate in the intervention due to the cluster randomisation approach. Adherence to the intervention was low.

Appendices

Appendix 1. Search strategies

Source	Search strategy
Airways Register (CRS) (Date of most recent search: 23 January 2020)	1 MeSH DESCRIPTOR Pulmonary Disease, Chronic Obstructive Explode All 2 MeSH DESCRIPTOR Bronchitis, Chronic 3 (obstruct*) near3 (pulmonary or lung* or airway* or airflow* or bronch* or respirat*) 4 COPD:MISC1 5 (COPD OR COAD OR COBD):TI,AB,KW 6 #1 OR #2 OR #3 OR #4 OR #5 7 MeSH DESCRIPTOR Self Care Explode All 8 MeSH DESCRIPTOR Education 9 MeSH DESCRIPTOR Patient Education as Topic 10 educat* 11 self‐manag* 12 self manag* 13 self‐car* or "self car*" 14 train* or instruct* 15 patient cent* or patient‐cent* 16 patient‐focus* or "patient focus*" 17 patient‐education or "patient education" 18 management plan or management‐plan 19 management* NEAR1 program* 20 behavior* or behaviour* 21 disease* NEAR2 management* 22 self‐efficac* 23 empower* 24 #7 or #8 or #9 or #10 or #11 or #12 or #13 or #14 or #15 or #16 or #17 or #18 or #19 or #20 or #21 or #22 or #23 25 #6 and #24
CENTRAL (CRS) (Date of most recent search: 23 January 2020)	1 MeSH DESCRIPTOR Pulmonary Disease, Chronic Obstructive Explode All 2 MeSH DESCRIPTOR Bronchitis, Chronic 3 (obstruct*) near3 (pulmonary or lung* or airway* or airflow* or bronch* or respirat*) 4 COPD:MISC1 5 (COPD OR COAD OR COBD):TI,AB,KW 6 #1 OR #2 OR #3 OR #4 OR #5 7 MeSH DESCRIPTOR Self Care Explode All 8 MeSH DESCRIPTOR Education 9 MeSH DESCRIPTOR Patient Education as Topic 10 educat* 11 self‐manag* 12 self manag* 13 self‐car* or "self car*" 14 train* or instruct* 15 patient cent* or patient‐cent* 16 patient‐focus* or "patient focus*" 17 patient‐education or "patient education" 18 management plan or management‐plan 19 management* NEAR1 program* 20 behavior* or behaviour* 21 disease* NEAR2 management* 22 self‐efficac* 23 empower* 24 #7 or #8 or #9 or #10 or #11 or #12 or #13 or #14 or #15 or #16 or #17 or #18 or #19 or #20 or #21 or #22 or #23 25 #6 and #24
MEDLINE (Ovid) (Date of most recent search: 23 January 2020)	1. exp Pulmonary Disease, Chronic Obstructive/ 2. (obstruct$ adj3 (pulmonary or lung$ or airway$ or airflow$ or bronch$ or respirat$)).tw. 3. (COPD or AECOPD or AECB).tw. 4. or/1‐3 5. Self Care/ 6. Self‐Management/ 7. Education/ 8. Patient Education as Topic/ 9. educat$.ti,ab. 10. (self‐manag$ or self manage$).tw. 11. (self‐car$ or self car$).tw. 12. (patient cent$ or patient‐cent$).tw. 13. (patient‐focus$ or patient focus$).tw. 14. (management adj2 (plan$ or program$)).tw. 15. (behavior$ or behaviour$).tw. 16. (disease* adj2 management$).tw. 17. (self‐efficac$ or self efficac$).tw. 18. empower$.tw. 19. or/5‐18 20. 4 and 19 21. (controlled clinical trial or randomized controlled trial).pt. 22. (randomized or randomised).ab,ti. 23. placebo.ab,ti. 24. randomly.ab,ti. 25. trial.ab,ti. 26. groups.ab,ti. 27. or/21‐26 28. Animals/ 29. Humans/ 30. 28 not (28 and 29) 31. 27 not 30 32. 20 and 31 33. limit 32 to yr="2011 ‐Current"
EMBASE (Ovid) (Date of most recent search: 23 January 2020)	1. chronic obstructive lung disease/ 2. (obstruct$ adj3 (pulmonary or lung$ or airway$ or airflow$ or bronch$ or respirat$)).tw. 3. (COPD or AECOPD or AECB).tw. 4. or/1‐3 5. exp self care/ 6. education/ 7. patient education/ 8. educat$.ti,ab. 9. (self‐manag$ or self manage$).tw. 10. (self‐car$ or self car$).tw. 11. (patient cent$ or patient‐cent$).tw. 12. (patient‐focus$ or patient focus$).tw. 13. (management adj2 (plan$ or program$)).tw. 14. (behavior$ or behaviour$).tw. 15. (disease* adj2 management$).tw. 16. (self‐efficac$ or self efficac$).tw. 17. empower$.tw. 18. or/5‐17 19. 4 and 18 20. Randomized Controlled Trial/ 21. randomization/ 22. controlled clinical trial/ 23. Double Blind Procedure/ 24. Single Blind Procedure/ 25. Crossover Procedure/ 26. (clinica$ adj3 trial$).tw. 27. ((singl$ or doubl$ or trebl$ or tripl$) adj3 (mask$ or blind$ or method$)).tw. 28. exp Placebo/ 29. placebo$.ti,ab. 30. random$.ti,ab. 31. ((control$ or prospectiv$) adj3 (trial$ or method$ or stud$)).tw. 32. (crossover$ or cross‐over$).ti,ab. 33. or/20‐32 34. exp animals/ or exp invertebrate/ or animal experiment/ or animal model/ or animal tissue/ or animal cell/ or nonhuman/ 35. human/ or normal human/ or human cell/ 36. 34 and 35 37. 34 not 36 38. 33 not 37 39. 19 and 38 40. limit 39 to yr="2011 ‐Current"
Clinicaltrial.gov (Date of most recent search: 23 January 2020)	Study type: Interventional Condition: copd Intervention: self‐management
WHO trials registry (Date of most recent search: 23 January 2020)	Condition: copd Intervention: self‐management

Data and analyses

Comparison 1. Self‐management versus usual care (primary outcomes).

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1.1 Health‐related quality of life (HRQoL): adjusted SGRQ total score (primary analysis)	14	2778	Mean Difference (IV, Random, 95% CI)	‐2.86 [‐4.87, ‐0.85]
1.2 Health‐related quality of life (HRQoL): adjusted SGRQ total score (secondary analysis)	14		Mean Difference (IV, Random, 95% CI)	Subtotals only
1.2.1 HRQoL: adjusted SGRQ total short term follow‐up (≤ 6 months)	7	1461	Mean Difference (IV, Random, 95% CI)	‐2.36 [‐4.92, 0.20]
1.2.2 HRQoL: adjusted SGRQ total medium term follow‐up (>6‐≤12 months)	13	2651	Mean Difference (IV, Random, 95% CI)	‐2.65 [‐4.78, ‐0.52]
1.3 Health‐related quality of life (HRQoL): CRQ domain scores (primary analysis)	5		Mean Difference (IV, Random, 95% CI)	Subtotals only
1.3.1 HRQoL: CRQ ‐ dyspnoea	5	738	Mean Difference (IV, Random, 95% CI)	0.13 [‐0.10, 0.35]
1.3.2 HRQoL: CRQ ‐ mastery	5	738	Mean Difference (IV, Random, 95% CI)	0.12 [‐0.09, 0.33]
1.3.3 HRQoL: CRQ ‐ fatigue	5	738	Mean Difference (IV, Random, 95% CI)	0.23 [‐0.01, 0.47]
1.3.4 HRQoL: CRQ ‐ emotional function	5	738	Mean Difference (IV, Random, 95% CI)	0.20 [‐0.06, 0.46]
1.4 Health‐related quality of life (HRQoL): CRQ domain scores (secondary analyses)	4		Mean Difference (IV, Random, 95% CI)	Subtotals only
1.4.1 HRQoL: CRQ ‐ dyspnoea (≤ 6 months)	3	386	Mean Difference (IV, Random, 95% CI)	0.23 [‐0.15, 0.61]
1.4.2 HRQoL: CRQ ‐ mastery (≤6 months)	3	386	Mean Difference (IV, Random, 95% CI)	0.12 [‐0.33, 0.57]
1.4.3 HRQoL: CRQ ‐ fatigue (≤6 months)	3	386	Mean Difference (IV, Random, 95% CI)	0.11 [‐0.55, 0.77]
1.4.4 HRQoL: CRQ ‐ emotional function (≤6 months)	3	386	Mean Difference (IV, Random, 95% CI)	0.22 [‐0.37, 0.82]
1.4.5 HRQoL: CRQ ‐ dyspnoea (>6‐≤12 months)	3	557	Mean Difference (IV, Random, 95% CI)	0.13 [‐0.09, 0.34]
1.4.6 HRQoL: CRQ ‐ mastery (>6‐≤12 months)	3	557	Mean Difference (IV, Random, 95% CI)	0.17 [‐0.07, 0.41]
1.4.7 HRQoL: CRQ ‐ fatigue (>6‐≤12 months)	3	557	Mean Difference (IV, Random, 95% CI)	0.22 [‐0.15, 0.59]
1.4.8 HRQoL: CRQ ‐ emotional function (>6‐≤12 months)	3	557	Mean Difference (IV, Random, 95% CI)	0.12 [‐0.27, 0.51]
1.5 Healthcare utilisation: respiratory‐related hospital admissions (number of participants with at least one admission) (primary analysis)	15	3263	Odds Ratio (M‐H, Random, 95% CI)	0.75 [0.57, 0.98]
1.6 Healthcare utilisation: respiratory‐related hospital admissions (number of participants with at least one admission) (secondary analyses)	15		Odds Ratio (M‐H, Random, 95% CI)	Subtotals only
1.6.1 Healthcare utilisation: respiratory‐related hospital admissions ≤ 6 months (number of participants with at least one admission)	4	437	Odds Ratio (M‐H, Random, 95% CI)	0.84 [0.45, 1.55]
1.6.2 Healthcare utilisation: respiratory‐related hospital admissions > 6 to ≤ 12 months (number of participants with at least one admission)	11	2826	Odds Ratio (M‐H, Random, 95% CI)	0.74 [0.53, 1.03]
1.7 Healthcare utilisation: respiratory‐related hospital admissions (mean number per participant) (primary analysis)	7	1572	Mean Difference (IV, Random, 95% CI)	‐0.29 [‐0.60, 0.01]
1.8 Healthcare utilisation: respiratory‐related hospital admissions (mean number per participant) (secondary analyses)	7		Mean Difference (IV, Random, 95% CI)	Subtotals only
1.8.1 Healthcare utilisation: respiratory‐related hospital admissions ≤ 6 months (mean number per participant)	3	709	Mean Difference (IV, Random, 95% CI)	‐0.01 [‐0.04, 0.02]
1.8.2 Healthcare utilisation: respiratory‐related hospital admissions >6 ‐ ≤12 months (mean number per participant)	6	1548	Mean Difference (IV, Random, 95% CI)	‐0.33 [‐0.68, 0.01]
1.9 Mortality: respiratory‐related mortality (primary analysis)	8	1572	Risk Difference (M‐H, Random, 95% CI)	0.01 [‐0.02, 0.04]
1.10 Mortality: respiratory‐related mortality (secondary analyses)	7		Risk Difference (M‐H, Random, 95% CI)	Subtotals only
1.10.1 Respiratory‐related ≤ 6 months mortality	4	309	Risk Difference (M‐H, Random, 95% CI)	‐0.01 [‐0.05, 0.03]
1.10.2 Respiratory‐related > 6 to ≤ 12 months mortality	6	1322	Risk Difference (M‐H, Random, 95% CI)	0.01 [‐0.01, 0.03]
1.11 Mortality: all‐cause mortality (primary analysis)	24	5719	Risk Difference (M‐H, Random, 95% CI)	‐0.01 [‐0.03, 0.01]
1.12 Mortality: all‐cause mortality (secondary analyses)	24		Risk Difference (M‐H, Random, 95% CI)	Subtotals only
1.12.1 All‐cause ≤ 6 months mortality	9	1678	Risk Difference (M‐H, Random, 95% CI)	0.00 [‐0.01, 0.01]
1.12.2 All‐cause > 6 to ≤ 12 months mortality	21	5240	Risk Difference (M‐H, Random, 95% CI)	‐0.01 [‐0.03, 0.01]

Comparison 2. Self‐management versus usual care (secondary outcomes).

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
2.1 Healthcare utilisation: all‐cause hospital admissions (number of participants with at least one admission)	10	2633	Odds Ratio (M‐H, Random, 95% CI)	0.88 [0.71, 1.08]
2.2 Healthcare utilisation: all‐cause hospital admissions (mean number per participant)	7	1914	Mean Difference (IV, Random, 95% CI)	‐0.01 [‐0.06, 0.04]
2.3 Healthcare utilisation: respiratory‐related hospitalisation days (per participant)	4	819	Mean Difference (IV, Random, 95% CI)	‐0.62 [‐2.27, 1.03]
2.4 Healthcare utilisation: all‐cause hospitalisation days (per participant)	6	2073	Mean Difference (IV, Random, 95% CI)	‐0.51 [‐1.85, 0.84]
2.5 Healthcare utilisation: emergency department visits (number of participants with at least one visit)	5	865	Odds Ratio (M‐H, Random, 95% CI)	0.53 [0.32, 0.87]
2.6 Healthcare utilisation: emergency department visits (mean number per participant)	6	1939	Mean Difference (IV, Random, 95% CI)	‐0.52 [‐0.89, ‐0.15]
2.7 Healthcare utilisation: GP visits (mean number per participant)	4	1113	Mean Difference (IV, Random, 95% CI)	‐0.21 [‐0.68, 0.25]
2.8 COPD exacerbations (mean number per participant)	7		Mean Difference (IV, Random, 95% CI)	Subtotals only
2.8.1 COPD exacerbations (regardless of definition)	7	1401	Mean Difference (IV, Random, 95% CI)	‐0.06 [‐0.26, 0.15]
2.8.2 COPD exacerbations (symptom based)	4	1047	Mean Difference (IV, Random, 95% CI)	0.05 [‐0.22, 0.31]
2.9 Courses of oral steriods (number of participants who used at least one course)	3	881	Odds Ratio (M‐H, Random, 95% CI)	4.19 [0.35, 50.65]
2.10 Courses of antibiotics (number of participants who used at least one course)	3	1012	Odds Ratio (M‐H, Random, 95% CI)	3.95 [1.37, 11.43]
2.11 Health status: modified Medical Research Council Dyspnoea Scale (mMRC)	3	356	Mean Difference (IV, Random, 95% CI)	‐0.31 [‐1.23, 0.60]
2.12 Health status: Hospital Anxiety and Depression Scale (HADS)	9		Mean Difference (IV, Random, 95% CI)	Subtotals only
2.12.1 Health status: Hospital Anxiety and Depression Scale (HADS) ‐ anxiety	9	1647	Mean Difference (IV, Random, 95% CI)	‐0.57 [‐1.01, ‐0.13]
2.12.2 Health status: Hospital Anxiety and Depression Scale (HADS) ‐ depression	9	1653	Mean Difference (IV, Random, 95% CI)	‐0.45 [‐0.80, ‐0.10]
2.13 Exercise capacity: six‐minute walk test (6MWT)	6	772	Mean Difference (IV, Random, 95% CI)	45.14 [9.16, 81.13]

Comparison 3. Subgroup analyses: self‐management versus usual care.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
3.1 Healthcare utilisation: respiratory‐related hospital admissions (subgroup by smoking cessation component)	15	3263	Odds Ratio (M‐H, Random, 95% CI)	0.75 [0.57, 0.98]
3.1.1 Smoking cessation component	6	1662	Odds Ratio (M‐H, Random, 95% CI)	1.10 [0.68, 1.79]
3.1.2 No smoking cessation component	9	1601	Odds Ratio (M‐H, Random, 95% CI)	0.60 [0.44, 0.82]

Characteristics of studies

Characteristics of included studies [ordered by study ID]

Benzo 2016.

Study characteristics
Methods	Design: RCT Follow‐up: 12 months Control group: usual care
Participants	Recruitment: hospital (inpatient) Assessed for eligibility: 969 Randomly assigned: SM: 108; UC: 107 Completed: SM: 92; UC: 106 Mean age: SM: 67.9 (SD 9.8) years; UC: 68.1 (SD 9.2) years Gender (% male): SM: 43; UC: 48 COPD diagnosis: GOLD criteria, confirmed with spirometry (FEV1/FVC ratio < 0.7) Inclusion of participants in acute phase: yes, during hospitalisation Major inclusion criteria: admission for a COPD exacerbation Major exclusion criteria: medical conditions that would impair their ability to participate in the study or to provide informed consent; receiving hospice care
Interventions	Mode: individual sessions at hospital outpatient clinics, telephone calls, educational booklet Duration: two face‐to‐face individual sessions (first visit 120 min, second visit not reported) and 6 phone calls (mean duration 28.6 min (SD 10.0) Professional: (respiratory) nurse, respiratory therapist Assignment of case manager: yes, accessible to participant during the complete follow‐up period Self‐management components: self‐recognition of COPD exacerbations, use of a COPD exacerbation action plan, home based exercise or physical activity component, coping with breathlessness Self‐management topics: (maintenance) medication Behavioural change techniques: 5 clusters: goals and planning, feedback and monitoring, social support, shaping knowledge, repetition and substitution
Outcomes	1. Rate of COPD hospitalisation 2. Quality of life 3. Physical activity 4. Number of COPD exacerbations, based on emergency department visits, nurse triage, or urgent care clinics
Notes	Source of funding: supported by NHLBI grant R01 HL09468 (RB, principal investigator) from the National Institutes of Health Conflict of interest: none declared

Bischoff 2012.

Study characteristics
Methods	Design: RCT Follow‐up: 24 months Control group: usual care
Participants	Recruitment: general practice Assessed for eligibility: 748 Randomly assigned: SM: 55; UC: 55 Completed: SM: 49; UC: 44 Mean age: SM: 65.5 (SD 11.5) years; UC: 63.5 (SD 10.3) years Gender (% male): SM: 67; UC: 51 COPD diagnosis: GOLD criteria, confirmed with spirometry (FEV1/FVC ratio < 0.7) Inclusion of participants in the acute phase: not reported Major inclusion criteria: aged at least 35 years, post‐bronchodilator ratio of FEV1/FVC < 0.70 Major exclusion criteria: post‐bronchodilator FEV1 < 30% predicted, treatment by a respiratory physician, severe comorbid conditions with a reduced life expectancy, inability to communicate in the Dutch language, and objections to one or more of the modes of disease management used in the study
Interventions	Mode: individual sessions at the general practice, paper modules "Living well with COPD", telephone calls Duration: 2 to 4 individual face‐to‐face sessions of one hour each, scheduled over 4 to 6 consecutive weeks; 6 telephone calls to reinforce self‐management skills Professional: practice nurse of each participating practice Assignment of case managers: yes, accessible to participants during the complete follow‐up period Self‐management components: smoking cessation, self‐recognition of COPD exacerbations, COPD exacerbation action plan, exercise / physical activity component (optional), diet, medication, coping with breathlessness, managing anxiety and stress Self‐management topics: keeping a healthy and fulfilling lifestyle Behavioural change techniques: 2 clusters: goals and planning, feedback and monitoring
Outcomes	1. Change from baseline in health‐related quality of life (CRQ) 2. Change in CRQ domain scores 3. Exacerbation frequency and management, based on symptoms 4. Total and five domain scores for self‐efficacy (CSES)
Notes	A third group of participants (n = 55) were assigned to routine monitoring through scheduled periodic monitoring visits as an adjunct to usual care. However, this group did not include an action plan. Source of funding: this study was funded by the Netherlands Organisation for Health Research and Development (ZonMw) and Partners in Care Solutions for COPD (PICASSO). Conflict of interest: no authors received any support from any company for the submitted work; no authors have any relationship with any company that might have an interest in the submitted work in the previous three years; no other relationships or activities that could appear to have influenced the submitted work.

Bösch 2007.

Study characteristics
Methods	Design: RCT Follow‐up: 12 months Control group: usual care
Participants	Recruitment: outpatient clinic Assessed for eligibility: not reported Randomly assigned: SM: 38; UC: 12 Completed: SM: 30; UC: 11 Mean age: SM: 63.8 (SD 8.4) years; UC: 64.6 (SD 6.8) years Gender (% male): 63% of 41 participants who completed the study; the distribution of males per group is not reported COPD diagnosis: GOLD criteria (FEV1/FVC ratio < 0.7), confirmed by authors Inclusion of participants in the acute phase: not reported Major inclusion criteria: diagnosis of COPD with obstruction proven by spirometry and a FEV1/FVC < 70% Major exclusion criteria: comorbidities which significantly influence symptoms, capacity or spirometry (symptomatic cardiopulmonary disease)
Interventions	Mode: group sessions (six to eight participants) at the participants' homes Duration: four face‐to‐face group sessions of two hours each with the final session scheduled six weeks later Professional: respiratory nurse under supervision of a respiratory specialist Assignment of case managers: yes, accessible to participants during the complete follow‐up period Self‐management components: smoking cessation (optional), self‐recognition of COPD exacerbations, use of a COPD exacerbation action plan, home‐based exercise or physical activity component, diet, COPD medication intake (i.e. adherence, inhalation technique), coping with breathlessnes, leisure activities and travelling Self‐management topics: not reported Behavioural change techniques: 2 clusters: goals and planning, feedback and monitoring
Outcomes	1. mMRC 2. Courses of antibiotics 3. FEV1 (L) 4. Hospital admissions 5. 6MWT 6. COPD exacerbations, based on treatment with antibiotics
Notes	Sources of funding: not reported Conflict of interest: none declared

Bourbeau 2003.

Study characteristics
Methods	Design: RCT Follow‐up: 12 and 24 months Control group: usual care
Participants	Recruitment: hospital (outpatient) Assessed for eligibility: not reported Randomly assigned: SM: 96; UC: 95 Completed: SM: 86; UC: 79 Mean age: SM: 69.4 (SD 6.5) years; UC: 69.6 (SD 7.4) years Gender (% male): SM: 52; UC: 59 COPD diagnosis: FEV₁ after the use of a bronchodilator between 25% and 70% of the predicted normal value and FEV₁/FVC ratio less than 70% Inclusion of participants in the acute phase: no Major inclusion criteria: hospitalised at least once in the preceding year for an exacerbation; stable COPD (respiratory symptoms and medication unchanged for at least 4 weeks before enrolment); at least 50 years of age; current or previous smoker (at least 10 pack‐years); FEV₁ after the use of a bronchodilator between 25% and 70% of the predicted normal value 14 and FEV₁/FVC ratio < 70%; no previous diagnosis of asthma, left congestive heart failure, terminal disease, dementia, or uncontrolled psychiatric illness; no participation in a respiratory rehabilitation programme in the past year; and no long‐term‐care facility stays Major exclusion criteria: participants with asthma as a primary diagnosis and those with major comorbidities (documented left ventricular failure and any terminal disease), dementia or uncontrolled psychiatric illness
Interventions	Mode: individual sessions at the participant's home, "Living well with COPD" programme with patient workbook, telephone calls Duration: seven face‐to‐face individual sessions of one hour each, scheduled in seven to eight consecutive weeks, 18 telephone calls (weekly calls for the eight weeks' educational period; after eight weeks, monthly phone calls for 12 months) Professional: experienced health professionals (nurses, respiratory therapists, a physiotherapist) who acted as case managers with the supervision and collaboration of the treating physician Assignment of case managers: "The programme was supervised by experienced and trained health professionals..." (Bourbeau 2006, p. 586) “Half‐day training sessions were dedicated to interactive lecturing sessions on each aspect of COPD given by different members of the multidisciplinary team. The rest of the training days included workshops oriented toward how to assess patient needs and the acquisition of motivational and teaching skills using group discussion, demonstration and practice of techniques, case scenarios, and role modeling" (Bourbeau 2006, p. 1705). The case‐manager was accessible to participants during the complete follow‐up period. Self‐management components: self‐recognition of COPD exacerbations, use of a COPD exacerbation action plan, home‐based exercise or physical activity component (optional), diet, COPD medication intake (i.e. adherence, inhalation technique), coping with breathlessness, leisure activities and travelling, energy conservation during day‐by‐day activities, relaxation exercises, adopting a healthy lifestyle, long‐term oxygen (optional) Self‐management topics: smoking cessation, exercise Behavioural change techniques: 7 clusters: goals and planning, feedback and monitoring, shaping knowledge, comparison of behaviour, associations, repetition and substitution, antecedents
Outcomes	1. Hospital admissions 2. Scheduled and unscheduled physician visits 3. Emergency department visits 4. Health‐related quality of life (SGRQ) 5. Pulmonary function 6. Functional exercise capacity 7. COPD exacerbations, based on symptoms
Notes	Completed first year of follow‐up: N = 165 (based on hospital registry database) Completed second year of follow‐up: N = 175 (based on provincial health insurance and hospitalisation database records) Source of funding: this study was funded by an unrestricted grant from Boehringer Ingelheim Canada, Burlington, Ontario, in partnership with the Fonds de la Recherche en Santé du Québec (FRSQ), Montreal, Quebec. Conflict of interest: none declared

Bringsvor 2018.

Study characteristics
Methods	Design: RCT Follow‐up: 3 months Control group: usual care
Participants	Recruitment: hospital (outpatient) Assessed for eligibility: 2309 Randomly assigned: SM: 92; UC: 90 Completed: SM: 55; UC: 70 Mean age: SM: 68.5 (SD 8.16) years; UC: 69.3 (SD 9.02) years Gender (% male): SM: 59; UC: 63 COPD diagnosis: GOLD criteria, confirmed with spirometry (FEV1/FVC ratio <0.7) Inclusion of participants in acute phase: no Major inclusion criteria: registered ICD‐10 code J44.0, 1, 8, or 9 after 1 January 2010; age ≥ 18 years; confirmed COPD grade II to IV, according to the GOLD; and the ability to read and speak Norwegian Major exclusion criteria: substantial cognitive impairment reported in a medical journal (e.g. severe dementia, severe Alzheimer’s disease), substantial alcohol or drug abuse, or both, or a life expectancy < 12 months due to comorbidity
Interventions	Mode: group sessions at meeting locations in the participants' home municipalities Duration: 11 face‐to‐face group sessions (120 min) scheduled weekly Professional: (respiratory) nurse, physiotherapist (co‐moderator) Assignment of case managers: yes, accessible to participants during the complete follow‐up period Self‐management components: smoking cessation (optional), self‐recognition of COPD exacerbations, use of a COPD exacerbation action plan, home‐based exercise or physical activity component, diet, COPD medication intake (i.e. adherence, inhalation technique), coping with breathlessness, psychological issues, information about the healthcare system, including local, regional and national “offers” for persons with COPD Self‐management topics: not reported Behavioural change techniques: 2 clusters: goals and planning, feedback and monitoring
Outcomes	1. Lung function 2. Dyspnoea (mMRC) 3. CAT 4. Self‐management (HeiQ version 2) 5. GSE 6. SOC‐13
Notes	Source of funding: this work was supported by the Western Norway Regional Health Authority [grant number 2013/911836] and the Norwegian Extra Foundation for Health and Rehabilitation [grant number 2015/RB13639] Conflict of interest: none declared

Bucknall 2012.

Study characteristics
Methods	Design: RCT Follow‐up: 12 months Control group: usual care
Participants	Recruitment: hospital (inpatient) Assessed for eligibility: 1405 Randomly assigned: SM: 232; UC: 232 Completed: SM: 211; UC: 200 Mean age: SM: 70.0 (SD 9.3) years; UC: 68.3 (SD 9.2) years Gender (% male): SM: 38; UC: 35 COPD diagnosis: chronic irreversible airflow limitation with FEV₁ < 70% predicted and a FEV₁ /FVC ratio of < 70% Inclusion of participants in the acute phase: not reported Major inclusion criteria: admitted to hospital with an acute exacerbation of COPD Major exclusion criteria: a history of asthma or left ventricular failure, evidence of active malignant disease or any evidence of confusion/poor memory, assessed with the abbreviated mental test (scores of 9/10 or 10/10 required)
Interventions	Mode: individual sessions at the participant's home, adapted "Living well with COPD" booklets, telephone calls Duration: four face‐to‐face individual sessions of 40 minutes each, scheduled fortnightly, over a two‐month period. There were also 828 phone calls to the intervention group participants (mean 4.6 phone calls per intervention participant). There were at least 6 subsequent home visits (but more frequently on request) thereafter for a total of 12 months Professional: study nurse Assignment of case managers: "Study nurses’ training was based on self regulation theory" (Bucknall 2012, p. 2). "Nurses were trained to deliver a structured self‐management programme in four fortnightly home visits (…). Nurses without previous respiratory training completed three half day training sessions" (p. 3). Case managers were accessible to participants during the complete follow‐up period. Self‐management components: smoking cessation, self‐recognition of COPD exacerbations, use of a COPD exacerbation action plan, diet (optional), COPD medication intake (i.e. adherence, inhalation technique), coping with breathlessness Self‐management topics: exercise Behavioural change techniques: 8 clusters: goals and planning, feedback and monitoring, social support, shaping knowledge, natural consequences, comparison of behaviour, repetition and substitution, self‐belief
Outcomes	1. Time to first acute hospital admission with a COPD exacerbation 2. Death due to COPD within 12 months of randomisation 3. Morbidity (change from baseline at 6 and 12 months in SGRQ) 4. Likelihood of anxiety or depression (HADS) 5. Sense of self‐efficacy (CSES) 6. Quality of life (EuroQol 5D)
Notes	Self‐management materials based on the "Living Well with COPD" programme and previously adapted for the UK population and healthcare setting by an iterative process, were used (p. 2). Extra information from author: "We used adapted “Living with COPD” booklets and daily diary cards (Stockley et al. – originally developed for use in Bronchiecistasis, piloted these and adapted them for this study, to include a line for recording steroid and antibiotic usage." Source of funding: in addition to funding from the Chief Scientist Office, Scottish Health Department (CZH/4/246), this study was supported by educational grants from Boehringer Ingelheim, GlaxoSmithKline, and Astra Zeneca. Conflict of interest: in addition to the Chief Scientist Office grant (CZH/4/246), CEB’s institution received financial support for the employment of a research fellow from Boehringer Ingelheim, GlaxoSmithKline, and Astra Zeneca, and JC holds other grants; no financial relationships with any organisations that might have an interest in the submitted work in the previous three years; no other relationships or activities that could appear to have influenced the submitted work.

Coultas 2005.

Study characteristics
Methods	Design: RCT Follow‐up: 6 months Intervention 1: nurse‐assisted medical management (MM) Intervention 2: nurse‐assisted collaborative management (CM) Control group: usual care
Participants	Recruitment: primary care clinics Assessed for eligibility: 217 Randomly assigned: MM: 72; CM: 72; UC: 73 Completed: MM: 49; CM: 51; UC: 51 Mean age: MM: 68.3 (SD 6.6) years; CM: 70.1 (SD 7.0) years; UC 68.8 (SD 10.4) years Gender (% male): MM: 42.9%; CM: 32.7%; UC: 53.8% COPD diagnosis: COPD‐related diagnosis code (International Classification of Diseases, Ninth Revision: codes 491, 492, 496), FEV1 < 80%; FEV1/FVC < 70%, confirmed by authors Inclusion of participants in acute phase: no Major inclusion criteria: current or former smoker with at least a 20‐pack‐year smoking history, at least one respiratory symptom (e.g. cough, shortness of breath or wheeze), airflow obstruction (i.e. FEV1/FVC ratio, 70%; and FEV1, 80% predicted) during the past 12 months Major exclusion criteria: not reported
Interventions	Mode: MM: enhance participant knowledge. CM: enhance participant knowledge and facilitating the adoption of healthy behaviour including lifestyle and self‐management skills Duration: 1 face‐to‐face individual session (mean 64 min ± 23.1), mean 6.0 ± 1.8 telephone calls (10.0 min ± 5.4) Professional: nurse Assignment of case manager: no Self‐management components: smoking cessation (optional), self‐recognition of COPD exacerbations (optional), use of a COPD exacerbation action plan (optional), COPD medication intake (i.e. adherence, inhalation technique) (optional) Self‐management topics: coping with breathlessness, review of symptoms and medications, education about COPD symptoms and medications Behavioural change techniques: MM: 2 clusters: goals and planning, feedback and monitoring. CM: 3 clusters: goals and planning, feedback and monitoring, social support
Outcomes	1. Health status 2. SGRQ 3. SF‐36 4. Perceived illness intrusiveness 5. Doctor visits 6. ER visits 7. Hospital admissions
Notes	Note 1: baseline characteristics are given only for the group of participants who completed the six‐month follow‐up period. Note 2: dropout percentages are high: MM: 32.0%; UC: 30.1%. Note 3: participants who dropped out of the study had more severe airflow obstruction, higher levels of distress and lower quality of life compared with participants who completed the study. Note 4: content of the interventions is not described properly, whereas the training of the nurses providing the intervention was described in detail. Note 5: outcome measures of self‐efficacy and social support and BSI‐18 and CES‐D scores were measured but not reported in the article. Source of funding: a grant from Robert Wood Johnson Foundation Conflict of interest: not reported

Emery 1998.

Study characteristics
Methods	Design: RCT Follow‐up: 2.5 months Control group: usual care
Participants	Recruitment: announcements, word of mouth, advertisements in weekly newspapers for older adults and physician referral Eligible: 92 Randomly assigned: SM: 25; UC: 25 Completed: SM: 23; UC: 25 Mean age: SM: 67.4 (SD 5.9) years; UC: 67.4 (SD 7.1) years Gender (% male): SM: 40; UC: 48 COPD diagnosis: airflow obstruction demonstrated on spirometry (i.e. the FEV1/FVC) Inclusion of participants in acute phase: yes, during hospitalisation Major inclusion criteria: stable COPD; > 50 years; FEV1/VC < 70; > six months of clinical symptoms of COPD Major exclusion criteria: significant cardiac disease; other diseases affecting exercise tolerance or learning skills last three months; asthma without fixed obstruction
Interventions	Mode: group education sessions Duration: 26 face‐to‐face group sessions (16 lectures of 60 min and 10 management sessions of 60 min) Professional: clinical psychologist Assignment of case managers: not reported Self‐management components: self‐recognition of COPD exacerbations, COPD medication intake (i.e. adherence, inhalation technique), coping with breathlessness, relaxation exercises, coping skills training Self‐management topics: not reported Behavioural change techniques: 3 clusters: goals and planning, feedback and monitoring, regulation and substitution
Outcomes	1. Health status 2. SIP 3. HRQoL‐MHLC 4. Health knowledge test 5. FEV₁ % predicted
Notes	We disregarded the third arm because it was focused on pulmonary rehabilitation. Source of funding: this work was supported by grants from the National Heart, Lung and Blood Institute (HL45290) and the National Institute on Aging (AG00029). Conflict of interest: not reported

Fan 2012.

Study characteristics
Methods	Design: RCT Follow‐up: 12 months Control group: guideline‐based usual care
Participants	Recruitment: outpatient clinic Assessed for eligibility: 467 Randomly assigned: SM: 209; UC: 217 Completed: SM: 193; UC: 203 Mean age: SM: 66.2 (SD 8.4) years; UC: 65.8 (SD 8.2) years Gender (% male): SM: 97.6; UC: 96.3 COPD diagnosis: GOLD, a post‐bronchodilator ratio of FEV₁/FVC < 0.70 with an FEV₁ < 80% predicted. At baseline and 1‐year study visits, post‐bronchodilator spirometry performed according to ATS criteria. Inclusion of participants in the acute phase: no Major inclusion criteria: hospitalised for COPD in the 12 months before enrolment, post‐bronchodilator ratio of FEV₁ to FVC < 0.70 with an FEV₁ < 80% predicted, older than 40 years, current or past history of cigarette smoking (> 10 pack‐years), at least 1 visit in the past year to either a primary care or pulmonary clinic at a Veterans Affairs medical centre, no COPD exacerbation in the past 4 weeks, ability to speak English, and access to a telephone Major exclusion criteria: primary diagnosis of asthma or any medical conditions that would impair ability to participate in the study or to provide informed consent
Interventions	Mode: individual and group sessions at hospital outpatient clinics, telephone calls, educational booklet Duration: four face‐to‐face individual sessions of 90 minutes each, scheduled weekly. The individual lessons were reinforced during a group session and by six phone calls, one per month for three months and every three months thereafter. Professional: case manager (various health‐related professionals). Assignment of case managers: before starting the study, all case managers received a three‐day training course with workshops covering detailed aspects of the self‐management programme, and all were supervised by the site investigator. Case managers were accessible to participants during the complete follow‐up period. Self‐management components: self‐recognition of COPD exacerbations, use of a COPD exacerbation action plan, COPD medication intake (i.e. adherence, inhalation technique) Self‐management topics: not reported Behavioural change techniques: 9 clusters: goals and planning, feedback and monitoring, social support, shaping knowledge, natural consequences, associations, repetition and substitution, regulation, antecedents
Outcomes	1. Time from randomisation to first COPD hospitalisation 2. All‐cause mortality 3. Number of COPD exacerbations, based on symptoms 4. Health‐related quality of life 5. Patient satisfaction 6. Medication adherence 7. COPD‐related knowledge, skill acquisition and self‐efficacy
Notes	This multi site RCT of an educational and acute care management programme was stopped early when a safety monitoring board noted excess mortality in the intervention group. The mean follow‐up time was 250 days. Source of funding: Veterans Affairs Cooperative Study Program Conflict of interest: none declared

Ferrone 2019.

Study characteristics
Methods	Design: RCT Follow‐up: 12 months Control group: usual care
Participants	Recruitment: general practice Assessed for eligibility: 1186 Randomly assigned: SM: 84; UC: 84 Completed: SM: 72; UC: 74 Mean age: SM: 68.6 (SD 9.6) years; UC: 67.9 (SD 9.8) years Gender (% male): SM: 40.5; UC: 52.4 COPD diagnosis: GOLD criteria, post‐bronchodilator FEV1 of ≤ 70% after four puffs of salbutamol and FEV1/FVC ratio < 0.7 Inclusion of participants in acute phase: no Major inclusion criteria: ≥ 40 years of age, current or ex‐smokers with a minimum 10 pack‐year smoking history, a post‐bronchodilator FEV1 of ≤ 70% after four puffs of salbutamol and FEV1/FVC ratio < 0.7, a history of at least 2 exacerbations in the past 3 years or 1 exacerbation in the past year Major exclusion criteria: COPD exacerbation in the past 4 weeks, diagnosis of asthma prior to the age of 40 years, use of long‐term supplemental oxygen, comorbid illness that would interfere with study participation, scheduled for COPD rehabilitation, terminal illness
Interventions	Mode: individual sessions at general practice, phone calls Duration: 2 face‐to‐face individual sessions (first visit 60 min (baseline evaluation) and 5 to7 min (encounter with physician) and second visit of 45 min after 3 months) and either a phone call or face‐to‐face visit at 6 and 9 months (15 to 30 min each) Professional: respiratory specialist, CRE Assignment of case managers: yes, accessible to participants during the complete follow‐up period Self‐management components: smoking cessation (optional), self‐recognition of COPD exacerbations, use of a COPD exacerbation action plan, home‐based exercise or physical activity component, COPD medication intake (i.e. adherence, inhalation technique), coping with breathlessness Self‐management topics: diet, energy conservation, advanced care/end‐of‐life planning, travel planning, COPD pathophysiology Behavioural change techniques: 6 clusters: goals and planning, feedback and monitoring, shaping knowledge, natural consequences, associations, regulation
Outcomes	1. COPD‐related quality of life (CAT and CCQ) 2. Knowledge (Bristol Knowledge Questionnaire) 3. Predicted FEV1 and FEV1/FVC ratio 4. COPD exacerbations (required prednisolone or antibiotics, or both) 5. COPD‐related health service utilisation (including unscheduled physician and ED visits, and hospitalisation)
Notes	Source of funding: this study was funded by Asthma Research Group Windsor Essex Inc. through unrestricted project grants by GlaxoSmithKline and Pfizer Canada Ltd. Conflict of interest: MF and ZR reported grants from Pfizer and GlaxoSmithKline during the conduct of the study; outside the current work. CJL reported grants and personal fees from AstraZeneca, Boehringer Ingelheim, and Novartis; grants from Pfizer and Bayer; and personal fees from GlaxoSmithKline, outside the current work. The remaining authors declare no competing interests.

Gallefoss 1999.

Study characteristics
Methods	Design: RCT Follow‐up: 12 months Control group: usual care
Participants	Recruitment: hospital (outpatient) Assessed for eligibility: not reported Randomly assigned: SM: 31; UC: 31 Completed: SM: 26; UC: 27 Mean age: SM: 57 (SD 9) years; UC: 58 (SD 10) years Gender (% male): SM: 48; UC: 52 COPD diagnosis: FEV₁ equal to or higher than 40% and lower than 80% of predicted Inclusion of participants in the acute phase: not reported Major inclusion criteria: participants with COPD, < 70 years of age, a FEV₁ equal to or higher than 40% and lower than 80% of predicted Major exclusion criteria: not suffering from any serious disease such as unstable coronary heart disease, heart failure, serious hypertension, diabetes mellitus, kidney or liver failure
Interventions	Mode: individual and group sessions at an outpatient clinic Duration: 1 or 2 face‐to‐face individual sessions with a nurse and 1 or 2 face‐to‐face individual sessions with a physiotherapist of 40 minutes each. Two 2‐hour group education sessions (five to eight persons) were scheduled on two separate days. Professional: nurse, physiotherapist, pharmacist, medical doctor Assignment of case managers: specially trained nurse, accessible to participants during the complete follow‐up period Self‐management components: smoking cessation (optional), self‐recognition of COPD exacerbations, COPD medication intake (i.e. adherence, inhalation technique), coping with breathlessness Self‐management topics: exercise, diet Behavioural change techniques: 6 clusters: goals and planning, feedback and monitoring, shaping knowledge, natural consequences, associations, regulation
Outcomes	1. Health‐related quality of life (SGRQ and four simple questions) 2. Hospital admissions 3. Days lost from work 4. GP consultation 5. FEV₁ % predicted
Notes	Source of funding: Norwegian Medical Associations Fund for Quality Improvement Conflict of interest: not reported

Hernández 2015.

Study characteristics
Methods	Design: RCT Follow‐up: 12 (and 72 months passive follow‐up thereafter) Control group: usual care
Participants	Recruitment: hospital (outpatient) Assessed for eligibility: 860 Randomly assigned: SM: 71; UC: 84 Completed: SM: 54; UC: 55 Mean age: SM: 73 (SD 8) years; UC: 75 (SD 9) years Gender (% male): SM: 83; UC: 86 COPD diagnosis: a person not involved in the study identified the cases with COPD (ICD9‐CM 491, 492, 493 or 496) as the primary diagnosis for admission. However, lung function testing was also assessed before randomisation. COPD confirmed with spirometry (FEV1/FVC < 70%). Inclusion of participants in the acute phase: no Major inclusion criteria: clinically stable COPD participants with a history of at least two hospital admissions owing to severe respiratory exacerbations during two consecutive years. "We considered a broad spectrum of COPD diagnostic terms that include chronic obstructive inflammatory diseases; namely, emphysema, asthma, tuberculosis, chronic bronchitis and COPD, aged above 45 years and living at home within the healthcare area of the hospital (Barcelona‐Esquerra)" (p. 2). Major exclusion criteria: nursing home or not living in the area, participants in another randomised controlled trial, died prior to contact
Interventions	Mode: individual and group sessions at an outpatient clinic and at participants' homes Duration: at least one face‐to‐face individual session of 40 minutes at the participant's home within 72 hours after entry into the study by the primary care team (participants without mobility problems), four face‐to‐face individual sessions of 15 minutes education each at the participant's home by the primary care team (participants with mobility problems), one two‐hour individual or group educational programme of 40 minutes. Three group sessions for participants without mobility problems (two comprehensive assessments of 90 minutes each at the outpatient clinic and one 2‐hour educational programme) and for participants with mobility problems, the programme was done at home. In all visits, the nurses dedicated 15 minutes for education. Professional: specialised respiratory nurse, primary care team (physician, nurse and social worker) Assignment of case managers: the community care teams received training: a 2‐hour face‐to‐face educational training and 1‐day stay at the hospital ward, aiming at enhancing home‐based management of frail COPD participants. Case managers were accessible to participants during the complete follow‐up period. Self‐management components: smoking cessation (optional), self‐recognition of COPD exacerbations, use of a COPD exacerbation action plan, home‐based exercise or physical activity component, diet, COPD medication intake (i.e. adherence, inhalation technique), coping with breathlessness, comorbid condition (no further explanation regarding content) Self‐management topics: vaccination Behavioural change techniques: 7 clusters: goals and planning, feedback and monitoring, social support, shaping knowledge, comparison of behaviour, associations, antecedents
Outcomes	1. Mental status 2. Activities of daily living (Lawton index) 3. Anxiety and depression (HADS) 4. Health‐related quality of life (SGRQ) 5. Sleepiness (Epworth sleepiness scale) 6. 6MWT 7. Nocturnal pulse oximetry and body mass distribution 8. Exacerbations
Notes	Source of funding: this study was funded by NEXES (Supporting Healthier and Independent Living for Chronic Patients and Elderly). Conflict of interest: none declared

Johnson‐Warrington 2016.

Study characteristics
Methods	Design: RCT Follow‐up: 3 months Control group: usual care
Participants	Recruitment: hospital (inpatient) Assessed for eligibility: 464 Randomly assigned: SM: 38; UC: 39 Completed: SM: 35; UC: 36 Mean age: SM: 67.6 (SD 8.5) years; UC: 68.3 (SD 7.7) years Gender (% male): SM: 38.4; UC: 33.3 COPD diagnosis: COPD confirmed with spirometry (FEV1/FVC ratio <0.7) Inclusion of patients in acute phase: yes, during hospitalisation Major inclusion criteria: established diagnosis of COPD and grade 2–5 dyspnoea according to the Medical Research Council Major exclusion criteria: reason for admission was not an acute exacerbation of COPD, unable to safely participate in unsupervised exercise (i.e. due to psychiatric, locomotive, cardiac or neurological impairments), involved in other research, unable to read English, had previously received SPACE (Self‐management Program of Activity Coping and Education) for COPD or completed pulmonary rehabilitation within the previous 6 months, had four or more admissions in the previous 12 months
Interventions	Mode: individual session at the hospital, written educational information Duration: 1 face‐to‐face individual session (30 to 45 min) and 6 phone calls (5 to 20 min each) Professional: physiotherapist Assignment of case managers: yes, accessible to participants during the complete follow‐up period Self‐management components: smoking cessation (optional), self‐recognition of COPD exacerbations, use of a COPD exacerbation action plan, home‐based exercise, COPD medication intake, coping with breathlessness Self‐management topics: diet, correct device use Behavioural change techniques: 11 clusters: goals and planning, feedback and monitoring, social support, shaping knowledge, natural consequences, comparison of behaviour, associations, repetition and substitution, regulation, antecedents, identity
Outcomes	1. Respiratory‐related hospital readmission at 3 months 2. Quality of life (CRQ‐SR) 3. Anxiety and depression (HADS) 4. Bristol COPD Knowledge Questionnaire 5. ISWT 6. ESWT 7. Pulmonary Rehabilitation Adapted Index of Self‐Efficacy 8. Ready for home survey
Notes	Source of funding: SJS and KR were supported by the Collaboration for Leadership in Applied Health Research and Care, East and West Midlands, respectively, and the NIHR Leicester Respiratory Biomedical Research Unit (BRU). Conflict of interest: none declared

Jolly 2018.

Study characteristics
Methods	Design: RCT Follow‐up: 12 months Control group: usual care
Participants	Recruitment: general practice Assessed for eligibility: 1146 Randomly assigned: SM: 289; UC: 288 Completed: SM: 247; UC: 281 Mean age: SM: 70.7 (SD 8.8) years; UC: 70.2 (SD 7.8) years Gender (% male): SM: 63; UC: 64 COPD diagnosis: according to UK guidelines (FEV1/FVC ratio <0.7), confirmed by authors Inclusion of participants in acute phase: no Major inclusion criteria: on the practice COPD register, mild dyspnoea (MRC grades 1 (only breathless on strenuous exercise) or 2 (only get short of breath when hurrying on level ground or up a slight hill)), FEV1/FVC < 0.7 after post‐bronchodilator spirometry, aged 18 years or over Major exclusion criteria: level of dyspnoea of MRC grade 3 or greater, terminal disease or severe psychiatric disorder (confirmed by their GP)
Interventions	Mode: individually tailored written supportive materials (i.e. information leaflet, standard written information), followed by telephone calls Duration: 4 individual phone calls (first call 35 to 60 min, other calls 15 to 20 min) scheduled at 3, 7 and 11 weeks Professional: nurse Assignment of case managers: yes, accessible to participants during the complete follow‐up period Self‐management components: smoking cessation (optional), self‐recognition of COPD exacerbations (optional), use of a COPD exacerbation action plan (optional), physical activity, COPD medication intake (i.e. adherence, inhalation technique) Self‐management topics: coping with breathlessness Behavioural change techniques: 8 clusters: goals and planning, feedback and monitoring, social support, shaping knowledge, natural consequences, comparison of outcomes, regulation, antecedents
Outcomes	1. SGRQ‐C 2. MRC dyspnoea scale 3. Self‐reported physical activity 4. Psychological morbidity 5. Self‐efficacy (Stanford self‐efficacy scale) 6. Health state utility (EuroQoL 5 Dimensions 5 Levels)
Notes	‐

Jonsdottir 2015.

Study characteristics
Methods	Design: RCT Follow‐up: 12 months Control group: usual care
Participants	Recruitment: 291 Assessed for eligibility: 291 Randomly assigned: SM: 60; UC: 59 Completed: SM: 52; UC: 48 Mean age: SM: 59.4 (SD 4.7) years; UC: 58.7 (SD 4.4) years Gender (% male): SM: 39.6; UC: 51.9 COPD diagnosis: GOLD criteria (FEV1/FVC ratio < 0.7), confirmed by authors Inclusion of participants in acute phase: no Major inclusion criteria: aged 45 to 65 with mild and moderate COPD (grade II and III) as the primary disease Major exclusion criteria: another major disease (among them, individuals with asthma who had more than 200 mL or 12% increase in FEV1 after inhalation of 200 µg albuterol in the postbronchodilator spirometry), non‐Icelandic speaking, not capable of travelling to the treatment site, participated in a structured rehabilitation programme for people with COPD 6 months prior to the screening
Interventions	Mode: group and individual sessions at a clinical research centre located on a university‐hospital campus, followed by telephone calls Duration: 1 face‐to‐face group session (120 min), 3 to 4 face‐to‐face individual sessions (30 to 45 min), and 4 phone calls (5 to 10 min each) Professional: (respiratory) nurse, peer led, research team Assignment of case managers: no Self‐management components: smoking cessation (optional), self‐recognition of COPD exacerbations, physical exercise, diet, COPD medication intake (i.e. adherence, inhalation technique), coping with breathlessness, utilisation of health care, prevention of further decline of disease with the aim of enhancing health of participant and family, and coping with feelings of shame and guilt Self‐management topics: skills in managing treatment and consequences in daily life, knowledge about and skills in maintaining safe environment (pollution, cold/hot weather, smoke‐free environment, infections), skills in communication with family, relatives and health professionals Behavioural change techniques: 10 clusters: goals and planning, feedback and monitoring, social support, shaping knowledge, natural consequences, comparison of behaviour, repetition and substitution, regulation, identity, self‐belief
Outcomes	1. SGRQ‐C 2. IIRS 3. IPAQ short version 4. COPD exacerbations (self‐reported), measured by the question: ‘How often during the previous 6 months have you had a serious exacerbation of the lungs?’
Notes	Source of funding: this research was funded by the Icelandic Research Fund, University of Iceland’s Research Fund, Landspitali‐University Hospital’s Research Fund, Icelandic Nurses’ Association’s Research Fund, and the Oddur Olafsson Fund. Conflict of interest: none declared

Kessler 2018.

Study characteristics
Methods	Design: RCT Follow‐up: 12 months Control group: usual care
Participants	Recruitment: hospital (outpatient) Assessed for eligibility: not reported Randomly assigned: SM: 172; UC: 173 Completed: SM: 137; UC: 128 Mean age: SM: 67.3 (SD 8.9) years; UC: 66.6 (SD 9.6) years Gender (% male): SM: 69.4; UC: 69.8 COPD diagnosis: GOLD criteria, confirmed with spirometry (FEV1/FVC ratio < 0.7) Inclusion of participants in acute phase: not reported Major inclusion criteria: COPD patients (a post‐bronchodilator FEV1/FVC ratio ≤ 70%; an FEV1 < 50% of the predicted value), aged ≥ 35 years, a ≥ 10 pack‐year smoking history, at least one severe exacerbation in the previous year, could receive all relevant COPD treatments including long‐term oxygen therapy and home mechanical ventilation Major exclusion criteria: not expected to survive longer than 6 months, unable to read or speak the country language or had cognitive/psychiatric disease, on continuous treatment of > 10 mg per day prednisone or equivalent for more than 6 weeks, living in a nursing home
Interventions	Mode: 'Living Well with COPD programme', group and individual sessions, phone calls Duration: 1 face‐to‐face group session (90 to 120 min), 4 face‐to‐face individual sessions (60 to 90 min), and multiple phone calls (duration not specified) Professional: respiratory specialist, case manager Assignment of case managers: yes, accessible to participants during the complete follow‐up period Self‐management components: smoking cessation (optional), self‐recognition of COPD exacerbations, use of a COPD exacerbation action plan, exercise programme (optional), diet, COPD medication intake, coping with breathlessness Self‐management topics: not reported Behavioural change techniques: 3 clusters: goals and planning, feedback and monitoring, shaping knowledge
Outcomes	1. Unplanned all‐cause hospitalisation days 2. Number of COPD exacerbations, based on symptoms 3. 6MWD 4. BODE index 5. Anxiety and depression (HADS) 6. Health status (SGRQ‐C) 7. Safety (adverse events, serious adverse events and deaths)
Notes	Source of funding: Air Liquide Healthcare. Conflict of interest: JB, IDZ, PC, DK, ST, JLV, RWDN, and RK were investigators in the COMET trial and have received honoraria from Air Liquide Healthcare, sponsors of the COMET trial. DG was an employee of Air Liquide Healthcare at the time when the study was conducted. SR is a director of HEVA HEOR, which received consulting fees from Air Liquide Healthcare to perform a health economic analysis. The authors report no other conflicts of interest in this work.

Lenferink 2019.

Study characteristics
Methods	Design: RCT Follow‐up: 12 months Control group: usual care
Participants	Recruitment: hospital (outpatient) Assessed for eligibility: 1586 Randomly assigned: SM: 102; UC: 99 Completed: SM: 85; UC: 84 Mean age: SM: 68.8 (SD 9.0) years; UC: 68.2 (SD 8.9) years Gender (% male): SM: 64.7; UC: 63.6 COPD diagnosis: GOLD criteria, confirmed with spirometry (FEV1/FVC ratio < 0.7) Inclusion of participants in acute phase: no. Major inclusion criteria: a clinical diagnosis of COPD according to the GOLD criteria (FEV1 80% of the predicted value and FEV1/FVC < 0.70); 1 or more diagnostic comorbidities (ischaemic heart disease, history of myocardial infarction, angina pectoris, heart failure (defined according to the ESC guidelines), diabetes (steroid‐induced or stable diabetes type 1 or 2)); active symptoms of anxiety or depression, or both (using a cut‐off score of ≥ 11 from the HADS and/or having symptoms that are currently being treated); 3 or more COPD exacerbations, defined as respiratory problems that required a course of oral corticosteroids/antibiotics in the two years preceding study entry; and/or 1 or more hospitalisations for respiratory problems in the two years preceding study entry; ≥ 40 years of age; stable at the time of inclusion (at least 4 weeks post‐exacerbation, 6 weeks post‐hospitalisation or post‐rehabilitation); able to understand and read English or Dutch Major exclusion criteria: terminal cancer, end stage of COPD or another serious disease with low survival rate (expected survival < 12 months), other serious lung disease (e.g. α1‐antitrypsin deficiency; interstitial lung diseases), people with cognitive impairment (MMSE < 24), people who are currently enrolled in other randomised controlled trials or intensive case management programmes
Interventions	Mode: group and individual sessions at the hospital, written symptom diary and action plan, telephone calls Duration: 2 to 3 face‐to‐face group sessions (120 to 240 min), 2 face‐to‐face individual sessions (60 min), and 3 phone calls (10 to 15 min each) Professional: respiratory nurse and cardiac, mental health and/or diabetes nurses Assignment of case managers: yes, accessible to participants during the complete follow‐up period. Self‐management components: self‐recognition of COPD exacerbations, use of a COPD exacerbation action plan, COPD medication intake (i.e. adherence, inhalation technique), coping with breathlessness, self‐recognition of increase in comorbid symptoms and use of an action plan for these comorbidities (CHF, IHD, anxiety and depression) Self‐management topics: exercise, diet, knowledge regarding COPD and comorbidities, immunisations, physical fitness and relaxation exercises Behavioural change techniques: 5 clusters: goals and planning, feedback and monitoring, social support, shaping knowledge, repetition and substitution
Outcomes	1. COPD exacerbation days, based on symptoms 2. Number of COPD exacerbations, based on symptoms, per participant 3. Duration of COPD exacerbations, based on symptoms, per participant per year
Notes	Source of funding: this study was supported by the Lung Foundation Netherlands (grant number 3.4.11.061), Lung Foundation Australia (Australian Lung Foundation Boehringer Ingelheim COPD Research Fellowship 2010), Repat Foundation, GlaxoSmithKline (unrestricted grant) and Stichting Astma Bestrijding. Funding information for this article has been deposited with the Crossref Funder Registry. Conflict of interest: A Lenferink reports grants from Stichting Astmabestrijding and GlaxoSmithKline (unrestricted grant), during the conduct of the study. J van der Palen reports grants from Netherlands Lung Foundation, during the conduct of the study. PDLPM van der Valk has nothing to disclose. P Cafarella has nothing to disclose. A van Veen has nothing to disclose. S Quinn has nothing to disclose. CGM Groothuis‐Oudshoorn has nothing to disclose. MG Burt has nothing to disclose. M Young has nothing to disclose. PA Frith has nothing to disclose. TW Effing reports grants from The Repat Foundation, Australian Lung Foundation and Dutch Asthma Foundation, during the conduct of the study.

Liang 2019.

Study characteristics
Methods	Design: CRT Follow‐up: 12 months Control group: usual care
Participants	Recruitment: general practice Assessed for eligibility: 1050 Randomly assigned: SM: 157; UC: 115 Completed: SM: 138; UC: 100 Mean age: SM: 66.6 (SD 10.8) years; UC: 61.7 (SD 10.1) years Gender (% male): SM: 60.5; UC: 62.6 COPD diagnosis: GOLD criteria, confirmed with spirometry (FEV1/FVC ratio < 0.7) Inclusion of participants in acute phase: no Major inclusion criteria: ≥ 40 years old, ≥ 2 clinic visits during the previous year, self‐reported being a current/ex‐smoker (≥ 10 pack‐year smoking history), documented diagnosis of COPD on clinic records or were being treated with COPD‐specific medications Major exclusion criteria: terminal illness (anticipated survival < 12 months), unable to provide informed consent (e.g. cognitive impairment), pre‐existing interstitial lung disease, unstable cardiovascular status, comorbidities preventing participation in an exercise training programme, contraindications to spirometry, completed pulmonary rehabilitation in the previous 24 months
Interventions	Mode: individual sessions at the general practice; phone calls Duration: 3 face‐to‐face individual sessions (duration not specified), and 9 phone calls (duration not specified) Professional: physiotherapist, research assistants, pharmacist Assignment of case managers: no Self‐management components: smoking cessation (optional), home based exercise, COPD medication intake (i.e. adherence, inhalation technique) Self‐management topics: not reported Behavioural change techniques: 5 clusters: goals and planning, feedback and monitoring, social support, natural consequences, reward and threat
Outcomes	1. HRQoL (SGRQ) 2. CAT 3. Dyspnoea (mMRC) 4. Lung function (FEV1 % predicted) 5. Anxiety and depression (HADS) 6. HSI 7. Smoking abstinence
Notes	Source of funding: this study was supported by Boehringer Ingelheim, Eastern Melbourne Primary Health Network, Lung Foundation Australia and National Health and Medical Research Council. Conflict of interest: MJ Abramson reports grants from Boehringer Ingelheim, during the conduct of the study; grants from Pfizer, assistance with conference attendance and personal fees for consultancy from Sanofi, outside the submitted work. G Russell has nothing to disclose. AE Holland is a current member of the Lung Foundation Australia COPD‐X: Concise Guide for Primary Care Advisory Committee. NA Zwar is a current member of the Lung Foundation Australia COPD Guidelines Committee. B Bonevski has nothing to disclose. A Mahal has nothing to disclose. P Eustace has nothing to disclose. E Paul has nothing to disclose. K Phillips is the Lung Foundation Australia General Manager of Consumer Programs. The Lung Foundation Australia works in collaboration and receives funding from pharmaceutical companies outlined in the foundation’s annual reports (available at lungfoundation.com.au/about‐us/annual‐reports/). NS Cox has nothing to disclose. S Wilson has nothing to disclose. J George reports grants from Boehringer Ingelheim, during the conduct of the study; grants from Pfizer, and personal fees for consultancy from GlaxoSmithKline, outside the submitted work; and is a current member of the Lung Foundation Australia COPD Guidelines Committee. J Liang has nothing to disclose.

Martin 2004.

Study characteristics
Methods	Design: RCT Follow‐up: 12 months Control group: usual care
Participants	Recruitment: general practice Assessed for eligibility: not reported Randomly assigned: SM: 44; UC: 49 Completed: SM: 35; UC: 45 Mean age: SM: 71.1 (95% CI 68.7 to 73.5) years; UC: 69.1 (95% CI 63.5 to 74.7) years Gender (% male): SM: 34.1; UC: 65.3 COPD diagnosis: GOLD criteria (FEV1/FVC ratio < 0.7), confirmed by authors Inclusion of participants in acute phase: no (use of the plan was commenced at a time when each participant was in a stable condition) Major inclusion criteria: diagnosis of COPD, aged 55 or over, at least one hospital admission or two acute exacerbations of COPD requiring GP care during the previous 12 months, an MMSE score > 22 Major exclusion criteria: terminally ill, coexisting lung cancer, admission to hospital with cardiac disease within previous 12 months, receiving home oxygen therapy
Interventions	Mode: individual sessions at a general practice, hospital, ambulance service, emergency department or home‐based Duration: 4 face‐to‐face individual sessions and respiratory nurse visits at 3, 6 and 12 months Professional: respiratory physician, respiratory nurse, GP, ED consultant, medical staff Assignment of case managers: no Self‐management components: use of a COPD exacerbation action plan, COPD medication intake, guidance regarding treatment for coexisting conditions (e.g. when/how to use oxygen therapy, and when to use diuretics) Self‐management topics: smoking cessation, coping with breathlessness/breathing techniques, self‐recognition of exacerbations Behavioural change techniques: 3 clusters: goals and planning, feedback and monitoring, shaping knowledge
Outcomes	1. Health care utilisation (GP visits, hospital admissions, ambulance calls) 2. Quality of life (SGRQ) 3. Medication use (courses of oral steroids and antibiotics)
Notes	Three participants subsequently withdrew for personal reasons. However, it was not reported from which group(s) they withdrew. A further 13 people died during the follow‐up period (nine in the intervention group and four in the control group). Source of funding: this study was supported by South Link Health Inc., a nonprofit consortium of general practitioners. Conflict of interest: none declared

Mitchell 2014.

Study characteristics
Methods	Design: RCT Follow‐up: 6 months Control group: usual care
Participants	Recruitment: general practice Assessed for eligibility: 326 Randomly assigned: SM: 89; UC: 95 Completed: SM: 65; UC: 79 Mean age: SM: 69 (SD 8 years); UC: 69 (SD 10.1) years Gender (% male): SM: 60.7; UC: 49.5 COPD diagnosis: COPD confirmed by spirometry, with a FEV₁/FVC ratio < 0.7 Inclusion of participants in the acute phase: no Major inclusion criteria: have a diagnosis of COPD confirmed by spirometry, with a FEV₁/FVC ratio < 0.7, grade 2‐5 MRC dyspnoea scale, clinically stable for 4 weeks Major exclusion criteria: unable to undertake an exercise regime due to neurological, musculoskeletal or cognitive comorbidities, unable to read English to the reading age of an 8‐year‐old, completed pulmonary rehabilitation within the previous 12 months
Interventions	Mode: individual sessions at a GP's office or home‐based, telephone calls, workbook Duration: one face‐to‐face individual session for 30 to 45 minutes by a physiotherapist and two telephone calls at two and four weeks into the programme to reinforce skills and provide encouragement to progress Professional: physiotherapist, trainee health psychologist Assignment of case managers: yes, but after a second phone call, no access to the case manager Self‐management components: smoking cessation (optional), self‐recognition of COPD exacerbations, use of a COPD exacerbation action plan, home‐based exercise, management of psychological consequences (e.g. dealing with anger, depression, disease acceptance) Self‐management topics: diet, (maintenance) medication, coping with breathlessness Behavioural change techniques: 11 clusters: goals and planning, feedback and monitoring, social support, shaping knowledge, natural consequences, comparison of behaviour, associations, repetition and substitution, regulation, antecedents, identity
Outcomes	1. Health status (CRQ dyspnoea domain) 2. Fatigue, emotion and mastery domains of the CRQ 3. Disease knowledge (Bristol COPD Knowledge Questionnaire) 4. Anxiety and depression (HADS) 5. Exercise capacity (ISWT, ESWT) 6. Self‐efficacy (Pulmonary Rehabilitation Adapted Index of Self‐Efficacy) 7. Healthcare utilisation (admissions, GP visits, ED visits, nurse home visits) 8. Medication use (courses of antibiotics) 8. Self‐reported smoking status
Notes	Source of funding: National Institute for Health Research (NIHR) Conflict of interest: none declared

Rice 2010.

Study characteristics
Methods	Design: RCT Follow‐up: 12 months Control group: usual care
Participants	Recruitment: hospital (Veterans Affairs medical centres) Assessed for eligibility: 1739 Randomly assigned: SM: 372; UC: 371 Completed: SM: 336; UC: 323 Mean age: SM: 69.1 (SD 9.4) years; UC: 70.7 (SD 9.7) years Gender (% male): SM: 97.6%; UC: 98.4% COPD diagnosis: clinical diagnosis of COPD with post‐bronchodilator spirometry showing a FEV₁ < 70% predicted and a FEV₁/FVC < 0.70 Inclusion of participants in the acute phase: not reported Major inclusion criteria: a diagnosis of COPD at high risk of hospitalisation as predicted by one or more of the following during the previous year: hospital admission or ED visit for COPD, chronic home oxygen use or course of systemic corticosteroids for COPD Major exclusion criteria: inability to have access to a home telephone line or sign a consent form, any condition that would preclude effective participation in the study or likely to reduce life expectancy to less than a year
Interventions	Mode: group sessions at an outpatient clinic, one‐page handout summary and number for help line, telephone calls Duration: one face‐to‐face group session (60 to 90 min) by a respiratory therapist case manager, 12 monthly phone calls (10 to 15 minutes each) Professional: respiratory therapist case manager Assignment of case managers: "case managers were respiratory therapists who had completed a one‐day training session.” Appendix 1, p. 2. The case manager was accessible to participants during the complete follow‐up period. Self‐management components: smoking cessation (optional), self‐recognition of COPD exacerbations, use of a COPD exacerbation action plan, COPD medication intake Self‐management topics: exercise, oximetry, recommendation concerning influenza and pneumococcal vaccinations, instruction in hand hygiene Behavioural change techniques: 4 clusters: goals and planning, feedback and monitoring, social support, shaping knowledge
Outcomes	1. Hospital admissions and ED visits for COPD 2. All‐cause hospitalisations and all‐cause ED visits 3. Hospital and intensive care unit lengths of stay 4. Respiratory medication use 5. Change in respiratory quality of life (SGRQ) 6. All‐cause mortality
Notes	Source of funding: this study was supported by an unrestricted grant from the Veterans Integrated Service Network 23 Primary Care and Research Services and by the Center for Chronic Disease Outcomes Research, a Veterans Affairs Health Services Research and Development Center of Excellence Conflict of interest: several study authors (i.e. K.L.R., M.C., D.E.N.) reported that they or family members received financial benefits from a commercial entity. The other study authors (i.e. H.E.B., J.G., T.M.S., D.B.N., S.K., M.T., L.J.G., C.B., do not have financial relationships with a commercial entity that has an interest in the subject of this manuscript

Rose 2018.

Study characteristics
Methods	Design: RCT Follow‐up: 12 months Control group: usual care
Participants	Recruitment: hospital (inpatient) Assessed for eligibility: 8696 (2100 documented COPD diagnosis) Randomly assigned: SM: 237; UC: 238 Completed: SM: 207; UC: 191 Mean age: SM: 71 (SD 9.2) years; UC: 71 (SD 9.7) years Gender (% male): SM: 50; UC: 44 COPD diagnosis: GOLD criteria, confirmed with spirometry (FEV1/FVC ratio < 0.7) Inclusion of participants in acute phase: yes, on emergency department presentation and/or hospital admission for COPD exacerbation, or during attendance at respirology outpatient clinic Major inclusion criteria: COPD diagnosis according to GOLD criteria and published Canadian reference values confirmed by a respirologist or internist, ≥ 50 years of age, 1 or more emergency department visits or hospital admissions for COPD exacerbation in previous 12 months, and ≥ 2 prognostically‐important COPD‐associated comorbidities (as defined by GOLD and Canadian Thoracic Society Guidelines) identified via medical record screening Major exclusion criteria: primary diagnosis of asthma (action plans differ substantially), terminal diagnosis, dementia, uncontrolled psychiatric illness, inability to understand English, no telephone access, inability to attend follow‐up, resident in a long‐term care facility, enrolled in the provincial tele‐home monitoring programme, and no family physician
Interventions	Mode: individual session (standardised education session based on 'Living Well with COPD') at an outpatient clinic; telephone calls Duration: 1 face‐to‐face individual session (40 min), 21 phone calls (duration not specified). Professional: Case manager (nurse practitioner or respiratory therapist, both trained as COPD educators) Assignment of case managers: yes, accessible to participants during the complete follow‐up period Self‐management components: smoking cessation (optional), self‐recognition of COPD exacerbations, use of a COPD exacerbation action plan, COPD medication intake (i.e. adherence, inhalation technique), advance care planning Self‐management topics: exercise Behavioural change techniques: 3 clusters: goals and planning, feedback and monitoring, social support
Outcomes	1. Number of ED visits 2. Number of hospital admissions and hospitalised days 3. Mortality 4. Time to first ED presentation 5. BODE index 6. Quality of life (EQ‐5D‐3L, SGRQ) 7. Anxiety and depression (HADS) 8. Self‐efficacy (CSES) 9. Satisfaction (CSQ8) 10. Caregiver impact
Notes	Source of funding: this trial was funded through the Building Bridges to Integrate Care (BRIDGES) program led by the University of Toronto’s Departments of Medicine and Family and Community Medicine and funded through the Ministry of Health and Long Term Care. L. Rose holds a CIHR New Investigator Award. Funding information for this article has been deposited with the Crossref Funder Registry. Conflict of interest: none declared

Sanchez‐Nieto 2016.

Study characteristics
Methods	Design: RCT Follow‐up: 12 months Control group: usual care
Participants	Recruitment: hospital (outpatient) Assessed for eligibility: 250 Randomly assigned: SM: 51; UC: 45 Completed: SM: 47; UC: 38 Mean age: SM: 68.2 (SD 7.2) years; UC: 67.1 (SD 6.8) years Gender (% male): SM: 92.2; UC: 88.9 COPD diagnosis: post‐bronchodilator FEV1/FVC < 70% Inclusion of participants in acute phase: no Major inclusion criteria: clinical stability (at least in the 3 months prior to randomisation, with no change in medication or usual symptoms); active smoker or prior history of smoking of at least 10 pack‐years; post‐bronchodilator FEV1/FVC < 70%; normal cognitive status (assessed by the intersecting pentagons test) to read and understand written texts, and receive training in inhalation techniques or self‐care education sessions; physical status that allows for regular walking or exercise; no diagnoses of asthma, advanced heart failure, unstable ischaemic heart disease, terminal disease, dementia or uncontrolled psychiatric disorders; ability to read texts; no participation in any pulmonary rehabilitation program in the previous year Major exclusion criteria: not reported
Interventions	Mode: group and individual sessions at the hospital; written material with treatment instructions Duration: 1 face‐to‐face group session (40 min), and 3 face‐to‐face individual sessions (20 min each) Professional: respiratory specialist, nurse, physiotherapist Assignment of case managers: yes (telephone assistance to intervention participants), accessible to participants during the complete follow‐up period Self‐management components: self‐recognition of COPD exacerbations, use of a COPD exacerbation action plan, home‐based physical exercise, COPD medication intake Self‐management topics: main characteristics of the disease Behavioural change techniques: 3 clusters: goals and planning, feedback and monitoring, shaping knowledge
Outcomes	1. Combined number of hospital admissions, and A&E department visits for COPD exacerbations 2. Hospitalisatons for COPD exacerbations 3. A&E visits for COPD exacerbations 4. Lengths of stay 5. Antibiotic or glucocorticoid treatment 6. All‐cause mortality
Notes	Source of funding: Gas Medi SA, Boehringer Ingelheim, Chiesi, Menarini Conflict of interest: none declared

Tabak 2014.

Study characteristics
Methods	Design: RCT Follow‐up: 9 months Control group: usual care
Participants	Recruitment: hospital, primary care physiotherapy practices Assessed for eligibility: not reported (101 participants eligible) Randomly assigned: SM: 15; UC: 14 Completed: SM: 10; UC: 2 Mean age: SM: 64.1 (SD 9.0) years; UC: 62.8 (SD 7.4) years Gender (% male): SM: 50.0; UC: 50.0 COPD diagnosis: GOLD II‐IV, a clinical diagnosis of COPD according to the GOLD criteria, confirmed with spirometry (FEV1/FVC ratio < 0.7) Inclusion of participants in the acute phase: no Major inclusion criteria: fulfill COPE‐II study (effects of self‐treatment and an exercise programme within a self‐management programme in outpatients with COPD) criteria: no exacerbation in the month prior to enrolment, three or more exacerbations or one hospitalisation for respiratory problems in the 2 years preceding study entry, a computer with Internet access at home Major exclusion criteria: other serious disease with a low survival rate, other diseases influencing bronchial symptoms and/or lung function, severe psychiatric illness, uncontrolled diabetes mellitus or a hospitalisation for diabetes mellitus in the 2 years preceding the study, need for regular oxygen therapy, maintenance therapy with antibiotics, known Alpha‐1 antitrypsin deficiency, disorders or progressive disease seriously influencing walking ability
Interventions	Mode: individual and group sessions at the outpatient clinic, primary care physiotherapy practices and at the participant's home, web‐based teleconsultation module Duration: at least 1 face‐to‐face individual session by the primary care physiotherapist (no protocol for education, offered as blended care, depending on physiotherapist and participant) and a teleconsultation module. For research purposes, there was one intake by a physiotherapist for baseline measure activity coach and explanations. Furthermore, there were additional meetings after 1, 3, 6 and 9 months. Before the start of the programme, participants had to attend 2 group sessions of 90 minutes each by a nurse practitioner. Professional: respiratory nurse practitioner, respiratory physiotherapist Assignment of case managers: no Self‐management components: smoking cessation (optional), self‐recognition of COPD exacerbations, use of a COPD exacerbation action plan, home‐based exercise (web‐based) Self‐management topics: diet, (maintenance) medication, coping with breathlessness/breathing techniques Behavioural change techniques: 3 clusters: goals and planning, feedback and monitoring, shaping knowledge
Outcomes	1. Use of application 2. Adherence (online diary, exercise scheme) 3. Satisfaction (Client Satisfaction Questionnaire) 4. Hospitalisations (number and length of stay) 5. Emergency department visits 6. COPD exacerbations, based on symptoms 7. Level of activity (activity coach, accelerometer) 8. Self‐perceived activity levels (Baecke Phsycial Activity Questionnaire) 9. Exercise tolerance (6MWT) 10. Fatigue (Multidimensional Fatigue Inventory 20) 11. Health status (CCQ) 12. Dyspnoea (MRC) 13. Quality of life (EuroQol‐5D)
Notes	Source of funding: NL Agency, a division of the Dutch Ministry of Economic Affairs (grant CALLOP9089) Conflict of interest: none declared

Titova 2015.

Study characteristics
Methods	Design: RCT Follow‐up: 24 months Control group: usual care
Participants	Recruitment: hospital (inpatient) Assessed for eligibility: 199 Randomly assigned: SM: 91; UC: 81 Completed: SM: 51; UC: 49 Mean age: SM: 74.1 (SD 9.26) years; UC: 72.6 (SD 9.33) years Gender (% male): SM: 42.9; UC: 43.2 COPD diagnosis: GOLD criteria (FEV1/FVC ratio <0.7), confirmed by authors Inclusion of participants in the acute phase: yes, during hospitalisation Major inclusion criteria: admission due to AECOPD, COPD (GOLD stage III or IV, 2007), living in the Trondheim municipality, ability to communicate in Norwegian, ability to sign the informed consent form Major exclusion criteria: any serious diseases that might cause a very short lifespan (expected survival time less than six months)
Interventions	Mode: individual sessions at the participant’s home, telephone calls, e‐learning programme, “My COPD book” Duration: six face‐to‐face individual sessions (one at discharge, five joint visits at home at approximately 3 days, 14 days, 6 months, 12 months, and 24 months post‐ discharge) by the specialist nurse, one interactive 15‐minute e‐learning programme, at least 24 telephone calls (routine phone calls at least once a month and during COPD exacerbations) Professional: specialist nurse Assignment of case managers: not reported Self‐management components: self‐recognition of COPD exacerbations, use of a COPD exacerbation action plan, coping with breathlessness Self‐management topics: smoking cessation, (maintenance) medication Behavioural change techniques: 4 clusters: goals and planning, feedback and monitoring, social support, repetition and substitution
Outcomes	1. Hospital utilisation (admissions caused by AECOPD, in‐hospital days due to AECOPD) 2. Mortality 3. Inhaled medication use (long‐acting bronchodilators)
Notes	Source of funding: Central Norway Regional Health Authority and the Research Council of Norway Conflict of interest: none declared

Walters 2013.

Study characteristics
Methods	Design: CRT Follow‐up: 12 months Control group: usual care
Participants	Recruitment: general practice Assessed for eligibility: 1207 Randomly assigned: SM: 90; UC: 92 Completed: SM: 74; UC: 80 Mean age: SM: 68.2 (SD 7.9) years; UC: 67.3 (SD 7.6) years Gender (% male): SM: 54; UC: 51 COPD diagnosis: postbronchodilator FEV1/FVC < 0.7, FEV1 30‐80% Inclusion of participants in acute phase: not reported Major inclusion criteria: smoking history > 10 pack‐years, postbronchodilator FEV1/FVC < 0.7, FEV1 30% to 80%; able to complete procedures and provide informed consent Major exclusion criteria: unable to participate in self‐care activities due to mental or physical incapacity, end‐stage cancer, poor English language skills and nursing home resident
Interventions	Mode: mentor telephone call sessions Duration: 16 individual phone calls (30 min each) Professional: community health nurses Assignment of case managers: yes, accessible to participants during the complete follow‐up period Self‐management components: smoking cessation (optional), self‐recognition of COPD exacerbations, use of a COPD exacerbation action plan, physical activity (optional), diet (optional), COPD medication intake (optional), alcohol (optional), psychosocial (optional) Self‐management topics: not reported Behavioural change techniques: 5 clusters: goals and planning, feedback and monitoring, social support, shaping knowledge, self‐belief
Outcomes	1. Quality of life (SF‐36 and SGRQ) 2. Patients' self‐management behaviour and knowledge (PIH scale) 3. Self‐efficacy (SEMCD) 4. Anxiety and depression (HADS, CES‐D and PCL‐C) 5. Well‐being (SWLS) 6. Hospital admissions
Notes	Source of funding: this work was supported by the National Health and Medical Research Council (NHMRC) project grant ID490028, a Royal Hobart Hospital Research Foundation grant and a University of Tasmania Institutional Research Grant. Conflict of interest: Lung Foundation Australia/Boehringer Ingelheim chronic obstructive pulmonary disease (COPD) Research Fellowship for JW

Wang 2019.

Study characteristics
Methods	Design: RCT Follow‐up: 12 months Control group: usual care
Participants	Recruitment: hospital (inpatient) Assessed for eligibility: 479 Randomly assigned: SM: 77; UC: 77 Completed: SM: 72; UC: 71 Mean age: SM: 68.7 (SD 6.2) years; UC: 69.2 (SD 6.1) years Gender (% male): SM: 76.6; UC: 80.5 COPD diagnosis: GOLD criteria, confirmed with spirometry (FEV1/FVC ratio < 0.7) Inclusion of participants in acute phase: yes, during hospitalisation Major inclusion criteria: aged 40 years or older; diagnosis of Global Initiative for Chronic Obstructive Lung Disease (GOLD) stage II, III or IV; COPD documented by pulmonary function testing; participants hospitalised for acute exacerbation of COPD; willing to sign an informed consent form Major exclusion criteria: severe sensory or cognitive impairment or symptomatic ischaemic heart disease; a coexisting respiratory condition (e.g. asthma or lung cancer); inability to be contacted by phone/mobile phone; participation in another research program or inability to provide informed consent
Interventions	Mode: individual sessions at the hospital and at home; booklet; telephone calls Duration: 5 to 6 face‐to‐face individual sessions (45 min each), 3 home visits (45 to 60 min each), and weekly phone calls scheduled over 3 months (10 to 15 min each) Professional: nurse Assignment of case managers: not reported Self‐management components: smoking cessation (optional), home‐based exercise or physical activity, COPD medication intake (i.e. adherence, inhalation technique), coping with breathlessness, respiratory muscle training (pursed lip breathing and abdominal breathing), coughing techniques, long‐term home oxygen therapy (optional) Self‐management topics: not reported Behavioural change techniques: 5 clusters: goals and planning, feedback and monitoring, social support, shaping knowledge, natural consequeces
Outcomes	1. COPD‐related hospital admissons 2. Emergency department visits 3. Exercise tolerance (6MWT) 4. Health‐related quality of life (SGRQ) 5. Participant satisfaction (CTCPSQ)
Notes	Source of funding: the work described in this paper was supported by a grant from the Education Department of Guizhou Province, China. Conflict of interest: none declared

AECOPD: acute exacerbations of COPD; ATS: American Thoracic Society; A&E: accident and emergency; BODE: Body‐mass index, airflow Obstruction, Dyspnea, and Exercise; BSI‐18: Brief Symtom Inventory 18; CAT: COPD Assessment Test; CCQ: Clinical COPD Questionnaire; CES‐D: Centers for Epidemiologic Studies – Depression; CHF: congestive heart failure; COPD: chronic obstructive pulmonary disease; CM: collaborative management; CRE: certified respiratory educator; CRQ: Chronic Respiratory (Disease) Questionnaire; CSES: COPD Self‐Efficacy Scale; CSQ: Client Satisfaction Questionnaire; CTCPSQ: COPD Transitional Care Patient Satisfaction Questionnaire; ED: emergency department; ER: emergency room; ESC: European Society of Cardiology; ESWT: Endurance‐Shuttle Walk Test; EuroQol 5D: European Quality of Life Five Dimension; FEV1/FEV₁: forced expiratory volume in one second (measured in litres (L)); FVC: forced vital capacity; GOLD: Global initiative for chronic Obstructive Lung Disease; GSE: General Self‐Efficacy Scale; GP: general practitioner; HADS: Hospital Anxiety and Depression Scale; HeiQ: Health education impact Questionnaire; HRQoL: health‐related quality of life; HSI: Heaviness of Smoking Index; ICD: International Classification of Diseases; IHD: ischaemic heart disease; IIRS: Illness Intrusiveness Rating Scale; IPAQ: International Physical Activity Questionnaire; ISWT: Incremental Shuttle Walking Test; MHLC: Multidimensional Health Locus of Control; min: minute(s); MM: medical management; MMSE: Mini Mental State Examination; (m)MRC: (modified) Medical Research Council; PCL‐C: Post‐traumatic Stress Disorder Checklist; PIH: Partners in Health Scale; RCT: randomised controlled trial; SD: standard deviation; SEMCD: Self‐Efficacy for Managing Chronic Disease; SF‐36: 36‐Item Short Form Health Survey; SGRQ: St. George's Respiratory Questionnaire; SGRQ‐C: St. George's Respiratory Questionnaire ‐ COPD‐Specific Version; SIP: Sickness Impact Profile; SM: self‐management; SOC‐13: Sense of Coherence Scale – 13; SR: self‐reported; SWLS: Satisfaction With Life Scale; UC: usual care; 6MWD: six‐minute walk distance

Characteristics of excluded studies [ordered by study ID]

Study	Reason for exclusion
Aboumatar 2019	No verification of COPD
ACTRN12616001039471	One intervention component
Akinci 2011	Included in previous review update; non‐RCT / CRT
Ali 2018	No stratification on COPD
Altenburg 2015	No iterative process
Anonymous 2012	No self‐management / home‐based exercise programme
Ansari 2017	Non‐RCT / CRT
Apps 2013	Non‐RCT / CRT
Arbillaga‐Etxarri 2018	One intervention component
Barberan‐Garcia 2014	Non‐RCT / CRT
Barnestein‐Fonseca 2011	Not all participants have had their COPD diagnosis verified (with spirometry)
Baron 2011	No usual care / control group
Barradell 2017	No usual care / control group
Barradell 2018	No usual care / control group
Basri 2017	No self‐management / home‐based exercise programme
Bausewein 2012	No stratification on COPD
Bavarsad 2015	One intervention component
Beekman 2014	No self‐management / home‐based exercise programme
Bentley 2014	One intervention component
Benzo 2017	No verification of COPD
Benzo 2019	No usual care / control group
Berkhof 2014	No self‐management / home‐based exercise programme
Berkhof 2015	No self‐management / home‐based exercise programme
Bhadhuri 2019	Non‐RCT / CRT
Bi 2021	No usual care / control group
Billington 2015	No usual care / control group
Bischoff 2011	Non‐RCT / CRT
Blackstock 2016	No usual care / control group
Bohingamu 2018	No self‐management / home‐based exercise programme
Boland 2014	No iterative process
Bosma 2011	No self‐management / home‐based exercise programme
Bove 2016	One intervention component
Bower 2012	No stratification on COPD
Browne 2013	No usual care / control group
Buckingham 2015	No verification of COPD
Bunker 2012	No verification of COPD
Cameron‐Tucker 2011	Abstract of pilot study; no published results available
Cameron‐Tucker 2016	No usual care / control group
Carcereny 2016	One intervention component
Casanas 2019	No stratification on COPD
Casas 2006	Included in previous review update; not all participants meet COPD spirometry criteria
Casey 2012	No self‐management / home‐based exercise programme
Cecere 2012	No self‐management / home‐based exercise programme
Chan 2011	One intervention component
Chan 2016	No usual care / control group
Chang 2019	No usual care / control group
Chatwin 2014	No usual care / control group; no self‐management / home‐based exercise programme
Chavannes 2009	Included in previous review update; non‐RCT / CRT
Chen 2011	One intervention component
Cheng 2017	One intervention component
Christenhusz 2012	No usual care / control group
Chuang 2011	Included in previous review update; non‐RCT / CRT
Collinsworth 2018	Not all participants have had their COPD diagnosis verified by spirometry
Cordova 2016	One intervention component
Coultas 2013	No usual care / control group
Coultas 2014	No usual care / control group
Coultas 2016	No usual care / control group
Coventry 2019	Non‐RCT / CRT
Csikesz 2016	No abstract or published results available
Cully 2012	No verification of COPD; no stratification on COPD
Cully 2017	No verification of COPD; no stratification on COPD
Dabrowska 2017	One intervention component
Davis 2016	Study was stopped due to participant recruitment issues
De Jongh 2013	No self‐management / home‐based exercise programme
Demeyer 2017a	One intervention component
Demeyer 2017b	No verification of COPD
Deng 2013	Non‐RCT / CRT
De Roos 2018	No self‐management / home‐based exercise programme
De San Miguel 2013	No self‐management / home‐based exercise programme
Dewan 2011b	No stratification on COPD
Dimitri 2012	No usual care / control group
Dogan 2017	Non‐RCT / CRT
Donesky 2012	No usual care / control group
Doward 2017	Non‐RCT / CRT
Drennan 2014	No self‐management / home‐based exercise programme
DRKS00006021	One intervention component
Due 2014	Non‐RCT / CRT
Durheim 2014	No usual care / control group
Durheim 2015	No usual care / control group
Dwinger 2013	No stratification on COPD
Effing 2009	Included in previous review update; no usual care / control group
Effing 2011	Included in previous review update; no usual care / control group
Emme 2014	No self‐management / home‐based exercise programme
Etxarri 2017	One intervention component
EUCTR2013‐002671‐18‐AT	No usual care / control group
Fairbrother 2011	Non‐RCT / CRT
Fairbrother 2013	Non‐RCT / CRT
Farmer 2014	No usual care / control group
Faulkner 2010	Included in previous review update; no home‐based exercise component
Ferreira 2016	No self‐management / home‐based exercise programme
Fish 2012	No stratification on COPD
Fitzsimmons 2011	No self‐management / home‐based exercise programme
Fitzsimmons 2016	No self‐management / home‐based exercise programme
Flink 2017	No stratification on COPD
Folch‐Ayora 2019	Not all participants have had their COPD diagnosis verified with spirometry
Foot 2017	No self‐management / home‐based exercise programme
Fors 2018	No iterative process
Fortin 2013	No stratification on COPD
Freund 2016	No verification of COPD; no stratification on COPD
Frith 2017a	No usual care / control group
Frith 2017b	No usual care / control group
Gaeckle 2016	No self‐management / home‐based exercise programme
Garcia‐Aymerich 2007	No verification of COPD
Gellis 2012	No stratification on COPD
Goossens 2014	No self‐management / home‐based exercise programme
Goris 2013	One intervention component
Grabenhorst 2013	No self‐management / home‐based exercise programme
Granados‐Santiago 2019	No self‐management / home‐based exercise programme
Gurgun 2011	No self‐management / home‐based exercise programme
Hæsum 2017	No iterative process
He 2015	Abstract only; no full‐text article available
Heaton 2019	No iterative process
Hegelund 2019	No iterative process
Heslop‐Marshall 2018	No iterative process
Hilberink 2011	No verification of COPD
Ho 2016	No self‐management / home‐based exercise programme
Houben 2014	No self‐management / home‐based exercise programme
Houben 2019	No self‐management / home‐based exercise programme
Howard 2014	No usual care / control group
Huang 2017	Not all participants have had their COPD diagnosis verified with spirometry
ISRCTN30110012	No usual care / control group
ISRCTN32281812	Study was stopped due to participant recruitment issues
ISRCTN77785397	No usual care / control group
Janaudis‐Ferreira 2018	No self‐management / home‐based exercise programme
Jarab 2012	No iterative process
Jennings 2015	One intervention component
Ji 2019	One intervention component
Jokar 2012	No verification of COPD
Jonkers 2012	No verification of COPD
Kalter‐Leibovici 2018	No usual care / control group
Kanabar 2015	No self‐management / home‐based exercise programme
Kara 2004	Included in previous review update; no home‐based exercise programme.
Kato 2017	No self‐management / home‐based exercise programme
Kenealy 2015	No self‐management / home‐based exercise programme
Kennedy 2013	No stratification on COPD
Khan 2019	No self‐management / home‐based exercise programme
Kheirabadi 2008	Included in previous review update; no verification of COPD
Khoshkesht 2015	No verification of COPD
Kiser 2012	No verification of COPD
Ko 2016	No self‐management / home‐based exercise programme
Koff 2009	Included in previous review update; no home‐based exercise programme
Korsbakke 2016	No self‐management / home‐based exercise programme
Kruis 2011	No iterative process
Kruis 2014a	No iterative process
Kruis 2014b	No iterative process
Labrecque 2011	Non‐RCT / CRT
Lahham 2018	No usual care / control group
Lainscak 2013	No self‐management / home‐based exercise programme
Lam 2011	Non‐RCT / CRT
Larson 2017a	No usual care / control group
Larson 2017b	No usual care / control group
La Torre 2018	No stratification on COPD
Lavesen 2012	No self‐management / home‐based exercise programme
Lavesen 2016	No self‐management / home‐based exercise programme
Lee 2015	No iterative process
Leiva‐Fernandez 2011	One intervention component
Leiva‐Fernandez 2012	One intervention component
Leiva‐Fernandez 2014	Not all participants have had their COPD diagnosis verified with spirometry
Li J 2015	Abstract only; no full‐text article available
Lilholt 2015	No self‐management / home‐based exercise programme
Lilholt 2016	No self‐management / home‐based exercise programme
Li P 2015	No iterative process
Li Z 2015	No published documents or information available
Lopez‐Lopez 2019	No self‐management / home‐based exercise programme
Luhr 2018	One intervention component
Marchioro 2011	No self‐management / home‐based exercise programme
Maricoto 2019	No stratification on COPD
Martinez 2014	Non‐RCT / CRT
Martinez 2019	One intervention component
McDonald 2011	Non‐RCT / CRT
Moayeri 2019	No verification of COPD
Monninkhof 2003	Included in previous review update; no home‐based exercise component
Morganroth 2011	No verification of COPD (based on ICD codes)
Morganroth 2016	No verification of COPD (based on ICD codes)
Moriyama 2015	Non‐RCT / CRT
Moullec 2008	Included in previous review update; non‐RCT / CRT
Moy 2015a	No verification of COPD (based on ICD codes)
Moy 2015b	No verification of COPD (based on ICD codes)
Moy 2016	No verification of COPD (based on ICD codes)
Mozaffari 2018	No verification of COPD
Murphy 2011	No self‐management / home‐based exercise programme
NCT01543217	No usual care / control group
NCT01867970	No stratification on COPD
NCT01871025	One intervention component
NCT01897298	One intervention component
NCT01921556	No self‐management / home‐based exercise programme
NCT01985529	No usual care / control group
NCT02035566	No published documents or other information available
NCT02078622	No verification of COPD
NCT02085161	No usual care / control group
NCT02567474	No iterative process
NCT02742597	No verification of COPD; no stratification on COPD
NCT02754232	Non‐RCT / CRT
NCT03387735	No stratification on COPD
NCT03654092	One intervention component
Ng 2017	No iterative process
Nguyen 2008	Included in previous review update; no usual care / control group
Nguyen 2009	Included in previous review update; no usual care / control group
Nguyen 2011	No usual care / control group
Nguyen 2012	No usual care / control group
Nguyen 2013	No usual care / control group
Nguyen 2016	One intervention component
Nguyen 2018	One intervention component
Nikoletou 2016	No usual care / control group
Ninot 2011	Included in previous review update; no home‐based exercise programme
NTR3945	One intervention component
Nyberg 2017	Non‐RCT / CRT
Nyberg 2019	Non‐RCT / CRT
O'Donnel 2018	Non‐RCT / CRT
O'Dwyer 2016	No verification of COPD; no stratification on COPD
Orme 2016	One intervention component
Orme 2018	No verification of COPD
Özkaptan 2016	Non‐RCT / CRT
Paneroni 2016	No self‐management / home‐based exercise programme
Papp 2017	No stratification on COPD
Pascual 2011	No stratification on COPD
Peian 2013	No verification of COPD
Perkins‐Porras 2018	No verification of COPD (physician diagnosis)
Phan 2015	No self‐management / home‐based exercise programme
Pinnock 2012	Non‐RCT / CRT
Pinnock 2013	No usual care / control group
Pommer 2012	No verification of COPD; no stratification on COPD
Pothirat 2015	Non‐RCT / CRT
Poureslami 2016	No usual care / control group
Pradella 2015	No self‐management / home‐based exercise programme
Rea 2004	Included in previous review update; no verification of COPD (by spirometry)
Renn 2018	No verification of COPD (based on ICD codes)
Rice 2011	Non‐RCT / CRT
Ritchie 2012	No verification of COPD; no stratification on COPD
Ritchie 2016	No verification of COPD
Rixon 2017	No verification of COPD
Roberts 2011	No usual care / control group
Robinson 2019	No usual care / control group
Rojas‐Gomez 2014	No iterative process
Russo 2015	No usual care / control group
Saini 2018	Non‐RCT / CRT
Sanchez 2018	No self‐management / home‐based exercise programme
Sanders 2012	Non‐RCT / CRT
Sassi‐Dambron 1995	Included in previous review update; no usual care / control group
Scalvini 2016	No self‐management / home‐based exercise programme
Schmidt 2018	No usual care / control group
Schou 2012	No self‐management / home‐based exercise programme
Schuz 2015	Non‐RCT / CRT
Scuffham 2018	No stratification on COPD
Seyedi 2018	No self‐management / home‐based exercise programme
Siddique 2012	No iterative process
Silva 2018	No usual care / control group
Silver 2017	No iterative process
Sinclair 2017	No stratification on COPD
Sink 2018	No self‐management / home‐based exercise programme
Slok 2014	No self‐management / home‐based exercise programme
Slok 2016a	No self‐management / home‐based exercise programme
Slok 2016b	No self‐management / home‐based exercise programme
Smidth 2014	No usual care / control group
Sohanpal 2012	No usual care / control group
Song 2014	Not all participants have had their COPD diagnosis verified with spirometry
Sorensen 2015	No verification of COPD; no iterative process
Soriano 2018	No self‐management / home‐based exercise programme
Stamenova 2019	No verification of COPD (by spirometry)
Steinhauser 2017	No verification of COPD; no stratification on COPD
Stenlund 2019	No verification of COPD
Steurer‐Stey 2014	No self‐management / home‐based exercise programme
Stolz 2018	No self‐management / home‐based exercise programme
Stoop 2015	No verification of COPD
Stulbarg 2002	Included in previous review update; no usual care / control group
Talboom‐Kamp 2017a	No RCT / CRT
Talboom‐Kamp 2017b	Non‐RCT / CRT
Tang 2012	No verification of COPD
Tashkin 2012	One intervention component
Taylor 2012	No iterative process
Theander 2015	No stratification on COPD
Thom 2018	Not all participants have had their COPD diagnosis verified with spirometry
Thoonsen 2011	No verification of COPD; no stratification on COPD
Titova 2016	Non‐RCT / CRT
To 2019	One intervention component
Tommelein 2014	No verification of COPD
Tong 2012	No self‐management / home‐based exercise programme
Torres‐Sanchez 2018	No self‐management / home‐based exercise programme
Touchette 2012	No verification of COPD; no stratification on COPD
Trappenburg 2011	No usual care / control group
Troosters 2011	No usual care / control group
Tsai 2016	One intervention component
Udsen 2014	Study protocol; no iterative process
Ulrik 2013	Non‐RCT / CRT
Valderramas 2018	No self‐management / home‐based exercise programme
Valenza 2018	No self‐management / home‐based exercise programme
Van der Weegen 2015	No stratification on COPD
Van Wetering 2009	Included in previous review update; no home‐based exercise programme
Vasilopoulou 2017	No self‐management / home‐based exercise programme
Vayisoglu 2019	No self‐management / home‐based exercise programme
Velardo 2017	No usual care / control group
Verwey 2014	No stratification on COPD
Vianello 2016	No self‐management / home‐based exercise programme
Vivodtzev 2012	One intervention component
Voncken‐Brewster 2015	No stratification on COPD
Vorrink 2016	One intervention component
Vorrink 2017	One intervention component
Wadell 2013	No self‐management / home‐based exercise programme
Wakabayashi 2011	Included in previous review update; no usual care / control group
Walker 2018	No self‐management / home‐based exercise programme
Walters 2012	Non‐RCT / CRT
Wan 2017	No usual care / control group
Wan 2019	No verification of COPD
Wang 2012	No self‐management / home‐based exercise programme
Wang 2018	No self‐management / home‐based exercise programme
Wang CH 2014	No usual care / control group
Wang H 2017	No self‐management / home‐based exercise programme
Wang J‐X 2017	No self‐management / home‐based exercise programme
Wang K 2017	No self‐management / home‐based exercise programme
Wang Y 2014	No iterative process
Wei 2014	One intervention component
Weldam 2016	No usual care / control group
Weldam 2017	No usual care / control group
Whelan 2019	No usual care / control group
White 2019	Non‐RCT / CRT
Wilson 2011	No usual care / control group
Wilson 2015	No usual care / control group
Windisch 2018	No usual care / control group
Wood‐Baker 2012	Non‐RCT / CRT
Wootton 2014	No self‐management / home‐based exercise programme
Wootton 2017	One intervention component
Wu M 2018	One intervention component
Wu W 2017	One intervention component
Wu W 2018	No self‐management / home‐based exercise programme
Wu X 2016	One intervention component
Xi 2015	No published documents or information available
Xin 2016	No iterative process
Yamaguti 2012	No self‐management / home‐based exercise programme
Yan 2017	No published documents or information available
Yan 2018	No self‐management / home‐based exercise programme
Yan 2019	No usual care / control group
Yan J 2016	No usual care / control group
Yan XN 2016	No usual care / control group
Yazdani 2018	No self‐management / home‐based exercise programme
Yilmaz 2017	One intervention component
Ying 2013	One intervention component
Yu 2013	No usual care / control group
Yu 2014	Non‐RCT / CRT
Yuan 2015	No stratification on COPD
Zakrisson 2019	No verification of COPD (based on ICD codes)
Zambom 2011	No self‐management / home‐based exercise programme
Zhai 2016	No published documents or information available
Zhang 2012	No published documents or information available
Zhang 2013	No self‐management / home‐based exercise programme
Zhang 2014	No published documents or information available
Zhang H 2016	One intervention component
Zhang M 2016	No self‐management / home‐based exercise programme
Zhao 2017	No published documents or information available
Zheng 2019	One intervention component
Zhou 2016	No usual care / control group
Zhu 2018	No self‐management / home‐based exercise programme
Zuo 2015	No published documents or information available
Zwar 2012	No verification of COPD
Zwar 2016	No verification of COPD
Zwerink 2013	No usual care / control group

Characteristics of studies awaiting classification [ordered by study ID]

Abdulsalim 2017.

Methods	Design: RCT Follow‐up: 24 months Control group: usual care
Participants	Recruitment: not reported. Assessed for eligibility: 328. Randomly assigned: SM: 130; UC: 130. Completed: SM: 104; UC: 98. Mean age: SM: 60.6 (SD 7.9) years; UC: 61.1 (SD 8.4) years. Gender (% male): SM: 96.9; UC: 94.4. COPD diagnosis: GOLD criteria, no spirometry (FEV1/FVC ratio <0.7) reported. Inclusion of participants in acute phase: not reported. Major inclusion criteria: confirmed diagnosis of COPD as per GOLD guidelines. Major exclusion criteria: not reported.
Interventions	Mode: 6 monthly counselling sessions, monthly phone calls, information leaflets. Duration: counselling sessions 15‐20 min each. Professional: pharmacist. Assignment of case manager: unclear. Self‐management components: unclear whether each included participant received at least two intervention components including an iterative process. Self‐management topics: unclear. Behavioural change techniques: at least goals and planning, feedback and monitoring, other unclear.
Outcomes	1. MAQ
Notes	More information regarding COPD spirometry, intervention components and iterative process needed.

Aboumatar 2017.

Methods	Design: RCT Follow‐up: 6 months Control group: usual care
Participants	Recruitment: hospital (inpatient). Assessed for eligibility: 969. Randomly assigned: not reported. Completed: not reported. Mean age: not reported per group. Gender (% male): not reported per group. COPD diagnosis: COPD diagnosis based on ICD9 codes 491.x , 492.x, 493.2, and 496. Inclusion of participants in acute phase: yes, during hospitalisation. Major inclusion criteria: admitted with a diagnosis of an acute COPD exacerbation; or, had a previous COPD diagnosis (ICD9 codes 491.x , 492.x, 493.2, and 496) and are receiving additional treatment to control COPD symptoms – (e.g. nebulizer treatments, steroids) in the current hospitalization. Major exclusion criteria: terminal illness with less than 6 months life expectancy.
Interventions	Mode: unclear. Duration: unclear. Professional: (respiratory) nurse. Assignment of case manager: unclear. Self‐management components: Tailored Transition Support,Individualized COPD selfmanagement education and support, Facilitated access to services. Self‐management topics: unclear. Behavioural change techniques: at least goals and planning, feedback and monitoring, other unclear.
Outcomes	1. combined number of COPD‐related hospitalizations and ED visits per participant at 6 months post discharge 2. quality of life (SGRQ)
Notes	More information regarding COPD spirometry, intervention components and iterative process needed.

Alharbey 2019.

Methods	Design: RCT Follow‐up: 12 months Control group: usual care
Participants	Recruitment: unclear. Assessed for eligibility: not reported. Randomly assigned: not reported. Completed: not reported. Mean age: not reported. Gender (% male): not reported. COPD diagnosis: not reported. Inclusion of participants in acute phase: not reported. Major inclusion criteria: not reported. Major exclusion criteria: not reported.
Interventions	Mode: unclear. Duration: unclear. Professional: (respiratory) nurse. Assignment of case manager: unclear. Self‐management components: unclear. Self‐management topics: unclear. Behavioural change techniques: unclear.
Outcomes	1. perceived awareness (UCOPD) 2. self‐efficacy (10 measurement items from a validated and reliable COPD self‐efficacy scale by Wigal) 3. perceived severity (HBM instrument by Champion) 4. behavioral intention (1‐item scale: “I intended to engage in the COPD recommended behavior.”)
Notes	More information regarding COPD spirometry, intervention components and iterative process needed.

Efraimsson 2008.

Methods	Design: RCT Follow‐up: 3 to 5 months Control group: usual care
Participants	Recruitment: nurse‐led primary healthcare clinic. Assessed for eligibility: 110. Randomly assigned: SM: 26, UC: 26. Completed: SM: 26, UC: 26. Mean age: SM: 66 (SD 9.4) years; UC: 67 (SD 10.4) years. Gender (% male): SM: 50.0, UC: 50.0. COPD diagnosis: mild, moderate, severe or very severe COPD based on spirometry, lung capacity after bronchodilator use, based on GOLD criteria. Inclusion of participants in the acute phase: not reported. Major inclusion criteria: diagnosed with mild, moderate, severe or very severe COPD based on spirometry, lung capacity after bronchodilator use, based on GOLD criteria. Major exclusion criteria: diagnosed severe mental disorders such as schizophrenia, dementia or alcohol or drug abuse.
Interventions	Mode: individual sessions at the outpatient and nurse‐led primary healthcare clinic Duration: two face‐to‐face individual sessions for self‐care education during 3‐5 months for one hour each by the nurse Professional: COPD nurse, physician, if needed: dietician, medical social worker, physical therapist, occupational therapist Training of case managers: not reported Self‐management components: action plan COPD exacerbations, iterative process with feedback on actions, self‐recognition of COPD exacerbations, education regarding COPD, smoking cessation, exercise or physical activity component Self‐management topics: smoking cessation, exercise, diet, (maintenance) medication, correct device use, coping with breathlessness/breathing techniques, other: instructions on the coughing technique to prevent infections and exacerbations, measurement on oxygen saturation before and after exertion, psycho‐social counselling and support, counselling on infection prevention Exercise programme: yes (optional), dialogue on physical activity and exercise. When needed, a dietician, a medical social worker, a physical therapist and an occupational therapist were consulted. Smoking cessation programme: yes (optional), motivational dialogue on smoking cessation based on Prochaska and DiClementes’ transtheoretical model of the stages of change. The model is based on open questions to help participants reflect on their smoking habits and empower patients to quit smoking. Behavioural change techniques: ten clusters: goals and planning, feedback and monitoring, social support, shaping knowledge, natural consequences, comparison of behaviour, associations, repetition and substitution, comparison of outcomes, reward and threat, regulation, antecedents, identity, scheduled consequences, self‐belief, covert learning. Action plan components: self‐recognition of exacerbations, self‐treatment of exacerbations, contact healthcare providers for support
Outcomes	1. health‐related quality of life (SGRQ) 2. smoking 3. COPD knowledge
Notes	Included in previous review update; more information regarding intervention components and iterative process needed

Ghanem 2010.

Methods	Design: RCT Follow‐up: 2 months Control group: usual care
Participants	Recruitment: hospital (inpatient) Assessed for eligibility: not reported. Randomly assigned: 39. Completed: 39. Mean age: SM: 56.96 (SD 11.59) years; UC: 56.43 (SD 9.03) years. Gender (% male): not reported. COPD diagnosis: moderate to severe COPD according to GOLD. Inclusion of participants in acute phase: yes, during hospitalisation. Major inclusion criteria: admission for a COPD exacerbation. Major exclusion criteria: unable to read or write, locomotor problems, cognitive impairment, ischaemic heart disease, aortic valve disease, cancer or lung disease other than COPD.
Interventions	Mode: individual sessions, face‐to‐face, booklet, home‐based. Duration: two face‐to‐face individual sessions (first visit 120 min, second visit not reported) and 6 phone calls (mean duration 28.6 min (SD 10.0). Professional: respiratory nurse, respiratory specialist. Assignment of case manager: yes, accessible for participant during the complete follow‐up period. Self‐management components: education regarding the disease, exercise programme, advice about nutrition, advice about medication. Iterative process unclear. Self‐management topics: unclear. Behavioural change techniques: unclear.
Outcomes	1. CRQ 2. SF‐36 3. FEV1 (L and % of predicted) 4. FEV1/FVC 5. 6MWT
Notes	Included in previous review update; more information regarding COPD spirometry, intervention components and iterative process needed.

Heidari 2018.

Methods	Design: RCT Follow‐up: 3 months Control group: usual care
Participants	Recruitment: participants referred to the clinic. Assessed for eligibility: 85. Randomly assigned: SM: 25; UC: 25. Completed: SM: 22; UC: 29. Mean age: SM: 59.54 (SD 7.43) years; UC: 60.05 (SD 5.17) years. Gender (% male): SM: 86.4; UC: 84.2. COPD diagnosis: GOLD criteria, confirmed with spirometry (FEV1/FVC ratio < 0.7). Inclusion of participants in acute phase: not reported. Major inclusion criteria: certified diagnosis of moderate or severe COPD by a pulmonologist according to the GOLD criteria, aged 45–70 years, a BMI of <30, being literate, having a strong understanding of the Persian language, having a constant prescription drug regime, and not suffering from another serious and restrictive disease (such as a major psychological disorder, neural disease, musculoskeletal disease, cancer, or cardiac or angina attack in the last month). Major exclusion criteria: hospitalization during the study, requiring use of oxygen or spray during the 6‐minute walking test, dealing with serious stress, failure to attend any personal or group education sessions, and noncompliance with a practical program that was determined at monthly visits (for intervention group).
Interventions	Mode: self‐management plan based on the 5A model. Duration: unclear. Professional: physician, nurse. Assignment of case manager: unclear. Self‐management components: unclear whether each included participant received at least two intervention components including an iterative process. Self‐management topics: unclear. Behavioural change techniques: at least goals and planning, feedback and monitoring, other unclear.
Outcomes	1. spirometry 2. 6MWT 3. Borg scale
Notes	More information regarding COPD spirometry, intervention components and iterative process needed.

Hill 2010.

Methods	Design: RCT Follow‐up: 3 months Control group: usual care
Participants	Recruitment: primary care setting. Assessed for eligibility: 131. Randomly assigned: 110. Completed: 93. Mean age: SM: 63.4 (SD 9.6) years; UC: 65.7 (SD 9.9) years. Gender (% male): SM: 44.0; UC: 46.5. COPD diagnosis: postbronchodilator ratio of FEV1/FVC < 0.7 and FEV1 < 80% predicted. Inclusion of participants in acute phase: not reported. Major inclusion criteria: confirmed diagnosis of COPD as per GOLD guidelines. Major exclusion criteria: unable to perform spirometry for a medical reason; unable to communicate in written or spoken English.
Interventions	Mode: individual sessions, face‐to‐face, written teaching manual, primary care practice. Duration: two individual sessions of one hour. Professional: certified COPD educator. Assignment of case manager: unclear. Self‐management components: unclear whether each included participant received at least two intervention components including an iterative process. Self‐management topics: unclear. Behavioural change techniques: unclear.
Outcomes	1. Bristol COPD Knowledge Questionnaire
Notes	Included in previous review update; more information regarding intervention components and iterative process needed

Jiang 2012.

Methods	Design: RCT Follow‐up: 10 months Control group: usual care
Participants	Recruitment: hospital (outpatient). Assessed for eligibility: 295. Randomly assigned: SM: 50; UC: 50. Completed: SM: 49; UC: 47. Mean age: SM: 65.2 (SD 8.96) years; UC: 64.7 (SD 8.05) years. Gender (% male): SM: 71.4; UC: 68.1. COPD diagnosis: moderate or severe COPD according to the Global Initiative for Chronic Obstructive Lung Disease (GOLD). Inclusion of participants in acute phase: not reported. Major inclusion criteria: moderate or severe COPD according to the GOLD, disease duration < 2 years since COPD diagnosis, at least one COPD exacerbation (defined as a complex of respiratory events/symptoms (increase or new onset) with a duration of 3 days requiring a change in treatment). Major exclusion criteria: unable to communicate clearly and give informed consent, concurrent oncologic or psychiatric diseases, drug or alcohol abuse history.
Interventions	Mode: weekly phone calls, audio CD, the self‐help manual, instruction booklet. Duration: 35 min per phone call. Professional: nurse. Assignment of case manager: unclear. Self‐management components: unclear whether each included participant received at least two intervention components including an iterative process. Self‐management topics: unclear. Behavioural change techniques: at least goals and planning, feedback and monitoring, other unclear.
Outcomes	1. uncertainty level (Mishel’s Uncertainty in Illness Scale‐Adult) 2. CSQ 3. anxiety (State‐Trait Anxiety Inventory scale) 4. depression (HADS‐depression) 5. quality of life (SF‐36)
Notes	More information regarding COPD spirometry, intervention components and iterative process needed.

Khdour 2009.

Methods	Design: RCT Follow‐up: 12 months Control group: usual care
Participants	Recruitment: hospital (outpatient clinic). Assessed for eligibility: not reported. Randomly assigned: 173. Completed: 143. Mean age: SM: 65.6 (SD 10.1) years; UC: 67.3 (SD 9.2) years Gender (% male): SM: 43.7; UC: 44.2. COPD diagnosis: confirmed diagnosis of COPD (by the hospital consultant) for at least 1 year, having a FEV of 30‐80% of the predicted normal value Inclusion of participants in acute phase: no. Major inclusion criteria: confirmed diagnosis of COPD for at least 1 year, having a FEV of 30‐80% of the predicted normal value and > 45 years old. Major exclusion criteria: congestive heart failure; moderate to severe learning difficulties (as judged by hospital consultant); attended a pulmonary rehabilitation programme in the last six months; severe mobility problems or terminal illness.
Interventions	Mode: individual sessions, face‐to‐face, telephone, hospital (outpatient clinic). Duration: one session of one hour, reinforcement at each outpatient visit every six months, two telephone calls at three and nine months. Professional: clinical pharmacist. Accessibility of case managers: not reported. Self‐management components: unclear. Self‐management topics: unclear. Behavioural change techniques: at least 2 clusters: goals and planning, feedback and monitoring, other unclear.
Outcomes	1. SGRQ 2. FEV1 3. Hospital admissions for acute exacerbations 4. ED visits for acute exacerbations 5. GP visits, scheduled and unscheduled 6. COPD knowledge questionnaire 7. Adherence to prescribed medication
Notes	Included in previous review update; more information from authors needed on COPD diagnosis, intervention components including an iterative process.

Li 2014.

Methods	Design: RCT Follow‐up: 3 months Control group: usual care
Participants	Recruitment: hospital (inpatient). Assessed for eligibility: 114. Randomly assigned: SM: 57; UC: 57. Completed: SM: 56; UC: 56. Mean age: SM: 70.91 (SD 9.17) years; UC: 72.18 (SD 8.53) years. Gender (% male): SM: 87.5; UC: 91.1. COPD diagnosis: COPD characterized by inhaled bronchodilator FEV1/FVC < 70%, FEV 1% predicted percentage < 80%. Inclusion of participants in acute phase: yes, during hospitalisation. Major inclusion criteria: a diagnosis of COPD characterized by inhaled bronchodilator FEV1/FVC < 70%, FEV 1% predicted percentage < 80%; ability to care for themselves during stable periods; and willingness to sign an informed consent form. Major exclusion criteria: a co‐existent medical problem (e.g, bronchial asthma, suspected malignancy, cardiac failure); cognitive impairment or lack of social support; or limb movement disorder.
Interventions	Mode: home‐based rehabilitation programme, phone calls, home visits. Duration: phone calls at 3, 5, 7, and 9 weeks after discharge, and home visits at 72 hours and 3 months post‐discharge. Professional: respiratory nurse, community nurse. Assignment of case manager: unclear. Self‐management components: unclear whether each included participant received at least two intervention components including an iterative process. Self‐management topics: unclear. Behavioural change techniques: at least goals and planning, feedback and monitoring, other unclear.
Outcomes	1. quality of life (SGRQ) 2. GHQ‐12 3. BMI
Notes	More information needed from authors on COPD diagnosis, intervention delivery, intervention components and an iterative process.

Liu 2013.

Methods	Design: RCT Follow‐up: 4 months Control group: usual care
Participants	Recruitment: not reported. Assessed for eligibility: not reported. Randomly assigned: SM: 29; UC: 28. Completed: not reported. Mean age: SM: 69.4 (SD 3.3) years; UC: 68.8 (SD 1.4) years. Gender (% male): SM: 72.4; UC: 82.1. COPD diagnosis: COPD according to the 2007 guidelines of the Chinese Society of Respiratory Disease. Inclusion of participants in acute phase: not reported. Major inclusion criteria: COPD according to the 2007 guidelines of the Chinese Society of Respiratory Disease,17 their clinical condition was stable at the time of inclusion, there was no history of bronchial asthma, a test for bronchiectasis was negative, no oral glucocorticoid treatment had been taken within the previous three months, and a computer with Internet access was available in the home. Major exclusion criteria: not reported.
Interventions	Mode: 6 monthly counselling sessions, monthly phone calls, information leaflets. Duration: counselling sessions 15‐20 min each. Professional: pharmacist. Assignment of case manager: unclear. Self‐management components: pursed‐lip breathing, deep inspiration‐slow blowing – making a fist, deep inhale‐holdingslow exhale, global exercise. Unclear whether each included participant received at least two intervention components including an iterative process. Self‐management topics: unclear. Behavioural change techniques: unclear.
Outcomes	1. pulmonary function tests (FEV1/FVC ratio) 2. exercise capacity (6MWT) 3. quality of life (SGRQ)
Notes	More information regarding COPD spirometry, intervention components and iterative process needed.

Lou 2015.

Methods	Design: RCT Follow‐up: 48 months Control group: usual care
Participants	Recruitment: healthcare units/centres in rural areas. Assessed for eligibility: 8,217. Randomly assigned: Self‐management (SM): 4,197; Usual care (UC): 4,020. Completed: SM: 3,418; UC: 2,803. Mean age: SM: 71.2 ± 7.4 years; UC: 71.5 ± 7.8 years. Gender (% male): SM: 47.8; UC: 47.9. COPD diagnosis: the subjects had to have a diagnosis of COPD according to the criteria proposed by the GOLD . Inclusion of participants in acute phase: no. Major inclusion criteria: at baseline, the subjects had to have a diagnosis of COPD according to the criteria proposed by GOLD . Major exclusion criteria: presence of fever, active tuberculosis, changes in radiographic images or medication in the 4 weeks immediately preceding recruitment, primary diagnosis of asthma or obvious bronchiectasis, cystic fibrosis, interstitial lung disease, previous lung‐volume‐reduction surgery, lung transplantation, pneumonectomy, uncontrolled or serious conditions that could potentially affect spirometry tests, and refusal to fill out psychological questionnaires.
Interventions	Mode: group and individual face‐to‐face sessions. Duration: 104 group sessions of 40‐60 minutes lecture each every 2 weeks, 104 individual follow‐up sessions at least once every two weeks. Every 2 months, the professionals examined the subjects collectively at the health‐care units. Professional: respiratory specialist, nurse psychologist, (respiratory) physiotherapist, peer led dietician, GPs, psychiatrists, rehabilitation specialists, other experts. Assignment of case manager: unclear. Self‐management components: education regarding COPD, smoking cessation, exercise or physical activity component, other: psychological counselling, review and adjustment of outpatient COPD medication. Unclear whether each included participant received at least two intervention components including an iterative process. Self‐management topics: unclear. Behavioural change techniques: at least goals and planning, feedback and monitoring, other unclear.
Outcomes	1. health status (BODE index) 2. changes in COPD knowledge, awareness and risk factors (survey) 3. changes in anxiety and depression symptoms (HADS) 4. changes in hospital admissions and ED visits 5. changes in medication regimens
Notes	More information regarding COPD spirometry, intervention components and iterative process needed.

Ozturk 2020.

Methods
Participants
Interventions
Outcomes
Notes	Potential eligible study screened from updated database search (January 2020 to March 2021). This study will be incorporated into the review at the next update, if inclusion criteria are met. No data extraction has been performed.

COPD: Chronic Obstructive Pulmonary Disease; CRQ: Chronic Respiratory (Disease) Questionnaire; CSQ: Client Satisfaction Questionnaire; ED: emergency department; FEV1: forced expiratory volume in one second; FVC: forced vital capacity; GHQ: General Health Questionnaire; GOLD: Global Initiative for Chronic Obstructive Lung Disease; GP: General Practitioner; HADS: Hospital Anxiety and Depression Scale; HMB: Health Belief Model; ICD: International Classification of Diseases; MAQ: Multidimensional Anxiety Questionnaire; RCT: Randomised Controlled Trial; SD: Standard deviation; SF‐36: 36‐Item Short Form Health Survey; SGRQ: St. George's Respiratory Questionnaire; SM: Self‐management; UC: Usual care; UCOPD: understanding COPD questionnaire; 6MWD: six‐minute walk distance

Characteristics of ongoing studies [ordered by study ID]

Boer 2011.

Study name
Methods
Participants
Interventions
Outcomes
Starting date
Contact information
Notes	More information needed from authors about intervention components, including whether the intervention involves an iterative process.

Bourne 2017.

Study name
Methods
Participants
Interventions
Outcomes
Starting date
Contact information
Notes	More information needed from authors about intervention components, including whether the intervention involves an iterative process.

Cecere Feemster 2013.

Study name
Methods
Participants
Interventions
Outcomes
Starting date
Contact information
Notes	More information needed from authors about intervention components, including whether the intervention involves an iterative process.

Chen 2018.

Study name
Methods
Participants
Interventions
Outcomes
Starting date
Contact information
Notes	More information needed from authors about intervention components, including whether the intervention involves an iterative process.

ChiCTR1800018197.

Study name
Methods
Participants
Interventions
Outcomes
Starting date
Contact information
Notes	More information needed from authors about intervention components, including whether the intervention involves an iterative process.

ChiCTR‐TRC‐12002559.

Study name
Methods
Participants
Interventions
Outcomes
Starting date
Contact information
Notes	More information needed from authors about intervention components, including whether the intervention involves an iterative process.

Chien 2016.

Study name
Methods
Participants
Interventions
Outcomes
Starting date
Contact information
Notes	More information needed from authors about intervention components, including whether the intervention involves an iterative process.

Costa 2015.

Study name
Methods
Participants
Interventions
Outcomes
Starting date
Contact information
Notes	More information needed from authors about intervention components, including whether the intervention involves an iterative process.

Dewan 2011a.

Study name
Methods
Participants
Interventions
Outcomes
Starting date
Contact information
Notes	More information needed from authors about intervention components, including whether the intervention involves an iterative process.

Ding 2019.

Study name	MH‐COPD
Methods
Participants
Interventions
Outcomes
Starting date
Contact information
Notes	More information needed from authors about intervention components, including whether the intervention involves an iterative process.

Doheny 2013.

Study name
Methods
Participants
Interventions
Outcomes
Starting date
Contact information
Notes	More information needed from authors about intervention components, including whether the intervention involves an iterative process.

Duran 2017.

Study name
Methods
Participants
Interventions
Outcomes
Starting date
Contact information
Notes	More information needed from authors about intervention components, including whether the intervention involves an iterative process.

Ergan 2018.

Study name
Methods
Participants
Interventions
Outcomes
Starting date
Contact information
Notes	More information needed from authors about intervention components, including whether the intervention involves an iterative process.

Fleehart 2015.

Study name
Methods
Participants
Interventions
Outcomes
Starting date
Contact information
Notes	More information needed from authors about intervention components, including whether the intervention involves an iterative process.

Gonzalez 2015.

Study name
Methods
Participants
Interventions
Outcomes
Starting date
Contact information
Notes	More information needed from authors about intervention components, including whether the intervention involves an iterative process.

Hernandez 2016.

Study name
Methods
Participants
Interventions
Outcomes
Starting date
Contact information
Notes	More information needed from authors about intervention components, including whether the intervention involves an iterative process.

Imanalieva 2016.

Study name
Methods
Participants
Interventions
Outcomes
Starting date
Contact information
Notes	More information needed from authors about intervention components, including whether the intervention involves an iterative process.

IRCT201504149014N61.

Study name
Methods
Participants
Interventions
Outcomes
Starting date
Contact information
Notes	More information needed from authors about intervention components, including whether the intervention involves an iterative process.

IRCT2017030432764N2.

Study name
Methods
Participants
Interventions
Outcomes
Starting date
Contact information
Notes	More information needed from authors about intervention components, including whether the intervention involves an iterative process.

James 2012.

Study name
Methods
Participants
Interventions
Outcomes
Starting date
Contact information
Notes	More information needed from authors about intervention components, including whether the intervention involves an iterative process.

Ko 2015.

Study name
Methods
Participants
Interventions
Outcomes
Starting date
Contact information
Notes	More information needed from authors about intervention components, including whether the intervention involves an iterative process.

Moreno 2017.

Study name
Methods
Participants
Interventions
Outcomes
Starting date
Contact information
Notes	More information needed from authors about intervention components, including whether the intervention involves an iterative process.

NCT02258646.

Study name
Methods
Participants
Interventions
Outcomes
Starting date
Contact information
Notes	More information needed from authors about intervention components, including whether the intervention involves an iterative process.

NCT02924870.

Study name
Methods
Participants
Interventions
Outcomes
Starting date
Contact information
Notes	More information needed from authors about intervention components, including whether the intervention involves an iterative process.

NCT03012256.

Study name
Methods
Participants
Interventions
Outcomes
Starting date
Contact information
Notes	More information needed from authors about intervention components, including whether the intervention involves an iterative process.

NCT03084874.

Study name
Methods
Participants
Interventions
Outcomes
Starting date
Contact information
Notes	More information needed from authors about intervention components, including whether the intervention involves an iterative process.

NCT03216603.

Study name
Methods
Participants
Interventions
Outcomes
Starting date
Contact information
Notes	More information needed from authors about intervention components, including whether the intervention involves an iterative process.

NCT03721315.

Study name
Methods
Participants
Interventions
Outcomes
Starting date
Contact information
Notes	More information needed from authors about intervention components, including whether the intervention involves an iterative process.

NL3827 (NTR4009).

Study name
Methods
Participants
Interventions
Outcomes
Starting date
Contact information
Notes	More information needed from authors about intervention components, including whether the intervention involves an iterative process.

NL5277 (NTR5558).

Study name
Methods
Participants
Interventions
Outcomes
Starting date
Contact information
Notes	More information needed from authors about intervention components, including whether the intervention involves an iterative process.

Padilla‐Zarate 2013.

Study name
Methods
Participants
Interventions
Outcomes
Starting date
Contact information
Notes	More information needed from authors about intervention components, including whether the intervention involves an iterative process.

Paquin 2014.

Study name
Methods
Participants
Interventions
Outcomes
Starting date
Contact information
Notes	More information needed from authors about intervention components, including whether the intervention involves an iterative process.

Reguera 2017.

Study name
Methods
Participants
Interventions
Outcomes
Starting date
Contact information
Notes	More information needed from authors about intervention components, including whether the intervention involves an iterative process.

Sano 2016.

Study name
Methods
Participants
Interventions
Outcomes
Starting date
Contact information
Notes	More information needed from authors about intervention components, including whether the intervention involves an iterative process.

Siddharthan 2018.

Study name
Methods
Participants
Interventions
Outcomes
Starting date
Contact information
Notes	More information needed from authors about intervention components, including whether the intervention involves an iterative process.

Sirichana 2014.

Study name
Methods
Participants
Interventions
Outcomes
Starting date
Contact information
Notes	More information needed from authors about intervention components, including whether the intervention involves an iterative process.

Thomas 2019.

Study name
Methods
Participants
Interventions
Outcomes
Starting date
Contact information
Notes	More information needed from authors about intervention components, including whether the intervention involves an iterative process.

Zanaboni 2016.

Study name
Methods
Participants
Interventions
Outcomes
Starting date
Contact information
Notes	More information needed from authors about intervention components, including whether the intervention involves an iterative process.

Differences between protocol and review

We changed the title to 'Self‐management interventions for people with chronic obstructive pulmonary disease', in line with recent self‐management studies.

We redrafted the Background section.

We expanded the Objectives to include the outcomes of interest.

For inclusion in this review, participants with COPD in included studies needed to have a diagnosis of COPD in line with the GOLD classification criteria (GOLD 2021), meaning a post‐bronchodilator FEV1 to FVC ratio of less than 0.70 (confirmed by (provided) spirometry data or confirmed in writing by authors).

We included only randomised controlled trials and cluster‐randomised trials that compared intervention to usual care in this update.

We included COPD self‐management intervention studies only if they incorporated at least two intervention components with an iterative process between healthcare provider and participant, in line with the most recent definition of COPD self‐management interventions (Effing 2016).

We also removed reference to modes of delivering the intervention and reframed as requiring interventions to be aimed at behaviour change.

For primary outcomes, we performed primary (using end point scores) and secondary analyses (using scores from different follow‐up times).

For dichotomous outcomes with few events, we reported risk differences with corresponding 95% confidence intervals.

We added mortality as a primary outcome. Hospitalisations were defined as respiratory‐related (primary outcome) or all‐cause (secondary outcome). We defined dyspnoea, impact of COPD on life, anxiety and/or depression as sub‐outcomes of health status. We also defined the use of other healthcare facilities, and added self‐management behaviours, patient activation and health literacy as secondary outcomes.

We used the Cochrane risk of bias 2 tool for the 2021 update.

We removed ‘Courses of steroids’ and ‘Exercise’ from the summary of findings table, and added 'Respiratory‐related mortality' and 'COPD exacerbations'.

We calculated the number needed to treat for an additional beneficial outcome (NNTB) for primary outcomes.

We performed sensitivity analysis on studies with contradictory results to investigate the robustness of the effect size, by excluding those studies from a meta‐analysis.

We performed sensitivity analyses on cluster‐randomised controlled trials using intra‐cluster correlation coefficients of 0.02 and 0.04, in case authors did not adjust their data for clustering and did not report such a coefficient.

We added the following extra subgroup analyses.

• The duration of the intervention (< 8 weeks versus ≥ 8 weeks).

• Inclusion of participants in the acute phase (having an acute exacerbation of COPD) versus stable state (at least four weeks post exacerbation and six weeks post hospitalisation).

• COPD self‐management interventions delivered in different income countries (self‐management interventions in low‐ and middle‐income countries versus high‐income countries).

• COPD self‐management interventions delivered in different care settings: primary care versus secondary and tertiary care.

• Smoking cessation component (inclusion of a smoking cessation component in the self‐management intervention versus no smoking cessation component in the self‐management intervention).

• Diet component (e.g. evaluation and optimisation of participants' diet and nutritional intake) (inclusion of a diet component in the self‐management intervention versus no diet component in the self‐management intervention).

• COPD medication component (e.g. advice about medication intake, adherence, inhalation technique) (inclusion of a medication component in the self‐management intervention versus no medication component in the self‐management intervention).

• Coping with breathlessness component (inclusion of a coping with breathlessness component in the self‐management intervention versus no coping with breathlessness component in the self‐management intervention).

• Self‐recognition of COPD exacerbations component (inclusion of a self‐recognition of COPD exacerbations component in the self‐management intervention versus no self‐recognition of COPD exacerbations component in the self‐management intervention).

• The effects of COPD self‐management interventions with and without use of digital technology.

• The integration of behavioural change techniques in COPD self‐management interventions.

Contributions of authors

Jade Schrijver coordinated the review, independently assessed the eligibility of titles, abstracts and full‐text versions of potentially relevant reports, independently extracted data from the included studies, assessed risk of bias for included studies, managed and analysed the data, generated the summary of findings table and wrote the review update.

Anke Lenferink independently assessed the eligibility of titles, abstracts and full‐text versions of potentially relevant reports, independently extracted data from the included studies, assessed risk of bias for included studies, double‐checked data entry, supported data management and data analysis, provided a methodological perspective, and critically revised the review update.

Marjolein Brusse‐Keizer independently assessed the eligibility of titles, abstracts and full‐text versions of potentially relevant reports, independently extracted data from the included studies, assessed risk of bias for included studies, double‐checked data entry, provided a methodological perspective, and critically revised the review update.

Marlies Zwerink independently assessed the eligibility of titles, abstracts and full‐text versions of potentially relevant reports, independently extracted data from the included studies, assessed risk of bias, and critically revised the review update.

Paul van der Valk independently assessed the eligibility of titles and abstracts, provided a clinical perspective, and critically revised the review update.

Job van der Palen independently assessed the eligibility of titles, abstracts and full‐text versions of potentially relevant reports, independently extracted data from the included studies, assessed risk of bias for included studies, double‐checked data entry, provided a methodological perspective, and critically revised the review update.

Tanja Effing independently assessed the eligibility of titles, abstracts and full‐text versions of potentially relevant reports, independently extracted data from the included studies, assessed risk of bias for included studies, double‐checked data entry, supported data management and data analysis, provided a methodological perspective, helped write the review, and critically revised the review update.

Contributions of editorial team

Rebecca Fortescue (Co‐ordinating Editor) edited the review; advised on methodology; approved the review prior to publication.

Emma Dennett (Managing Editor): co‐ordinated the editorial process; advised on interpretation and content; edited the review.

Emma Jackson (Assistant Managing Editor): conducted peer review; edited reference sections and other sections generally in the protocol and the review.

Elizabeth Stovold (Information Specialist): designed the search strategy; ran the searches; edited the search methods section.

Lucy Goldsmith: checked the data entry prior to the full write‐up of the review.

Sources of support

Internal sources

• All authors, Other

  The authors declare that no such funding was received for this systematic review

External sources

• All authors, Netherlands

  Dutch Foundation for Asthma Prevention

Declarations of interest

JS: is a researcher, employed by Medisch Spectrum Twente, Enschede, the Netherlands. She received funding to complete work on this review from the Dutch Foundation for Asthma Prevention, that was in no way able to influence the results of the review.

AL: is an epidemiologist and an assistant professor at the Health Technology and Services Research section, University of Twente, the Netherlands. She is also a researcher at Medisch Spectrum Twente, Enschede, the Netherlands. She coordinated the Lenferink 2019* study, included in this review. She has no conflict of interest with regard to the current review.

MB‐K: is an epidemiologist, employed by Medisch Spectrum Twente, Enschede, the Netherlands. She is also a researcher at the Health Technology and Services Research section, University of Twente, the Netherlands. She was involved in the study of Tabak 2014**, included in this review. She has no conflict of interest with regard to the current review.

MZ: is a data‐analyst and an epidemiologist, employed by Medisch Spectrum Twente, Enschede, the Netherlands. She coordinated the previous update of this review (Zwerink 2014).

PvdV: is a medical doctor, pulmonologist and respiratory researcher at the Department of Pulmonary Medicine, Medisch Spectrum Twente, Enschede, the Netherlands. He was involved in the Lenferink 2019* and Tabak 2014** studies, both included in this review. He has no conflict of interest with regard to the current review.

JvdP: is Professor of Evaluation and Assessment in Health Research at University of Twente, Enschede, the Netherlands. He is also an epidemiologist and research coordinator at Medisch Spectrum Twente, Enschede, the Netherlands. He was involved in the Lenferink 2019* and Tabak 2014** studies, both included in this review. He has no conflict of interest with regard to the current review.

TE: is an epidemiologist affiliated with Flinders University and University of Adelaide, Adelaide, Australia. She was involved in the Lenferink 2019* study, included in this review. She has no conflict of interest with regard to the current review.

* The Lenferink 2019 study was supported by the Lung Foundation Netherlands (grant number 3.4.11.061), Lung Foundation Australia (Australian Lung Foundation Boehringer Ingelheim COPD Research Fellowship 2010), Repat Foundation, GlaxoSmithKline (unrestricted grant) and Dutch Foundation for Asthma Prevention.
** The Tabak 2014 study was supported by the NL Agency, a division of the Dutch Ministry of Economic Affairs (grant CALLOP9089).

Edited (no change to conclusions)