Pulmonary rehabilitation for interstitial lung disease

Abstract

Background

Interstitial lung disease (ILD) is characterised by reduced functional capacity, dyspnoea and exercise‐induced hypoxia. Pulmonary rehabilitation is often used to improve symptoms, health‐related quality of life and functional status in other chronic lung conditions. There is accumulating evidence for comparable effects of pulmonary rehabilitation in people with ILD. However, further information is needed to clarify the long‐term benefit and to strengthen the rationale for pulmonary rehabilitation to be incorporated into standard clinical management of people with ILD. This review updates the results reported in 2014.

Objectives

To determine whether pulmonary rehabilitation in people with ILD has beneficial effects on exercise capacity, symptoms, quality of life and survival compared with no pulmonary rehabilitation in people with ILD.

To assess the safety of pulmonary rehabilitation in people with ILD.

Search methods

We searched CENTRAL, MEDLINE (Ovid), Embase (Ovid), CINAHL (EBSCO) and PEDro from inception to April 2020. We searched the reference lists of relevant studies, international clinical trial registries and respiratory conference abstracts to look for qualifying studies.

Selection criteria

We included randomised controlled trials and quasi‐randomised controlled trials in which pulmonary rehabilitation was compared with no pulmonary rehabilitation or with other therapy in people with ILD of any origin.

Data collection and analysis

Two review authors independently selected trials for inclusion, extracted data and assessed risk of bias. We contacted study authors to request missing data and information regarding adverse effects. We specified a priori subgroup analyses for participants with idiopathic pulmonary fibrosis (IPF) and participants with severe lung disease (low diffusing capacity or desaturation during exercise). There were insufficient data to perform the prespecified subgroup analysis for type of exercise training modality.

Main results

For this update, we included an additional 12 studies resulting in a total of 21 studies. We included 16 studies in the meta‐analysis (356 participants undertook pulmonary rehabilitation and 319 were control participants). The mean age of participants ranged from 36 to 72 years and included people with ILD of varying aetiology, sarcoidosis or IPF (with mean transfer factor of carbon dioxide (TLCO) % predicted ranging from 37% to 63%). Most pulmonary rehabilitation programmes were conducted in an outpatient setting, with a small number conducted in home‐based, inpatient or tele‐rehabilitation settings. The duration of pulmonary rehabilitation ranged from three to 48 weeks. There was a moderate risk of bias due to the absence of outcome assessor blinding and intention‐to‐treat analyses and the inadequate reporting of randomisation and allocation procedures in 60% of the studies.

Pulmonary rehabilitation probably improves the six‐minute walk distance (6MWD) with mean difference (MD) of 40.07 metres, 95% confidence interval (CI) 32.70 to 47.44; 585 participants; moderate‐certainty evidence). There may be improvements in peak workload (MD 9.04 watts, 95% CI 6.07 to 12.0; 159 participants; low‐certainty evidence), peak oxygen consumption (MD 1.28 mL/kg/minute, 95% CI 0.51 to 2.05; 94 participants; low‐certainty evidence) and maximum ventilation (MD 7.21 L/minute, 95% CI 4.10 to 10.32; 94 participants; low‐certainty evidence). In the subgroup of participants with IPF, there were comparable improvements in 6MWD (MD 37.25 metres, 95% CI 26.16 to 48.33; 278 participants; moderate‐certainty evidence), peak workload (MD 9.94 watts, 95% CI 6.39 to 13.49; low‐certainty evidence), VO₂ (oxygen uptake) peak (MD 1.45 mL/kg/minute, 95% CI 0.51 to 2.40; low‐certainty evidence) and maximum ventilation (MD 9.80 L/minute, 95% CI 6.06 to 13.53; 62 participants; low‐certainty evidence). The effect of pulmonary rehabilitation on maximum heart rate was uncertain.

Pulmonary rehabilitation may reduce dyspnoea in participants with ILD (standardised mean difference (SMD) –0.36, 95% CI –0.58 to –0.14; 348 participants; low‐certainty evidence) and in the IPF subgroup (SMD –0.41, 95% CI –0.74 to –0.09; 155 participants; low‐certainty evidence). Pulmonary rehabilitation probably improves health‐related quality of life: there were improvements in all four domains of the Chronic Respiratory Disease Questionnaire (CRQ) and the St George's Respiratory Questionnaire (SGRQ) for participants with ILD and for the subgroup of people with IPF. The improvement in SGRQ Total score was –9.29 for participants with ILD (95% CI –11.06 to –7.52; 478 participants; moderate‐certainty evidence) and –7.91 for participants with IPF (95% CI –10.55 to –5.26; 194 participants; moderate‐certainty evidence). Five studies reported longer‐term outcomes, with improvements in exercise capacity, dyspnoea and health‐related quality of life still evident six to 12 months following the intervention period (6MWD: MD 32.43, 95% CI 15.58 to 49.28; 297 participants; moderate‐certainty evidence; dyspnoea: MD –0.29, 95% CI –0.49 to –0.10; 335 participants; SGRQ Total score: MD –4.93, 95% CI –7.81 to –2.06; 240 participants; low‐certainty evidence). In the subgroup of participants with IPF, there were improvements at six to 12 months following the intervention for dyspnoea and SGRQ Impact score. The effect of pulmonary rehabilitation on survival at long‐term follow‐up is uncertain. There were insufficient data to allow examination of the impact of disease severity or exercise training modality.

Ten studies provided information on adverse events; however, there were no adverse events reported during rehabilitation. Four studies reported the death of one pulmonary rehabilitation participant; however, all four studies indicated this death was unrelated to the intervention received.

Authors' conclusions

Pulmonary rehabilitation can be performed safely in people with ILD. Pulmonary rehabilitation probably improves functional exercise capacity, dyspnoea and quality of life in the short term, with benefits also probable in IPF. Improvements in functional exercise capacity, dyspnoea and quality of life were sustained longer term. Dyspnoea and quality of life may be sustained in people with IPF. The certainty of evidence was low to moderate, due to inadequate reporting of methods, the lack of outcome assessment blinding and heterogeneity in some results. Further well‐designed randomised trials are needed to determine the optimal exercise prescription, and to investigate ways to promote longer‐lasting improvements, particularly for people with IPF.

Plain language summary

Pulmonary rehabilitation for interstitial lung disease

Review question: we reviewed available evidence on the effects of pulmonary rehabilitation on exercise capacity, shortness of breath and quality of life in people with interstitial lung disease (ILD).

Background: people with ILD (a condition where the lungs become scarred and breathing becomes increasingly difficult) often have reduced exercise capacity and shortness of breath during exercise. Pulmonary rehabilitation can improve well‐being in people with other chronic lung diseases, but there is less information regarding the effectiveness of pulmonary rehabilitation in ILD. We wanted to discover whether pulmonary rehabilitation provided advantages over no pulmonary rehabilitation for people with ILD and whether it can be performed safely. We also looked at whether people with idiopathic pulmonary fibrosis (IPF), a type of ILD that can progress rapidly, could benefit from pulmonary rehabilitation.

Studies we found: we included 21 studies involving 909 people with ILD. We combined and compared the results of 16 studies (356 participants received pulmonary rehabilitation and 319 participants did not receive pulmonary rehabilitation). Nine studies included only people with IPF, three studies included only those with sarcoidosis (small patches of red and swollen tissue within the lungs), two studies included only those with occupational dust‐related ILD, and the other eight studies included people with a variety of ILDs. The average age of participants ranged from 36 to 72 years. All pulmonary rehabilitation programmes consisted of endurance training (stepping, walking, cycling or a combination of modalities) and some also included the addition of strength‐training exercises. Most pulmonary rehabilitation programmes lasted for eight to 12 weeks, with participants attending two or three sessions per week.

Key results: immediately following pulmonary rehabilitation, participants could walk further than those who had not undertaken pulmonary rehabilitation (on average, 40 metres further in six minutes). Participants also improved their maximum exercise capacity and reported less shortness of breath and improved quality of life. People with IPF experienced comparable improvements in exercise capacity, shortness of breath and quality of life following pulmonary rehabilitation. Six to 12 months following pulmonary rehabilitation, participants could still walk further than those who had not undertaken pulmonary rehabilitation (on average 37 metres further in six minutes) and they sustained some improvements in shortness of breath and quality of life. In people with IPF, it is less certain whether improvements are sustained six to 12 months following pulmonary rehabilitation. There were no studies that described any side effects of pulmonary rehabilitation.

Quality of the evidence: the quality of evidence was generally low to moderate. This was mainly due to inadequate reporting of methods, assessors knowing which treatment had been given and the variability in some results.

Conclusion: pulmonary rehabilitation probably improves exercise capacity, symptoms and quality of life, and can be performed safely in people with ILD, including those with IPF. These results support the inclusion of pulmonary rehabilitation as part of the management for people with ILD. Future studies should explore ways to promote longer‐lasting improvements following exercise training, in particular for those with IPF and which exercise‐training strategy leads to the greatest benefit.

This review is current to June 2020.

Summary of findings

Summary of findings 1. Pulmonary rehabilitation compared to no pulmonary rehabilitation for interstitial lung disease.

Pulmonary rehabilitation compared to no pulmonary rehabilitation for interstitial lung disease
Patient or population: interstitial lung disease Setting: pulmonary rehabilitation centres Intervention: pulmonary rehabilitation Comparison: no pulmonary rehabilitation
Outcomes	Anticipated absolute effects* (95% CI)		Relative effect (95% CI)	№ of participants (studies)	Certainty of the evidence (GRADE)	Comments
	Risk with no pulmonary rehabilitation	Risk with pulmonary rehabilitation
Change in 6MWD assessed with: 6MWT Follow‐up: range 3–48 weeks	The mean change in 6MWD ranged from –35 metres to 26 metres	MD 40.07 metres higher (32.70 higher to 47.44 higher)	—	585 (13 RCTs)	⊕⊕⊕⊝ Moderatea	Sensitivity analysis from studies at lower risk of bias was similar (MD 41.22 metres, 95% CI 26.80 to 55.64; 5 RCTs, 288 participants; I² = 35%).
Change in 6MWD at long‐term follow‐up assessed with: 6MWT Follow‐up: range 6–11 months	The mean change in 6MWD at long‐term follow‐up ranged from –49 metres to –6 metres	MD 32.43 metres higher (15.58 higher to 49.28 higher)	—	321 (6 RCTs)	⊕⊕⊕⊝ Moderateb	—
Change in peak work capacity assessed with: cardiopulmonary exercise test Follow‐up: range 8 weeks to 6 months	The mean change in peak work capacity ranged from –10 watts to 0.6 watts	MD 9.04 watts higher (6.07 higher to 12.0 higher)	—	159 (4 RCTs)	⊕⊕⊝⊝ Lowc,d	—
Change in dyspnoea score Follow‐up: range 8 weeks to 6 months	The mean change in dyspnoea score ranged from –0.2 to 0.4	SMD 0.36 SD lower (0.58 lower to 0.14 lower)	—	348 (7 RCTs)	⊕⊕⊝⊝ Lowe,f	Lower value post intervention is favourable, indicating improvement in dyspnoea. Sensitivity analysis from studies at lower risk of bias was similar (SMD –0.28, 95% CI –0.51 to –0.04; 5 RCTs, 288 participants; I² = 70%). SMD of –0.36 corresponds to MD of –0.32 points when re‐expressed on the modified Medical Research Dyspnoea Scale (0–4, 5‐point score, 0 indicates no dyspnoea).
Change in quality of life assessed with: SGRQ Total score Follow‐up: range 8–48 weeks	The mean change in quality of life ranged from –7 to 6 points	MD 9.29 points lower (11.06 lower to 7.52 lower)	—	478 (11 RCTs)	⊕⊕⊕⊝ Moderatea	Lower value post intervention is favourable, indicating improvement in quality of life. Sensitivity analysis from studies at lower risk of bias was similar (MD –8.13, 95% CI –11.24 to –5.02; 4 RCTs, 231 participants; I² = 21%).
Change in quality of life at long‐term assessed with: SGRQ Total score Follow‐up: 6–11 months	The mean change in quality of life at long‐term follow‐up ranged from –1 to 5 points	MD 4.93 points lower (7.81 lower to 2.06 lower)	—	240 (4 RCTs)	⊕⊕⊝⊝ Lowc,f	Lower value post intervention is favourable, indicating improvement in quality of life.
Long‐term survival (incidence of mortality) Follow‐up: range 6–11 months	Study population		OR 0.40 (0.14 to 1.12)	291 (4 RCTs)	⊕⊕⊝⊝ Lowc,g	Lower OR represents improved survival at long‐term follow‐up.
	85 per 1000	36 per 1000 (13 to 94)
*The risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). 6MWD: 6‐minute walk distance; 6MWT: 6‐minute walk test; CI: confidence interval; MD: mean difference; OR: odds ratio; RCT: randomised controlled trial; SD: standard deviation; SGRQ: St George's Respiratory Questionnaire; SMD: standardised mean difference.
GRADE Working Group grades of evidence High certainty: we are very confident that the true effect lies close to that of the estimate of the effect. Moderate certainty: 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 certainty: our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect. Very low certainty: we have very little confidence in the effect estimate: the true effect is likely to be substantially different from the estimate of effect.

^aDowngraded one level for detection bias (nine to 11 studies), attrition bias (five to eight studies) and selection bias (seven studies).
^bDowngraded one level for detection bias (two studies) and attrition bias (one study).
^cDowngraded one level for detection bias (two studies), attrition bias (one study) and small numbers of studies/participants in meta‐analysis.
^dDowngraded one level for inconsistency – high statistical heterogeneity detected (I² > 75%).
^eDowngraded one level for detection detection performance bias (four studies) and attrition bias (two studies).
^fDowngraded one level for inconsistency – substantial statistical heterogeneity detected (I² = 50% to 75%).
^gDowngraded one level for imprecision (wide CIs).

Summary of findings 2. Pulmonary rehabilitation compared to no pulmonary rehabilitation for idiopathic pulmonary fibrosis.

Pulmonary rehabilitation compared to no pulmonary rehabilitation for idiopathic pulmonary fibrosis
Patient or population: idiopathic pulmonary fibrosis Setting: pulmonary rehabilitation centres Intervention: pulmonary rehabilitation Comparison: no pulmonary rehabilitation
Outcomes	Anticipated absolute effects* (95% CI)		Relative effect (95% CI)	№ of participants (studies)	Certainty of the evidence (GRADE)	Comments
	Risk with no pulmonary rehabilitation	Risk with pulmonary rehabilitation
Change in 6MWD assessed with: 6MWT Follow‐up: range 3–12 weeks	The mean change in 6MWD ranged from –35 metres to 26 metres	MD 37.25 metres higher (26.16 higher to 48.33 higher)	—	278 (8 RCTs)	⊕⊕⊕⊝ Moderatea	—
Change in 6MWD at long‐term follow‐up assessed with: 6MWT Follow‐up: range 6–11 months	The mean change in 6MWD at long‐term follow‐up ranged from –49 metres to 4 metres	MD 1.64 metres higher (24.89 lower to 28.17 higher)	—	123 (3 RCTs)	⊕⊕⊝⊝ Lowb,c	—
Change in peak work capacity assessed with: cardiopulmonary exercise test Follow‐up: range 8–12 weeks	The mean change in peak work capacity ranged from –7 watts to –0.8 watts	MD 9.94 watts higher (6.39 higher to 13.49 higher)	—	62 (2 RCTs)	⊕⊕⊝⊝ Lowb,d,e	—
Change in dyspnoea score Follow‐up: range 8–12 weeks	The mean change in dyspnoea score ranged from –0.06 to 0.4	SMD 0.41 lower (0.74 lower to 0.09 lower)	—	155 (4 RCTs)	⊕⊕⊝⊝ Lowb,f	Lower value post intervention is favourable, indicating improvement in dyspnoea. SMD of –0.41 corresponds to MD of –0.37 points when re‐expressed on the modified Medical Research Dyspnoea Scale (0–4, 5‐point score, 0 indicates no dyspnoea).
Change in quality of life assessed with: SGRQ Total Follow‐up: range 8 weeks to 6 months	The mean change in quality of life ranged from –3 to 3 points	MD 7.91 points lower (10.55 lower to 5.26 lower)	—	194 (6 RCTs)	⊕⊕⊕⊝ Moderatea	Lower value post intervention is favourable, indicating improvement in quality of life.
Change in quality of life at long‐term assessed with: SGRQ Total score Follow‐up: range 6–11 months	The mean change in quality of life at long‐term follow‐up ranged from 1 to 4 points	MD 3.45 points lower (7.43 lower to 0.52 higher)	—	89 (2 RCTs)	⊕⊕⊝⊝ Lowb,e	Lower value post intervention is favourable, indicating improvement in quality of life.
Long‐term survival (incidence of mortality) Follow‐up: range 6–11 months	Study population		OR 0.32 (0.08 to 1.19)	127 (3 RCTs)	⊕⊕⊝⊝ Lowb,c	Lower OR represents improved survival at long‐term follow‐up.
	133 per 1000	47 per 1000 (12 to 155)
*The risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). 6MWD: 6‐minute walk distance; 6MWT: 6‐minute walk test; CI: confidence interval; OR: odds ratio; RCT: randomised controlled trial; SGRQ: St George's Respiratory Questionnaire.
GRADE Working Group grades of evidence High certainty: we are very confident that the true effect lies close to that of the estimate of the effect. Moderate certainty: 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 certainty: our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect. Very low certainty: we have very little confidence in the effect estimate: the true effect is likely to be substantially different from the estimate of effect.

^aDowngraded one level for detection bias (four or five studies), attrition bias (three or four studies) and selection bias (five studies)
^bDowngraded one level for detection bias (one or two studies), attrition bias (one study) and meta‐analysis was limited to 3‐4 studies
^cDowngraded one level for imprecision (wide CIs)
^dDowngraded one level for inconsistency – high statistical heterogeneity detected (I² > 75%)
^eDowngraded one level for imprecision ‐ meta‐analysis was limited to 2 studies
^fDowngraded one level for inconsistency – substantial statistical heterogeneity detected (I² = 50% to 75%)

Background

Description of the condition

Interstitial lung disease (ILD) is a highly disabling group of conditions including idiopathic pulmonary fibrosis (IPF), acute and chronic interstitial pneumonias, hypersensitivity pneumonitis, asbestosis, silicosis, sarcoidosis and connective tissue disease‐related disorders such as rheumatoid arthritis and scleroderma. People with ILD frequently experience breathlessness on exertion, which limits their ability to undertake daily activities. Patients report low levels of physical functioning and vitality, and high levels of dyspnoea and fatigue. Those with the greatest exercise limitations have the worst quality of life (Chang 1999). Treatment options for people with ILD are generally limited. Two antifibrotic therapies, pirfenidone and nintedanib, slow disease progression and potentially improve survival in IPF (King 2014; Richeldi 2014), and may also be beneficial for people with other types of progressive fibrosing ILD (Flaherty 2019). In addition, there is limited evidence to suggest these treatments can provide convincing benefits for exercise tolerance, quality of life or symptoms (Graney 2018; Kreuter 2020; Nathan 2019).

The mechanisms of reduced exercise capacity in ILD are multi‐factorial. Impaired gas exchange occurs as a result of destruction of the pulmonary capillary bed, resulting in ventilation‐perfusion mismatch and oxygen diffusion limitations (Agusti 1991). Circulatory limitation results from pulmonary capillary destruction and pulmonary vasoconstriction and leads to pulmonary hypertension and cardiac dysfunction in some patients (Hansen 1996). Ventilatory limitations to exercise may also occur, although these are not thought to be a major contributor in most patients (Harris‐Eze 1996). Peripheral muscle dysfunction may play a significant role in limiting exercise capacity as a result of physical deconditioning (Markovitz 1998). Patients who experience dyspnoea and fatigue with functional activity commonly reduce their activity levels, leading to a vicious cycle of worsening exercise capacity and increasing symptoms. In addition, treatments for ILD such as corticosteroids and immunosuppressive therapy may lead to drug‐induced myopathy.

Description of the intervention

Pulmonary rehabilitation includes patient assessment, regular participation in an exercise‐training programme, education and behavioural change (Spruit 2013). Exercise training is a fundamental component of pulmonary rehabilitation (Spruit 2013), and includes aerobic training as a core component, often comprising of walking, cycling or a combination of both. Resistance training is an important additional component for optimising improvements in muscle strength (Bolton 2013). Pulmonary rehabilitation can occur in several different settings such as hospital outpatient departments and community health centres, typically the most widely available of settings, inpatient stays or a home‐based environment. The role of pulmonary rehabilitation is well established in people with other chronic lung diseases such as chronic obstructive pulmonary disease (COPD), for whom it improves exercise performance and reduces symptoms (Spruit 2013). Individuals with ILD often present with similar symptoms to those seen in COPD, despite differences in underlying pathophysiology, such as dyspnoea, fatigue, reduced exercise tolerance and poor quality of life (Holland 2013). Given these similarities, and that many of these issues are modifiable in COPD, several authors have postulated that similar effects of pulmonary rehabilitation may be seen in people with ILD.

How the intervention might work

The mechanism by which pulmonary rehabilitation might improve outcomes in people with ILD has not been established. In people with other respiratory diseases, pulmonary rehabilitation may improve aerobic capacity and improves peripheral muscle performance (Spruit 2013). Effects on these outcomes in ILD are less established. Despite this, guidelines for pulmonary rehabilitation have advocated its use in 'individuals with chronic respiratory disorders other than COPD' including ILD as 'there is now more robust evidence to support inclusion of some of these patient groups in pulmonary rehabilitation programs' (Bolton 2013; Spruit 2013). However, it has been suggested that the benefits of pulmonary rehabilitation in ILD are smaller than those generally seen in COPD, that it may not be suitable for some patients due to variability across the disease spectrum and that its ongoing effects are not sustained beyond six months (Bolton 2013; Spruit 2013). Guidelines for clinical management of both ILD (Bradley 2008) and IPF (ATS 2011) indicate that more information is needed on the benefits of pulmonary rehabilitation for these patients. The greater prevalence of exercise‐induced hypoxia, pulmonary hypertension and arrhythmia compared with other chronic lung diseases in this patient population raises the possibility that response to exercise rehabilitation may also differ (ATS 2011).

Why it is important to do this review

The review authors undertook the original version of this Cochrane Review to establish the safety and efficacy of pulmonary rehabilitation in adults with ILD, and to determine the effects of pulmonary rehabilitation on exercise capacity, symptoms, quality of life and survival in this patient group. The original review and the second update in 2014 concluded that pulmonary rehabilitation resulted in significant improvements in exercise capacity, quality of life and symptoms. However, the number of RCTs was small (five to nine) and they were associated with methodological bias and the longer‐term benefit of pulmonary rehabilitation remained unclear. Poor exercise tolerance, dyspnoea and fatigue remain a major burden for people with ILD and interventions such as pulmonary rehabilitation can positively improve these aspects. This has led to a dramatic rise in the number of studies investigating the benefits of pulmonary rehabilitation in ILD, with some following strong methodological design. In addition, there has been increasing international acceptance that pulmonary rehabilitation can positively impact people with ILD, with the inclusion of ILD in pulmonary rehabilitation programmes recommended in international guidelines (Bolton 2013). This review aimed to provide more conclusive evidence of the benefit of pulmonary rehabilitation in ILD, as well as clarifying the longer‐term benefit and strengthening the rationale for the inclusion of pulmonary rehabilitation in standard care for people with ILD.

Objectives

To determine whether pulmonary rehabilitation in people with ILD has beneficial effects on exercise capacity, symptoms, quality of life and survival compared with no pulmonary rehabilitation in people with ILD.

To assess the safety of pulmonary rehabilitation in people with ILD.

Methods

Criteria for considering studies for this review

Types of studies

We included randomised controlled trials (RCTs) and quasi‐RCTs in which a prescribed regimen of pulmonary rehabilitation was compared with no pulmonary rehabilitation or with other therapy in study participants with ILD. We considered single‐blind and open studies for inclusion.

Types of participants

People with ILD of any origin, diagnosed according to investigator definitions. There were no exclusions based on age, gender or physiological status.

Types of interventions

We considered any type of prescribed exercise training, supervised or unsupervised, provided with or without education. We recorded, when possible, the precise nature of the training (intensity, frequency, duration and whether supplemental oxygen was applied). Trials in which pulmonary rehabilitation was combined with another intervention (e.g. pharmacological therapy) were eligible for inclusion.

Comparisons to be examined included the following.

• Pulmonary rehabilitation versus no pulmonary rehabilitation.

• Pulmonary rehabilitation versus another intervention.

• Pulmonary rehabilitation combined with another intervention versus no pulmonary rehabilitation.

Types of outcome measures

Primary outcomes

• Functional or maximal exercise capacity, measured during formal exercise tests (maximal oxygen uptake (VO₂ max), peak oxygen uptake (VO₂ peak), peak work capacity (peak watts), maximal ventilation (Ve max), maximum heart rate (HRmax)) or field exercise tests (increase in distance walked).

Secondary outcomes

• Dyspnoea: all measures of dyspnoea used.

• Quality of life: measured by generic or disease‐specific quality‐of‐life instruments. All quality‐of‐life instruments used.

• Adverse effects: adverse cardiovascular events during exercise training, musculoskeletal injuries and deaths.

• Survival.

Search methods for identification of studies

Electronic searches

We searched the Cochrane Airways Trials Register via the Cochrane Register of Studies, Cochrane Central Register of Controlled Trials (CENTRAL 2020, Issue 4) via the Cochrane Register of Studies, MEDLINE (OvidSP), Embase (OvidSP), the Cumulative Index to Nursing and Allied Health Literature (CINAHL) (EBSCO) and the Physiotherapy Evidence Database (PEDro) from inception to 16 April 2020. There were no language restrictions. The previously published version included searches up to June 2014. The search period for this update was June 2014 to April 2020.

The full database search strategies are listed in the appendices (Appendix 1; Appendix 2; Appendix 3; Appendix 4; Appendix 5).

Searching other resources

We handsearched the reference lists of relevant studies and related review papers for qualifying studies. We reviewed clinical trial registries (ClinicalTrials.gov: www.clinicaltrials.gov and the World Health Organization (WHO) trials portal: www.who.int/ictrp/en) to search for relevant planned, ongoing and unpublished trials. We reviewed annual conference abstracts for the American Thoracic Society (ATS), the European Respiratory Society (ERS), the Asian Pacific Society of Respirology (APSR) and the Thoracic Society of Australia and New Zealand (TSANZ) for relevant studies. In addition, we contacted the authors of RCTs to ask for information on other published and unpublished studies.

Data collection and analysis

Selection of studies

Two review authors (LD and AM) independently coded for relevant studies identified in the literature searches by examining titles, abstracts and keyword fields as follows.

• Include: study categorically met all review criteria.

• Unclear: study appeared to meet some review criteria, but available information was insufficient for review authors to categorically determine relevance.

• Exclude: study did not categorically meet all review criteria.

Two review authors (LD and AM) used full‐text copies of study papers categorised as 'include' and 'unclear' to decide on study inclusion. We resolved disagreements by consensus or involving a third review author (AH). We kept a full record of decisions, and calculated simple agreement and kappa statistics.

Data extraction and management

Two review authors (LD and AM) independently extracted data using a prepared checklist; one review author (LD) entered data into Review Manager 5 with random checks on accuracy. We resolved disagreements by consensus. Data included characteristics of included studies (methods, participants, interventions, outcomes) and results of included studies. We contacted authors of included studies to request details of missing data where applicable.

Assessment of risk of bias in included studies

Two review authors (LD and AM) independently assessed the risk of bias for all included studies. We assessed the risk of bias following the criteria provided by the Cochrane Handbook of Systematic Reviews of Interventions (Higgins 2020). The review authors assessed the internal validity of included studies using a component approach (including sequence generation for randomisation, allocation concealment, blinding of participants and personnel, incomplete outcome data, selective outcome reporting and other potential sources of bias). We judged the risk of bias for each study as 'low', 'high' or 'unclear' risk and resolved disagreements by consensus. We wrote to study authors to seek clarification when information was inadequate to judge the risk of bias.

Measures of treatment effect

For continuous variables, we recorded mean change from baseline or mean post intervention values and standard deviation (SD) for each group. We calculated SDs when 95% confidence intervals (CIs) or standard errors were reported using guidance in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2020). When SDs were missing and we were unable to obtain the results from study authors, we used a mean value for the SD of a similar study that reported the outcome to calculate the required SD using guidance in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2020). When measures of improvement had opposite directions of effect on different scales (e.g. dyspnoea), we recorded all improvements as negative values, and all deteriorations as positive values. We calculated mean differences (MDs) for outcomes measured with the same metrics or standardised mean differences (SMDs) for outcomes measured with different metrics with 95% CIs using Review Manager 5 and RevmanWeb. To facilitate interpretation of SMDs, we re‐expressed SMD estimates as MDs on more common measurement scales as described in Chapter 15 of the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2020). For binary outcome measures, we recorded the number of participants with each outcome event, by allocated treated group, to allow intention‐to‐treat analysis. We calculated odds ratios (ORs) with 95% CIs for each study.

Unit of analysis issues

The search identified no cluster RCTs that met the inclusion criteria for this systematic review. If future versions include cluster RCTs that have not been adjusted for clustering in their analysis, we will calculate the effective sample size for these studies based on the methods described in the Cochrane Handbook for Systematic Review of Interventions (Higgins 2020).

Dealing with missing data

Where possible, we contacted the trial authors if data were missing from included studies. When SDs were missing and we were unable to obtain the results from study authors, we used the SD of a similar study that reported the outcome to impute a SD using Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2020). We excluded studies that did not report mean change scores or enough data to calculate mean change scores from meta‐analyses. We considered studies that did not use intention‐to‐treat analysis and omitted data from participants due to withdrawal or incompleteness to have a high risk of bias.

Assessment of heterogeneity

We assessed statistical heterogeneity in each meta‐analysis using the Chi² test and the I² statistic (Higgins 2020). We used a P value of 0.10 to determine statistical significance. We regarded heterogeneity as low when the I² statistic was less than 30%, moderate when the I² statistic was 30% to 50%, as substantial when the I² statistic was 50% to 75% and of high statistical heterogeneity if the I² statistic was greater than 75%.

Assessment of reporting biases

We assessed relevant causes of bias on the analysis including publication bias, outcome reporting bias and methodological quality. When meta‐analyses included a minimum of 10 studies, we created funnel plots to investigate reporting biases (such as publication bias).

Data synthesis

We performed a pooled quantitative analysis when trials were clinically homogeneous. We used a fixed‐effect model or a random‐effects model depending on assessment of heterogeneity.

Subgroup analysis and investigation of heterogeneity

We conducted three subgroup analyses specified a priori to explore possible sources of heterogeneity.

• Type of ILD: IPF versus other: as a result of the progressive nature of IPF, pulmonary rehabilitation could be less effective in this form of ILD.

• Severity of lung disease: people with more advanced disease may be less able to participate in pulmonary rehabilitation. Participants were considered to have severe disease if diffusing capacity for carbon monoxide (transfer factor for carbon monoxide (TLCO)) was less than 45% predicted (Flaherty 2001). In addition, participants who desaturated during exercise testing (oxygen saturation (SpO₂) 88% or less) were compared with those who did not desaturate.

• Type of exercise: aerobic exercise training programmes may be more effective in improving symptoms and functional exercise tolerance than resistance training programmes. However, data were insufficient to allow review authors to perform this subgroup analysis.

Sensitivity analysis

We performed sensitivity analyses according to trial quality by repeating our analysis among only those studies judged to be of 'high quality’. For the purposes of this review, 'high‐quality’ trials were defined as trials with low risk of bias due to allocation concealment, and intention‐to‐treat analysis. We performed the sensitivity analyses for the primary outcome of functional or maximal exercise capacity (six‐minute walk distance (6MWD)) and the secondary outcomes of dyspnoea and health‐related quality of life.

Summary of findings and assessment of the certainty of the evidence

We used GRADE to assess the evidence for the primary outcome of functional or maximal exercise capacity (peak oxygen update, peak work rate) plus the secondary outcomes of dyspnoea, quality of life (St George's Respiratory Questionnaire (SGRQ)) and survival. We performed these analyses and presented the results in a 'Summary of findings' table for ILD (Table 1) and IPF (Table 2) generated using GRADEpro GDT software.

Results

Description of studies

Details are available in the Characteristics of included studies, Characteristics of excluded studies, Characteristics of studies awaiting classification, and Characteristics of ongoing studies tables.

Results of the search

See Figure 1 for the study flow diagram.

1.

Study flow diagram for 2014–2020 literature searches. HRQoL: health‐related quality of life.

The original version of the review identified 4783 records from the initial search of databases (Holland 2008). From the studies on this list, the review authors retrieved 15 full‐text articles for closer inspection. There were no additional studies identified upon handsearching of reference lists or contact with study authors. Review authors achieved agreement on 13/15 full‐text articles (87%) with kappa = 0.74, indicating substantial agreement. They resolved disagreement by consensus. Five articles met the inclusion criteria for the original review (Baradzina 2005; Holland 2008; Mejia 2000; Nishiyama 2008; Wewel 2005).

The 2014 updated search of databases returned 1901 potential studies (Dowman 2014). The review authors retrieved eight full‐text articles from this list for closer inspection. They identified six additional studies upon handsearching of reference lists and review of international clinical trial registries and annual international respiratory conference abstracts. The review authors achieved agreement on 13/14 full‐text articles (92%) with kappa = 0.81, indicating substantial agreement. They resolved disagreements by consensus. We included four additional studies in the review update (Jackson 2014; Menon 2011; Perez Bogerd 2018 (identified as Perez Bogerd 2011 in the 2014 update of this review as only preliminary results were available); Vainshelboim 2014). One article was awaiting classification and, therefore, was not included in the analysis (Dale 2014). Nine articles in total were included in the 2014 review update.

The search for the most recent and current update covered the period from June 2014 to April 2020. We identified 4453 references through the electronic database search and four additional studies upon handsearching of reference lists and review of international clinical trial registries and annual international respiratory conference abstracts. We retrieved 22 studies (36 references) from electronic databases for full‐text assessment. We achieved agreement between review authors on 19 of the full‐text articles (86%) with kappa = 0.70, indicating good agreement. We resolved disagreement by consensus or by involving a third review author. We included 12 additional studies (Dale 2014; De Las Heras 2019; Dowman 2017; Gaunaurd 2014; He 2016; Jarosch 2020; Ku 2017; Lanza 2019; Naz 2018; Shen 2016; Wallaert 2020; Xiao 2019). Figure 1 shows a study flow diagram.

Included studies

Twenty‐one studies met the inclusion criteria for this review; all were parallel RCTs. We included four studies in the previous version of the review as preliminary data (Dale 2014; Gaunaurd 2014; Perez Bogerd 2018; Vainshelboim 2014), but full published data for all four studies for for Vainshelboim 2014 were available for this update. Seven studies were in abstract form only (Baradzina 2005; De Las Heras 2019; Lanza 2019; Mejia 2000; Menon 2011; Shen 2016; Wewel 2005). Sample sizes ranged from 18 to 142 participants. Full details can be found in the Characteristics of included studies table.

Participants

Most studies included participants with a variety of ILDs (Dowman 2017; Holland 2008; Ku 2017; Mejia 2000; Menon 2011; Perez Bogerd 2018; Wewel 2005). One of these was stratified according to the three subgroups of IPF, dust‐related ILD and connective tissue disease‐related ILD (Dowman 2017) and one was stratified for IPF (Holland 2008). Nine studies included only participants with IPF (De Las Heras 2019; Gaunaurd 2014; He 2016; Jackson 2014; Jarosch 2020; Lanza 2019; Nishiyama 2008; Shen 2016; Vainshelboim 2014), whilst three studies included only participants with sarcoidosis (Baradzina 2005; Naz 2018; Wallaert 2020), and two studies included only participants with occupational dust‐related ILD (pneumoconiosis) (Dale 2014; Xiao 2019). All participants were adults with mean age ranging from 36 to 72 years. Three studies did not report mean age (Lanza 2019; Menon 2011; Shen 2016).

Interventions

All studies compared pulmonary rehabilitation versus no pulmonary rehabilitation or a sham training control group. Eighteen studies examined pulmonary rehabilitation programmes conducted in an outpatient setting (Baradzina 2005; Dale 2014; Dowman 2017; Gaunaurd 2014; He 2016; Holland 2008; Jackson 2014; Ku 2017; Lanza 2019; Mejia 2000; Menon 2011; Naz 2018; Nishiyama 2008; Perez Bogerd 2018; Shen 2016; Vainshelboim 2014; Wallaert 2020; Xiao 2019), one study evaluated pulmonary rehabilitation in an inpatient setting (Jarosch 2020), one study evaluated a home‐based pulmonary rehabilitation programme (Wewel 2005), and another study evaluated a tele‐rehabilitation model of pulmonary rehabilitation (De Las Heras 2019) (Table 3). The length of pulmonary rehabilitation programmes varied from five to 48 weeks for outpatient rehabilitation with 15 studies being eight to 12 weeks and the remaining three being five weeks, six months and 48 weeks. The length of pulmonary rehabilitation programmes were three weeks for inpatient rehabilitation, six months for home‐based rehabilitation and three months for tele‐rehabilitation (Table 3).

1. Study design.

Study	Follow‐up	Duration (weeks)	Sessions (per week)	Setting	Programme type
Baradzina 2005	5 weeks	5	5	Outpatient	Exercise + other
Dale 2014	8, 26 weeks	8	2	Outpatient	Exercise
De Las Heras 2019	12 weeks	12	5–7	Tele‐rehabilitation	Exercise
Dowman 2017	8 weeks, 6 months	8	2	Outpatient	Exercise + other
Gaunaurd 2014	12 weeks, 3 months	12	2	Outpatient	Exercise + other
He 2016	12 weeks	12	3–5	Outpatient	Exercise
Holland 2008	8, 26 weeks	8	2	Outpatient	Exercise
Jackson 2014	12 weeks, 3 months	12	2	Outpatient	Exercise + other
Jarosch 2020	3 weeks, 3 months	3	5–6	Inpatient	Exercise + other
Ku 2017	8 weeks	8	2	Outpatient	Exercise + other
Lanza 2019	12 weeks	12	2	Outpatient	Exercise
Mejia 2000	12 weeks	12	3	Outpatient	Exercise
Menon 2011	8 weeks	8	—	Outpatient	Exercise
Naz 2018	12 weeks	12	2	Outpatient	Exercise
Nishiyama 2008	9 weeks	9	2	Outpatient	Exercise
Perez Bogerd 2018	3, 6, 12 months	26	2–3	Outpatient	Exercise + other
Shen 2016	12 weeks	12	3	Outpatient	Exercise
Vainshelboim 2014	12 weeks	12	2	Outpatient	Exercise
Wallaert 2020	8 weeks	8	3	Outpatient	Exercise + other
Wewel 2005	6 months	26	7	Home	Exercise
Xiao 2019	48 weeks	48	4	Outpatient/home	Exercise + other

Five studies examined the effects of aerobic training (Baradzina 2005; Dale 2014; He 2016; Mejia 2000; Wewel 2005), 13 studies used a combination of aerobic and resistance training (De Las Heras 2019; Dowman 2017; Gaunaurd 2014; Holland 2008; Jackson 2014; Jarosch 2020; Ku 2017; Naz 2018; Nishiyama 2008; Perez Bogerd 2018; Vainshelboim 2014; Wallaert 2020; Xiao 2019), and the remaining studies did not specify the exercise modality used (Lanza 2019; Menon 2011; Shen 2016). No study evaluated resistance training alone; therefore, subgroup analyses for type of exercise were not possible. Nine studies comprised exercise training alone (Dale 2014; De Las Heras 2019; He 2016; Holland 2008; Mejia 2000; Naz 2018; Vainshelboim 2014; Wewel 2005; Xiao 2019), whereas nine studies added interventions to exercise training that were not offered to the control group (Table 3); these included educational lectures (Baradzina 2005; Dowman 2017; Gaunaurd 2014; Jackson 2014; Jarosch 2020; Ku 2017; Nishiyama 2008; Perez Bogerd 2018; Wallaert 2020), nutritional advice (Baradzina 2005; Gaunaurd 2014; Jarosch 2020; Ku 2017; Perez Bogerd 2018), stress management (Baradzina 2005), physiotherapy (Baradzina 2005), occupational therapy (Perez Bogerd 2018), and psychosocial support (Gaunaurd 2014; Jarosch 2020; Ku 2017; Perez Bogerd 2018; Wallaert 2020). Inclusion of additional interventions with exercise training was unclear in three studies (Lanza 2019; Menon 2011; Shen 2016).

Outcomes

All studies used a measure of functional exercise tolerance, most commonly the six‐minute walk test (Dale 2014; De Las Heras 2019; Dowman 2017; Gaunaurd 2014; He 2016; Holland 2008; Jackson 2014; Jarosch 2020; Ku 2017; Lanza 2019; Menon 2011; Naz 2018; Nishiyama 2008; Perez Bogerd 2018; Shen 2016; Vainshelboim 2014; Wewel 2005; Xiao 2019). Five studies also performed a cardiopulmonary exercise test (Dale 2014; Holland 2008; Perez Bogerd 2018; Vainshelboim 2014; Wewel 2005). Eighteen studies assessed quality of life, using the Chronic Respiratory Disease Questionnaire (CRQ) (Dale 2014; Dowman 2017; Holland 2008; Jarosch 2020; Mejia 2000; Perez Bogerd 2018), the SGRQ (Dale 2014; De Las Heras 2019; Ku 2017; Naz 2018; Nishiyama 2008; Perez Bogerd 2018; Shen 2016; Vainshelboim 2014; Wewel 2005; Xiao 2019), the St George's Respiratory Questionnaire IPF version (SGRQ‐I) (Dowman 2017; Gaunaurd 2014; Lanza 2019), the 36‐item Short Form Health Survey (SF‐36) (Jarosch 2020, Naz 2018), or the WHO questionnaire (Baradzina 2005). Twelve studies assessed dyspnoea using the modified Medical Research Council Scale (Dale 2014; Dowman 2017; Holland 2008; Ku 2017; Naz 2018; Perez Bogerd 2018; Vainshelboim 2014; Wallaert 2020), the Baseline Dyspnoea Index (Nishiyama 2008), and an unspecified measure (Baradzina 2005; Wewel 2005).

Excluded studies

Common reasons for exclusion were that eight studies were not RCTs, three studies included participants without lung disease, four studies included mixed disease groups, four studies did not include pulmonary rehabilitation and one study included an intervention‐based control group. Full details of 20 excluded studies, of studies that are ongoing and of studies awaiting classification can be found in the Characteristics of excluded studies, Characteristics of ongoing studies, and Characteristics of studies awaiting classification tables.

Risk of bias in included studies

An overview of the risk of bias for the domains listed below is provided in Figure 2.

2.

Methodological quality summary: review authors' judgements about each methodological quality item for each included study.

Allocation

Sequence generation

All studies reported random allocation to groups. Twelve studies described the methods used for generation of the randomisation sequence and were at low risk of bias (Dale 2014; De Las Heras 2019; Dowman 2017; Gaunaurd 2014; Holland 2008; Jarosch 2020; Ku 2017; Naz 2018; Perez Bogerd 2018; Vainshelboim 2014; Wallaert 2020; Xiao 2019). Nine studies did not specify the method by which the randomisation sequence was generated and these were considered to have unclear risk of bias (Baradzina 2005; He 2016; Jackson 2014; Lanza 2019; Mejia 2000; Menon 2011; Nishiyama 2008; Shen 2016; Wewel 2005).

Allocation concealment

Ten studies reported that the allocation sequence was concealed using sealed envelopes (Dale 2014; Dowman 2017; Holland 2008; Jackson 2014; Jarosch 2020; Naz 2018; Nishiyama 2008; Perez Bogerd 2018; Vainshelboim 2014; Wallaert 2020). Of the remaining studies, four did not specify whether the allocation sequence was concealed (Gaunaurd 2014; He 2016; Ku 2017; Xiao 2019), and seven were available only in abstract form, and did not provide sufficient information to permit assessment of whether the allocation sequence was concealed (Baradzina 2005; De Las Heras 2019; Gaunaurd 2014; He 2016; Ku 2017; Lanza 2019; Mejia 2000; Menon 2011; Shen 2016; Wewel 2005).

Blinding

Performance bias

Blinding of participants or personnel was not possible for the majority of the studies due to the physical nature of the intervention. All included studies except two were at high risk of performance bias. Two studies were at unclear risk since they provided sham exercise training (Mejia 2000), or simple exercise in the form of free movement and hospital‐led gymnastics (Xiao 2019), to the control group. It is possible the participants could be blinded to the intervention received. However, neither study provided specific details to confirm this. No studies reported whether data analysts were blinded to treatment allocation.

Detection bias

Four studies reported use of a blinded assessor for all outcome measures and at low risk of detection bias (Dale 2014; De Las Heras 2019; Dowman 2017; Holland 2008). Seven studies indicated that the assessors were unblinded; these were at high risk of bias (Gaunaurd 2014; Jackson 2014; Jarosch 2020; Ku 2017; Perez Bogerd 2018; Vainshelboim 2014; Wallaert 2020). There were insufficient data to show whether assessors were blinded in the other studies and were considered at unclear risk of bias (Baradzina 2005; He 2016; Lanza 2019; Mejia 2000; Menon 2011; Naz 2018; Nishiyama 2008; Shen 2016; Wewel 2005; Xiao 2019).

Incomplete outcome data

Ten studies reported dropouts and loss to follow‐up (Dale 2014; Dowman 2017; Gaunaurd 2014; Holland 2008; Jackson 2014; Jarosch 2020; Nishiyama 2008; Perez Bogerd 2018; Vainshelboim 2014; Xiao 2019). One of these reported that two participants in the exercise group withdrew before baseline data had been collected and was considered to have low risk of bias (Nishiyama 2008). Six studies reported that participants in the exercise group (one to four) and in the control group (zero to six) did not complete the intervention period (De Las Heras 2019; Gaunaurd 2014; Jackson 2014; Jarosch 2020; Vainshelboim 2014; Xiao 2019). Data from these participants were not included in the analysis in either study (De Las Heras 2019; Gaunaurd 2014; Jackson 2014; Jarosch 2020; Vainshelboim 2014; Xiao 2019), therefore these studies were at high risk of bias.

Eight studies were at low risk of bias for incomplete outcome data. Three studies reported no dropouts with all participants completing the intervention and assessments (He 2016; Ku 2017; Naz 2018). One study reported minimal dropouts (Dale 2014, 5% of people dropped out), three studies reported a moderate number of dropouts (Dowman 2017, 12%; Holland 2008, 20%; Wallaert 2020, 18%) and one study reported a significant number of dropouts (Perez Bogerd 2018, 40%). These five studies performed the data analysis according to the intention‐to‐treat principle (Dale 2014; Dowman 2017; Holland 2008; Perez Bogerd 2018; Wallaert 2020). One study used the last observation carried forward method (Holland 2008), and the other four studies used maximum likelihood estimation to account for missing data in the statistical analysis (Dale 2014; Dowman 2017; Perez Bogerd 2018; Wallaert 2020).

The remaining studies, all of which were published only in abstract form, did not report whether dropouts or losses to follow‐up occurred and were rated as having an unclear risk of bias (Baradzina 2005; Lanza 2019; Mejia 2000; Menon 2011; Shen 2016; Wewel 2005).

Selective reporting

Ten studies were listed on a clinical trial registry (Dale 2014; De Las Heras 2019; Dowman 2017; Gaunaurd 2014; Holland 2008; Jackson 2014; Jarosch 2020; Perez Bogerd 2018; Vainshelboim 2014; Wallaert 2020). Eight studies reported results for all outcomes at all time points (Dale 2014; Dowman 2017; Holland 2008; Jackson 2014; Jarosch 2020; Perez Bogerd 2018; Vainshelboim 2014; Wallaert 2020). Two studies did not report all outcome measures mentioned in the clinical registry (De Las Heras 2019; Gaunaurd 2014). One study was in abstract form, and it is likely that not all data are currently available (De Las Heras 2019). The other study (Gaunaurd 2014) was the publication of the quality‐of‐life data from the original study (Jackson 2014). This study (Gaunaurd 2014) and two others (Ku 2017; Xiao 2019) were judged to have a high risk of reporting bias as they did not report all quality‐of‐life outcomes (Gaunaurd 2014; Ku 2017; Xiao 2019), or mMRC Dyspnoea Score (Ku 2017). Three studies reported the results for all time points for the outcomes detailed in the methods and were considered at low risk of bias (He 2016; Naz 2018; Nishiyama 2008). It was not possible for review authors to determine whether all data were available for the other studies, all of which were provided only in abstract form (Baradzina 2005; Lanza 2019; Mejia 2000; Menon 2011; Shen 2016; Wewel 2005); therefore, these studies were considered at unclear risk of bias.

Other potential sources of bias

Other potential sources of bias may be present due to not all of the data being available in required format of mean change from baseline and SD. Change from baseline SDs were imputed using a correlation coefficient calculated from Nishiyama 2008 for three studies (He 2016; Shen 2016; Xiao 2019). In addition, there were some studies from which we could not obtain additional data, despite contacting the authors to request additional information and there were a number of studies that were provided in abstract form only.

Effects of interventions

See: Table 1; Table 2

Data and analyses tables summarise results of the meta‐analysis for comparison of pulmonary rehabilitation versus no pulmonary rehabilitation. Sixteen studies provided sufficient data for pooling in a meta‐analysis (Dale 2014; De Las Heras 2019; Dowman 2017; Gaunaurd 2014; He 2016; Holland 2008; Jackson 2014; Jarosch 2020; Ku 2017; Naz 2018; Nishiyama 2008; Perez Bogerd 2018; Shen 2016; Vainshelboim 2014; Wallaert 2020; Xiao 2019). Table 1 and Table 2 summarise the certainty of the evidence. For functional exercise capacity, maximal exercise capacity and the CRQ domains (quality of life), positive values reflect improvement. For measures of dyspnoea and the SGRQ domains (quality of life), negative values reflect improvement.

Primary outcomes

Functional exercise capacity

Nineteen studies reported functional exercise capacity. Of these, 17 trials including 802 participants reported that pulmonary rehabilitation resulted in an improvement in functional exercise capacity immediately following the programme (Baradzina 2005; Dale 2014; De Las Heras 2019; Dowman 2017; He 2016; Holland 2008; Jarosch 2020; Ku 2017; Lanza 2019; Mejia 2000; Menon 2011; Naz 2018; Nishiyama 2008; Perez Bogerd 2018; Vainshelboim 2014; Wewel 2005; Xiao 2019). There was no change in 6MWD following pulmonary rehabilitation in one study (Jackson 2014), and it was unclear if there was an improvement in 6MWD in the remaining study (Shen 2016). Thirteen trials provided sufficient data on the six‐minute walk test for meta‐analysis, with 309 participants in the pulmonary rehabilitation group and 276 participants in the control group (Dale 2014; De Las Heras 2019; Dowman 2017; He 2016; Holland 2008; Jackson 2014; Jarosch 2020; Ku 2017; Naz 2018: Nishiyama 2008; Perez Bogerd 2018; Vainshelboim 2014; Xiao 2019). Results of the meta‐analysis are shown in Figure 3 (Analysis 1.1). The common effect (MD) for change in distance walked was 40.07 metres in favour of the pulmonary rehabilitation group (95% CI 32.70 to 47.44). This effect exceeded the minimal important difference (MID) for the 6MWD of 30 metres to 33 metres for people with ILD (Holland 2014). There was also an effect in favour of pulmonary rehabilitation in the subgroup of participants with IPF (8 trials, 151 participants in pulmonary rehabilitation group, 127 participants in control group) with an MD of 37.25 metres (95% CI 26.16 to 48.33). This effect also exceeded the MID for the 6MWD of 29 metres to 34 metres for people with IPF (Holland 2014). Two studies provided sufficient data to show the effects of pulmonary rehabilitation among 84 participants with severe lung disease or in 103 participants who desaturated (Dowman 2017; Holland 2008). The effect of pulmonary rehabilitation on 6MWD was less certain for participants who desaturated (MD 20.21 metres, 95% CI –2.62 to 42.87; Analysis 1.1) or for participants with severe lung disease (MD 15.37 metres, 95% CI –10.71 to 41.43; Analysis 1.1).

3.

Forest plot of comparison: 1 Pulmonary rehabilitation versus no pulmonary rehabilitation, outcome: 1.1 Change in six‐minute walk test immediately following pulmonary rehabilitation.

1.1. Analysis.

Comparison 1: Pulmonary rehabilitation versus no pulmonary rehabilitation, Outcome 1: Change in 6‐minute walk distance immediately following pulmonary rehabilitation. Mean change from baseline, metres

Five studies reported results of the six‐minute walk test at long‐term (six to 12 months') follow‐up (Analysis 1.2) (Dale 2014; Dowman 2017; Holland 2008; Perez Bogerd 2018; Vainshelboim 2014). In those who participated in pulmonary rehabilitation, improvements in 6MWD were maintained six to 12 months following the intervention period with an MD of 32.43 metres (95% CI 15.58 to 49.28; 297 participants). This effect was within the MID range for 6MWD of 30 metres to 33 metres (Holland 2014). In the subgroup of participants with IPF, improvements in 6MWD were less evident at long‐term follow‐up with an MD of 1.64 metres (95% CI –24.89 to 28.17 metres; 3 studies, 123 participants; Analysis 1.2).

1.2. Analysis.

Comparison 1: Pulmonary rehabilitation versus no pulmonary rehabilitation, Outcome 2: Change in 6‐minute walk test at long‐term follow‐up. Mean change from baseline, metres

Sensitivity analysis using studies of high quality (low risk of bias) produced a similar estimate of the treatment effect for participants with ILD (MD 41.22 metres, 95% CI 26.80 to 55.64; 5 studies, 288 participants; Table 4). Tests of heterogeneity for all analyses of functional exercise capacity were not significant. A funnel plot of the complete data showed no evidence of asymmetry (Figure 4).

2. Summary of sensitivity analysis for interstitial lung disease.

Outcome	Subscale	Included studies	№ of participants	Heterogeneity	MD (95% CI)	Test of overall effect
6MWT	—	Dale 2014; Dowman 2017; Holland 2008; Naz 2018; Perez Bogerd 2018	288	I² = 35%, P = 0.19	41.22 metres (26.80 to 55.64)	P < 0.00001
Dyspnoea score	—	Dale 2014; Dowman 2017; Holland 2008; Naz 2018; Perez Bogerd 2018	288	I² = 70%, P = 0.01	–0.28 (–0.51 to –0.04)	P < 0.02
SGRQ	Symptoms	Dale 2014; Dowman 2017; Naz 2018; Perez Bogerd 2018	231	I² = 51%, P = 0.11	–13.76 (–18.49 to –9.04)	P < 0.00001
	Activity	Dale 2014; Dowman 2017; Naz 2018; Perez Bogerd 2018	231	I² = 21%, P = 0.28	–8.56 (–12.90 to –4.22)	P = 0.0001
	Impact	Dale 2014; Dowman 2017; Naz 2018; Perez Bogerd 2018	231	I² = 0%, P = 0.83	–7.91 (–11.54 to –4.29)	P < 0.0001
	Total	Dale 2014; Dowman 2017; Naz 2018; Perez Bogerd 2018	231	I² = 21%, P = 0.29	–8.13 (–11.24 to –5.02)	P < 0.00001
CRQ	Dyspnoea	Dale 2014; Dowman 2017; Holland 2008; Perez Bogerd 2018	270	I² = 41%, P = 0.18	0.61 (0.32 to 0.90)	P < 0.0001
	Fatigue	Dale 2014; Dowman 2017; Holland 2008; Perez Bogerd 2018	270	I² = 0%, P = 0.93	0.66 (0.40 to 0.92)	P < 0.00001
	Emotion	Dale 2014; Dowman 2017; Holland 2008; Perez Bogerd 2018	270	I² = 0%, P = 0.44	0.58 (0.35 to 0.81)	P < 0.00001
	Mastery	Dale 2014; Dowman 2017; Holland 2008; Perez Bogerd 2018	270	I² = 58%, P = 0.07	0.71 (0.44 to 0.98)	P < 0.00001

6MWT: six‐minute walk test; CI: confidence interval; CRQ: Chronic Respiratory Disease Questionnaire; MD: mean difference; SGRQ: St George's Respiratory Questionnaire.

4.

Funnel plot of comparison: 1 Pulmonary rehabilitation versus no pulmonary rehabilitation, outcome: 1.1 Change in six‐minute walk distance immediately following pulmonary rehabilitation. Mean change from baseline, metres.

Maximal exercise capacity

Four studies measured maximal exercise capacity using an incremental cycle ergometer test. Four studies provided sufficient data to conduct a meta‐analysis for peak work rate (Dale 2014; Holland 2008; Perez Bogerd 2018; Vainshelboim 2014), and three studies for VO₂ peak, maximum ventilation and maximum heart rate (Dale 2014; Holland 2008; Vainshelboim 2014). Peak work rate increased following pulmonary rehabilitation with an MD of 9.04 watts (95% CI 6.07 to 12.0; 81 participants in pulmonary rehabilitation group, 78 participants in control group; Figure 5; Analysis 1.3). There was an increase in peak work rate following pulmonary rehabilitation in the subgroup of participants with IPF with an MD of 9.94 watts (95% CI 6.39 to 13.49; 2 studies, 32 participants in pulmonary rehabilitation group, 30 participants in control group). This effect exceeded the MID of 4 watts proposed by Puhan 2011 for people with COPD. There was an increase in VO₂ peak between baseline and follow‐up with an MD of 1.28 mL/kg/minute in favour of the pulmonary rehabilitation group (95% CI 0.51 to 2.05; Analysis 1.4). There was a similar effect in the subgroup of participants with IPF with a common effect of 1.45 mL/kg/minute (95% CI 0.51 to 2.40; Analysis 1.4). Pulmonary rehabilitation resulted in an increase in maximum ventilation with an MD between groups of 7.21 L/minute in favour of the pulmonary rehabilitation group (95% CI 4.10 to 10.32; Analysis 1.5). The effect was more pronounced in the subgroup of participants with IPF (MD 9.80 L/minute, 95% CI 6.06 to 13.53; 2 studies, 62 participants; Analysis 1.5). There was an increase in peak watts (MD 5.4 watts, 95% CI 0.07 to 10.73; 1 study, 30 participants; Analysis 1.3) and in maximum ventilation (MD 6.95 L/minute, 95% CI 0.03 to 13.87; 1 study, 27 participants; Analysis 1.5) in favour of the pulmonary rehabilitation group in participants who desaturated. This effect on peak watts and maximum ventilation was not evident for participants with severe lung disease (Analysis 1.3; Analysis 1.4; Analysis 1.5). There was no evidence of an effect of pulmonary rehabilitation on VO₂ peak for participants with severe lung disease and for participants who desaturated (Analysis 1.4). There was no evidence of an effect of pulmonary rehabilitation on maximum heart rate (Analysis 1.6). Neither study reported data on maximal exercise capacity at long‐term follow‐up. Tests of heterogeneity were statistically high for peak work rate for both participants with ILD (I² = 89%, P < 0.00001) and the subgroup of participants with IPF (I² = 94%, P < 0.0001). There was substantial heterogeneity present for maximum ventilation for participants with ILD (I² = 68%, P = 0.05) and VO₂ peak for the IPF subgroup (I² = 73%, P = 0.05). The high heterogeneity within this analysis could have stemmed from the small number of studies and small number of participants.

5.

Forest plot of comparison: 1 Pulmonary rehabilitation versus no pulmonary rehabilitation, outcome: 1.3 Change in peak work rate immediately following pulmonary rehabilitation, watts.

1.3. Analysis.

Comparison 1: Pulmonary rehabilitation versus no pulmonary rehabilitation, Outcome 3: Change in peak work rate immediately following pulmonary rehabilitation, watts

1.4. Analysis.

Comparison 1: Pulmonary rehabilitation versus no pulmonary rehabilitation, Outcome 4: Change in VO ₂ peak immediately following pulmonary rehabilitation, mL/kg/minute

1.5. Analysis.

Comparison 1: Pulmonary rehabilitation versus no pulmonary rehabilitation, Outcome 5: Change in maximum ventilation (Ve max ) immediately following pulmonary rehabilitation, L/minute

1.6. Analysis.

Comparison 1: Pulmonary rehabilitation versus no pulmonary rehabilitation, Outcome 6: Change in maximum heart rate immediately following pulmonary rehabilitation, beats/minute

Secondary outcomes

Dyspnoea

Eleven studies (512 participants) measured dyspnoea, with four reporting reduced dyspnoea immediately following pulmonary rehabilitation (Baradzina 2005; Holland 2008; Naz 2018; Vainshelboim 2014), and six reporting no change (Dale 2014; Dowman 2017; Ku 2017; Nishiyama 2008; Perez Bogerd 2018; Wewel 2005). We pooled data from seven studies for meta‐analysis (178 participants in the pulmonary rehabilitation group and 170 participants in the control group) (Figure 6; Analysis 1.7). Six studies utilised the modified Medical Research Council Scale (Dale 2014; Dowman 2017; Holland 2008; Naz 2018; Perez Bogerd 2018; Vainshelboim 2014), and one used the Baseline Dyspnoea Index (Nishiyama 2008). The common effect (SMD) for change in dyspnoea was –0.36 in favour of the pulmonary rehabilitation group (95% CI –0.58 to –0.14). There was a greater reduction in dyspnoea among participants with IPF (80 participants in the pulmonary rehabilitation group and 75 participants in the control group), with an SMD of –0.41 (95% CI –0.74 to –0.09; 4 studies). If the pooled SMD estimate was re‐expressed on the modified Medical Research Council Scale (5‐point scale, 0 to 4), it corresponded to an MD of –0.32 points (95% CI –0.52 to –0.13) for ILD and an MD of –0.37 points (95% CI –0.67 to –0.08) for the IPF subgroup for the IPF subgroup. This effect was smaller than the MID of one point for the modified Medical Research Dyspnoea Scale (Jones 2013). There was a small effect of pulmonary rehabilitation on dyspnoea with an SMD of –0.39 in participants with severe disease (95% CI –0.79 to 0.00; 1 study, 103 participants; Analysis 1.7). This effect was less evident in participants who desaturated (Figure 6). Six studies reported dyspnoea at long‐term follow‐up (Dale 2014; Dowman 2017; Holland 2008; Perez Bogerd 2018; Vainshelboim 2014; Wallaert 2020) (Analysis 1.8). In participants who received pulmonary rehabilitation, there was a reduction in dyspnoea that was still evident at long‐term follow‐up with an MD of –0.29 (95% CI –0.49 to –0.10; 6 studies, 335 participants; Analysis 1.8). There was a greater reduction in dyspnoea in the subgroup of participants with IPF at long‐term follow‐up with an MD of –0.38 (95% CI –0.72 to –0.05; 3 studies, 123 participants; Analysis 1.8). There was no evidence of a difference at long‐term follow‐up of pulmonary rehabilitation on dyspnoea in participants with severe disease (MD 0.14, 95% CI –0.36 to 0.63; 2 studies, 84 participants) and in those who desaturated (MD –0.03, 95% CI –0.42 to 0.35; 2 studies, 103 participants) (Analysis 1.8). Tests of heterogeneity for all analyses of dyspnoea were statistically substantial for both the participants with ILD (I² = 71%, P = 0.002) and the IPF subgroup (I² = 67%, P = 0.03) immediately following pulmonary rehabilitation, although this was not significant at long‐term follow‐up. The reduction in heterogeneity at long‐term follow‐up could stem from results in the meta‐analysis including a majority of studies with low risk of bias.

6.

Forest plot of comparison: 1 Pulmonary rehabilitation versus no pulmonary rehabilitation, outcome: 1.6 Dyspnoea score immediately following pulmonary rehabilitation.

1.7. Analysis.

Comparison 1: Pulmonary rehabilitation versus no pulmonary rehabilitation, Outcome 7: Change in dyspnoea score immediately following pulmonary rehabilitation

1.8. Analysis.

Comparison 1: Pulmonary rehabilitation versus no pulmonary rehabilitation, Outcome 8: Change in dyspnoea score at long‐term follow‐up

Quality of life

Fifteen studies measured health‐related quality of life, with 11 studies reporting differences between groups immediately following pulmonary rehabilitation (Dale 2014; Dowman 2017; Gaunaurd 2014; Holland 2008; Jarosch 2020; Ku 2017; Lanza 2019; Naz 2018; Nishiyama 2008; Perez Bogerd 2018; Vainshelboim 2014). Two studies reported improvement in health‐related quality of life following pulmonary rehabilitation; however, it was unclear if this finding reached statistical significance (Shen 2016; Xiao 2019). In the remaining studies, there was no evidence of differences between groups (Baradzina 2005; Mejia 2000; Wewel 2005). Six studies utilised the CRQ (Dale 2014; Dowman 2017; Holland 2008; Jarosch 2020; Mejia 2000; Perez Bogerd 2018), 10 used the SGRQ (Dale 2014; De Las Heras 2019; Ku 2017; Naz 2018; Nishiyama 2008; Perez Bogerd 2018; Shen 2016; Vainshelboim 2014; Wewel 2005; Xiao 2019), three used the SGRQ‐I (Dowman 2017; Gaunaurd 2014; Lanza 2019), two used the SF‐36 (Jarosch 2020; Naz 2018), and one used the WHO questionnaire (Baradzina 2005). There were sufficient raw data to conduct meta‐analyses for all domains of the CRQ (Dyspnoea, Fatigue, Emotional Function and Mastery) and SGRQ (Symptoms, Activity, Impact and Total). We pooled the SGRQ and SGRQ‐I for the meta‐analyses since the SGRQ‐I and SGRQ have similar psychometric properties. The SGRQ‐I has been designed to be more responsive in people with IPF whereas the SGRQ was originally designed for people with COPD. Not all the studies provided the results for all four domains of the SGRQ questionnaire; therefore, the numbers of studies and participants varies per domain.

St George's Respiratory Questionnaire

Thirteen studies utilised either the SGRQ or SGRQ‐I to assess health‐related quality of life (Dale 2014; De Las Heras 2019; Dowman 2017; Gaunaurd 2014; Ku 2017; Lanza 2019; Naz 2018; Nishiyama 2008; Perez Bogerd 2018; Shen 2016; Vainshelboim 2014; Wewel 2005; Xiao 2019). Pulmonary rehabilitation improved SGRQ Symptoms (MD –15.58, 95% CI –19.54 to –11.62; 7 studies, 312 participants; Analysis 1.9), SGRQ Activity (MD –2.47, 95% CI –4.11 to –0.83; 7 studies, 312 participants; Analysis 1.10), SGRQ Impact (MD –8.81, 95% CI –11.17 to –6.46; 7 studies, 312 participants; Analysis 1.11) and SGRQ Total (MD –9.29, 95% CI –11.06 to –7.52; 11 studies, 478 participants; Figure 7; Analysis 1.12). There was a similar effect in favour of pulmonary rehabilitation in participants with IPF for SGRQ Symptoms (MD –13.92, 95% CI –19.68 to –8.17; 4 studies, 142 participants; Analysis 1.9), SGRQ Impact (MD –8.94, 95% CI –11.76 to –6.13; 4 studies, 142 participants; Analysis 1.11) and SGRQ Total (MD –7.91 95% CI –10.55 to –5.26; 6 studies, 194 participants; Figure 7; Analysis 1.12). Pulmonary rehabilitation had a smaller effect on SGRQ Activity for participants with IPF (MD –1.71, 95% CI –3.44 to 0.01; 4 studies, 142 participants; Analysis 1.10). The improvements in the SGRQ following rehabilitation exceeded the MID for Symptoms (MID = 8), Impact (MID = 7) and Total (MID = 7) score in participants with ILD and the subgroup of IPF (Swigris 2010). Data regarding effects on quality of life in participants with severe disease and in participants who desaturated were available from one study (Dowman 2017). There were improvements favouring pulmonary rehabilitation for SGRQ Activity (MD –8.20, 95% CI –15.55 to –0.85; Analysis 1.10) and Total score (MD of –6.00, 95% CI –11.56 to –0.44; Figure 7; Analysis 1.12) in participants who desaturated.

1.9. Analysis.

Comparison 1: Pulmonary rehabilitation versus no pulmonary rehabilitation, Outcome 9: Change in quality of life (St George's Respiratory Questionnaire (SGRQ) Symptoms) immediately following pulmonary rehabilitation

1.10. Analysis.

Comparison 1: Pulmonary rehabilitation versus no pulmonary rehabilitation, Outcome 10: Change in quality of life (SGRQ Activity) immediately following pulmonary rehabilitation

1.11. Analysis.

Comparison 1: Pulmonary rehabilitation versus no pulmonary rehabilitation, Outcome 11: Change in quality of life (SGRQ Impact) immediately following pulmonary rehabilitation

7.

Forest plot of comparison: 1 Pulmonary rehabilitation versus no pulmonary rehabilitation, outcome: 1.12 Change in quality of life (SGRQ Total) immediately following pulmonary rehabilitation.

1.12. Analysis.

Comparison 1: Pulmonary rehabilitation versus no pulmonary rehabilitation, Outcome 12: Change in quality of life (SGRQ Total) immediately following pulmonary rehabilitation

Four studies provided data regarding longer‐term effects on quality of life for meta‐analysis (Dale 2014; Dowman 2017; Perez Bogerd 2018; Vainshelboim 2014). The effects of pulmonary rehabilitation were still evident at long‐term follow‐up for all SGRQ domains except the Activity score (SGRQ Symptoms: MD –11.31, 95% CI –16.58 to –6.03; 240 participants; Analysis 1.13; SGRQ Impact: MD –4.73, 95% CI –7.76 to –1.69; 240 participants; Analysis 1.15; SGRQ Total score: MD –4.93, 95% CI –7.81 to –2.06; 240 participants; Analysis 1.16). In participants with IPF, the effects of pulmonary rehabilitation were evident at long‐term follow‐up for SGRQ Impact (MD –4.59, 95% CI –8.60 to –0.57; 2 studies, 89 participants; Analysis 1.15). Those with severe disease had improved SGRQ Symptoms scores compared to controls at long‐term follow‐up (MD –12.0, 95% CI –22.41 to –1.59; 1 study, 61 participants; Analysis 1.13). Heterogeneity was statistically high for the symptom score in the subgroup of IPF immediately following pulmonary rehabilitation (I² = 74%, P = 0.01). There was substantial heterogeneity for participants with ILD for the Activity score immediately following pulmonary rehabilitation (I² = 54%, P = 0.04). A funnel plot of the SGRQ Total data showed a tendency towards asymmetry, suggesting potential publication bias (Figure 8).

1.13. Analysis.

Comparison 1: Pulmonary rehabilitation versus no pulmonary rehabilitation, Outcome 13: Change in quality of life (SGRQ Symptoms) at long‐term follow‐up

1.15. Analysis.

Comparison 1: Pulmonary rehabilitation versus no pulmonary rehabilitation, Outcome 15: Change in quality of life (SGRQ Impact) at long‐term follow‐up

1.16. Analysis.

Comparison 1: Pulmonary rehabilitation versus no pulmonary rehabilitation, Outcome 16: Change in quality of life (SGRQ Total) at long‐term follow‐up

8.

Funnel plot of comparison: 1 Pulmonary rehabilitation versus no pulmonary rehabilitation, outcome: 1.12 Change in quality of life (SGRQ Total) immediately following pulmonary rehabilitation.

Chronic Respiratory Disease Questionnaire

Five studies provided data for meta‐analysis with a total of 175 participants in the pulmonary rehabilitation group and 146 participants in the control group (Dale 2014; Dowman 2017; Holland 2008; Jarosch 2020; Perez Bogerd 2018). Pulmonary rehabilitation improved CRQ Dyspnoea (MD 0.68, 95% CI 0.42 to 0.93; Figure 9; Analysis 1.17), CRQ Fatigue (MD 0.66, 95% CI 0.43 to 0.90; Analysis 1.18), CRQ Emotion (MD 0.63, 95% CI 0.42 to 0.84; Analysis 1.19) and CRQ Mastery (MD 0.67, 95% CI 0.44 to 0.90; Analysis 1.20). These improvements exceeded the MID of 0.5 points (Jaeschke 1989; Redelmeier 1996). There was a similar effect in favour of pulmonary rehabilitation in participants with IPF (3 studies, 169 participants) for CRQ Dyspnoea (MD 0.81, 95% CI 0.49 to 1.14; Figure 9; Analysis 1.17), CRQ Fatigue (MD 0.67, 95% CI 0.36 to 0.98; Analysis 1.18), CRQ Emotion (MD 0.64, 95% CI 0.33 to 0.95; Analysis 1.19) and CRQ Mastery (MD 0.63, 95% CI 0.33 to 0.94; Analysis 1.20). These improvements also exceeded the MID of 0.5 points (Jaeschke 1989; Redelmeier 1996). Two studies provided data regarding effects on quality of life in participants with severe disease and in those who desaturated (Dowman 2017; Holland 2008). There were improvements in CRQ Dyspnoea and CRQ Fatigue, that exceeded the MID of 0.5 points (Jaeschke 1989; Redelmeier 1996), for participants with severe disease and for those who desaturated. There were improvements in CRQ Mastery for those who desaturated (Analysis 1.17; Analysis 1.18; Analysis 1.19; Analysis 1.20).

9.

Forest plot of comparison: 1 Pulmonary rehabilitation versus no pulmonary rehabilitation, outcome: 1.17 Change in quality of life (Chronic Respiratory Disease Questionnaire (CRQ) Dyspnoea) immediately following pulmonary rehabilitation.

1.17. Analysis.

Comparison 1: Pulmonary rehabilitation versus no pulmonary rehabilitation, Outcome 17: Change in quality of life (Chronic Respiratory Disease Questionnaire (CRQ) Dyspnoea) immediately following pulmonary rehabilitation

1.18. Analysis.

Comparison 1: Pulmonary rehabilitation versus no pulmonary rehabilitation, Outcome 18: Change in quality of life (CRQ Fatigue) immediately following pulmonary rehabilitation.

1.19. Analysis.

Comparison 1: Pulmonary rehabilitation versus no pulmonary rehabilitation, Outcome 19: Change in quality of life (CRQ Emotion) immediately following pulmonary rehabilitation

1.20. Analysis.

Comparison 1: Pulmonary rehabilitation versus no pulmonary rehabilitation, Outcome 20: Change in quality of life (CRQ Mastery) immediately following pulmonary rehabilitation

Four studies provided data for meta‐analysis regarding longer‐term effects on quality of life (Dale 2014; Dowman 2017; Holland 2008; Perez Bogerd 2018). The effects of pulmonary rehabilitation were evident at long‐term follow‐up for CRQ Dyspnoea (MD 0.42, 95% CI 0.17 to 0.68; Analysis 1.21), CRQ Fatigue (MD 0.40 95% CI 0.09 to 0.70; Analysis 1.22), CRQ Emotion (MD 0.51, 95% CI 0.26 to 0.77; Analysis 1.23) and CRQ Mastery (MD 0.47, 95% CI 0.17 to 0.78; Analysis 1.24). The effects of pulmonary rehabilitation were no longer evident at long‐term follow‐up for any CRQ domain for the subgroup of participants with IPF (28 studies, 95 participants), for those with severe disease (2 studies, 84 participants) or those who desaturated (2 studies, 103 participants) (Analysis 1.21; Analysis 1.22; Analysis 1.23; Analysis 1.24). There was substantial heterogeneity only for participants with ILD for the CRQ Emotion domain at long‐term follow‐up (I² = 57%, P = 0.07). Sensitivity analysis using studies of high quality (low risk of bias) produced a similar estimate of the treatment effect for participants with ILD (Table 4).

1.21. Analysis.

Comparison 1: Pulmonary rehabilitation versus no pulmonary rehabilitation, Outcome 21: Change in quality of life (CRQ Dyspnoea) at long‐term follow‐up

1.22. Analysis.

Comparison 1: Pulmonary rehabilitation versus no pulmonary rehabilitation, Outcome 22: Change in quality of life (CRQ Fatigue) at long‐term follow‐up

1.23. Analysis.

Comparison 1: Pulmonary rehabilitation versus no pulmonary rehabilitation, Outcome 23: Change in quality of life (CRQ Emotion) at long‐term follow‐up

1.24. Analysis.

Comparison 1: Pulmonary rehabilitation versus no pulmonary rehabilitation, Outcome 24: Change in quality of life (CRQ Mastery) at long‐term follow‐up

Adverse effects

Ten studies provided information regarding adverse events (Dale 2014; Dowman 2017; Holland 2008; Jarosch 2020; Ku 2017; Naz 2018; Nishiyama 2008; Perez Bogerd 2018; Vainshelboim 2014; Wallaert 2020), none of which reported adverse events during the study period. Four studies reported the death of one pulmonary rehabilitation participant during the intervention period; however, this was believed to be unrelated to the intervention received (Jackson 2014; Jarosch 2020; Perez Bogerd 2018; Wallaert 2020).

Survival

Four studies (291 participants) reported long‐term (six to 12 months) survival (Dowman 2017; Holland 2008; Perez Bogerd 2018; Vainshelboim 2014). There were five deaths in the exercise training group and 12 deaths in the control group (Analysis 1.25).

1.25. Analysis.

Comparison 1: Pulmonary rehabilitation versus no pulmonary rehabilitation, Outcome 25: Long‐term survival

Sensitivity analysis

We performed a sensitivity analysis including only studies of high quality, for which random sequence generation, allocation concealment and incomplete outcome data were rated as low risk (see 'Risk of bias' table in Figure 2). Five studies met the criteria for high quality (Dale 2014; Dowman 2017; Holland 2008; Naz 2018; Perez Bogerd 2018). Three of the five studies also had low risk of detection bias (blinded outcome assessment) (Dale 2014; Dowman 2017; Holland 2008). The sensitivity analysis effect estimates were consistent with the overall summary effect estimates for functional exercise capacity (MD 41.221.223.44, 95% CI 26.806.809.64 to 55.645.647.24, 288 participants, Table 4), dyspnoea score, SGRQ Activity, SGRQ Impact, SGRQ Total and all four domains of the CRQ questionnaire (Table 4). Any heterogeneity that was previously present (Dyspnoea and SGRQ Activity) was reduced or no longer apparent, and all outcomes continued to be statistically significant when restricted to studies of high quality (Table 4). Sensitivity analysis for the subgroup of participants with IPF was not performed for any outcome as the number of included studies in the sensitivity analysis was limited to one or two studies.

Discussion

Summary of main results

We identified 21 eligible studies comparing pulmonary rehabilitation versus no pulmonary rehabilitation or sham control among people with ILD. There were no adverse effects of this treatment identified. Pulmonary rehabilitation probably improves functional exercise capacity, as measured by the six‐minute walk test, and maximum exercise capacity, represented by peak work rate. Reduction in dyspnoea and improvement in quality of life probably occurs immediately following pulmonary rehabilitation. The effects of pulmonary rehabilitation were comparable in the subgroup of participants with IPF. There were insufficient data available to allow conclusions regarding the effects of pulmonary rehabilitation among those with severe disease and those who desaturated. Sustained benefits in functional exercise capacity and quality of life are probable and improvements in dyspnoea may also be sustained at long‐term follow‐up. Sustained benefits in functional exercise capacity are less certain at long‐term follow‐up in those with IPF but there may be lasting effects of pulmonary rehabilitation in dyspnoea and quality of life.

The results of the previous version of this review supported pulmonary rehabilitation in the management of ILD and results of this current update reconfirm these findings, and add new evidence of sustained benefits at six to 12 months following pulmonary rehabilitation. Mean improvement in the six‐minute walk test following pulmonary rehabilitation was 40.07 metres, which is similar to the mean improvement of 43.93 metres seen in people with COPD who have undergone pulmonary rehabilitation (McCarthy 2015). This suggests that people with ILD receive comparable benefit from pulmonary rehabilitation. This improvement exceeds the MID for 6MWD among people with ILD, which ranges from 30 metres to 33 metres (Holland 2014). This indicates that providing a pulmonary rehabilitation programme that is well‐aligned with current guidelines for pulmonary rehabilitation (Spruit 2013) results in clinically important changes in functional capacity. There were also improvements in dyspnoea and health‐related quality of life following pulmonary rehabilitation. In addition, the magnitude of improvement for health‐related quality of life exceeded the MID in three of the four domains for the SGRQ (Swigris 2010), and all four domains of the CRQ (Jaeschke 1989; Redelmeier 1996). This supports the view that the observed improvements are clinically important and meaningful for patients.

Overall completeness and applicability of evidence

The update of this systematic review was substantial in that it included 12 additional studies and an increase in total study participants from 365 to 675 and number of studies from five to 16 for inclusion in the meta‐analysis. Included studies involved participants with a range of ILDs, and studies often included samples of participants with mixed diagnoses (Dowman 2017; Holland 2008; Ku 2017; Lanza 2019; Mejia 2000; Menon 2011; Perez Bogerd 2018; Wewel 2005). This recruitment strategy probably reflects the relatively uncommon nature and the shared pathophysiological features of many ILDs. Participants with IPF often have more severe physiological derangement and a more rapid disease course compared with those with other ILDs (Lama 2004), and subsequently it has been hypothesised that pulmonary rehabilitation might be less effective in people with IPF. However, our earlier review indicated that pulmonary rehabilitation is equally effective in people with IPF and the results of this current update strengthens these findings. Participants with IPF did achieve improvements in six‐minute walk test results, maximum exercise capacity, dyspnoea and health‐related quality of life. Improvement in the six‐minute walk test was comparable among participants with IPF (37.25 metres in participants with IPF versus 40.07 metres in all participants) and this mean improvement exceeded the MID for the 6MWD in people with IPF, which is in the range of 29 metres to 34 metres (Holland 2014). Changes in quality of life, dyspnoea and maximum exercise capacity were also comparable in the subgroup of participants with IPF. The comparable benefits in ILD and IPF suggest the response to exercise may not vary across the disease spectrum, despite the known heterogeneity between ILD subgroups.

Participants with IPF may experience sustained benefits in dyspnoea and quality of life, although the sustained improvement in functional capacity is less certain. IPF is a chronic progressive fibrotic lung disease, although the extent and rate of progression of IPF can vary, with rapid decline in some people and more gradual disease progression and periods of stability in others. This progressive nature of IPF may impact the long‐term benefit in functional capacity in some people with IPF, and, therefore, establishing strategies to manage this functional decline are essential.

It should be noted, however, that of the 16 studies contributing to the meta‐analysis, five included only participants with IPF (He 2016; Jackson 2014; Jarosch 2020; Nishiyama 2008; Vainshelboim 2014), whilst two others included a majority of participants with IPF (Dowman 2017; Holland 2008); thus, overall results of the meta‐analysis are strongly influenced by the response of participants with IPF.

All studies in this review utilised aerobic exercise training or a combination of aerobic and resisted exercise training. These strategies are well aligned with current guidelines for pulmonary rehabilitation (Spruit 2013), and the results, therefore, are readily applicable to clinical practice in pulmonary rehabilitation programmes. However, we were unable to draw inferences regarding the most effective exercise training strategy for people with ILD. Given the relatively modest improvements in exercise capacity documented here, this may be an important area for future research.

The included studies used a range of programme durations (three to 48 weeks) and training frequencies (two to five sessions per week). Longer programmes and more frequent sessions appear to yield greater benefit for people with other chronic lung diseases (Spruit 2013). To date the most effective dose of pulmonary rehabilitation for people with ILD has not been established.

Pulmonary rehabilitation by definition should include both an exercise and education component. There was substantial heterogeneity in the amount and type of education provided between the included studies and there was limited detail about the nature of the education. Therefore, it is difficult to ascertain the impact the inclusion or exclusion or type of education had on the outcomes of pulmonary rehabilitation.

A limitation of this review is it did not include the evaluation of pulmonary rehabilitation on psychological health. Anxiety and depression commonly occur in people with ILD, often further impacting health‐related quality of life. Pulmonary rehabilitation may promote an improvement in anxiety and depression for people with ILD. Therefore, we will include the assessment of the impact of pulmonary rehabilitation on anxiety and depression in the next review update.

Quality of the evidence

This review had several potential sources of bias. Of the 21 study identified, seven were available only in abstract form (Baradzina 2005; De Las Heras 2019; Lanza 2019; Mejia 2000; Menon 2011; Shen 2016; Wewel 2005). These publications provided limited data on the outcomes of interest, and it was not possible for review authors to obtain additional data from study authors. Data that could be pooled for meta‐analysis usually ranged from five to 13 studies; however, there were a number of meta‐analyses that were limited to two or three studies. These were often the subgroup analysis of participants with IPF, severe disease and those who desaturated and the analyses at long‐term follow‐up. Despite this limitation, there was consistency in most reported outcomes, with all but one study reporting improved functional exercise capacity immediately following pulmonary rehabilitation (Jackson 2014). Reasons for lack of improvement in functional exercise capacity in the study by Jackson 2014 are unclear but may have been related to the method by which the six‐minute walk test was conducted or may have involved the small numbers of included participants. As pulmonary rehabilitation is a physical intervention, it can be assumed that no participants were blinded. Therefore, the blinding of outcome assessors is critical to limit bias for outcomes such as exercise capacity, however, only four studies reported blinding of the assessor (Dale 2014; De Las Heras 2019; Dowman 2017; Holland 2008). Five studies reported use of an intention‐to‐treat analysis (Dale 2014; Dowman 2017; Holland 2008; Perez Bogerd 2018; Wallaert 2020). Three studies reported that all participants completed the study and were included in the analysis (He 2016; Ku 2017; Naz 2018). Given the progressive nature of many ILDs, a significant dropout rate is likely and may impact both the size of the reported treatment effect and the feasibility of the intervention.

Using the GRADE system, we rated the review outcomes as providing very low‐ to moderate‐certainty evidence. Review outcomes were rated as of moderate certainty (6MWD, quality of life assessed by SGRQ) and low certainty (maximum exercise capacity, dyspnoea and survival) for ILD and as moderate certainty (6MWD, quality of life assessed by SGRQ), low certainty (survival) and very low certainty (maximum exercise capacity, dyspnoea) for IPF. Risk of bias was increased by poor reporting of methods (selection bias), lack of assessor blinding (detection bias) and lack of intention‐to‐treat analyses (attrition bias). This may have overestimated the effect provided by the meta‐analyses, however, sensitivity analysis including only high‐quality studies revealed similar effects in favour of pulmonary rehabilitation. Imprecision was increased by the small numbers of included studies and participants for the subgroup analysis on participants with IPF with two to three studies and 62 to 169 participants contributing to some outcomes. Inconsistency was increased by the presence of heterogeneity in all CRQ domains and Dyspnoea score for participants with ILD and the subgroup with IPF.

Potential biases in the review process

One review author independently extracted data and a second review author thoroughly checked accuracy. We resolved any discrepancies through discussion. Two review authors independently assessed risk of bias ratings. Clarification on data extraction and risk of bias assessments was sought from other co‐authors as required.

We conducted a broad search, which included handsearching of reference lists, conference abstracts and trial registries. We included studies that were published only in abstract form to ensure that all available trials were included. We sought additional information from eight authors, six of whom provided information, to increase the accuracy of the data for inclusion. However, there were some studies from which we could not obtain additional data. This may have influenced assessment of trial quality and some estimates of effect.

Three of the review authors (AH, LD and CJH) conducted two of the studies in the review (Dowman 2017; Holland 2008). To limit the bias of the review process, LD and AM completed the risk of bias assessments for Holland 2008 and AM completed them for Dowman 2017.

Agreements and disagreements with other studies or reviews

This is the second update of this Cochrane Review, which was last published in 2014 (Dowman 2014). Data included in the previous review suggested that improvement in functional exercise capacity following pulmonary rehabilitation in ILD was comparable to that seen in people with COPD (Lacasse 2006; McCarthy 2015). The results of this update reconfirm these findings (McCarthy 2015), and extend the findings to show persistent benefits at six to 12 months following pulmonary rehabilitation. The earlier version found similar, clinically important effects of rehabilitation in both the overall ILD group and the subgroup of IPF. This review reconfirms these findings with equivalent or greater improvements in the subgroup for participants with IPF. In addition, this review found the improvements were similar for the CRQ and greater for the SGRQ, except for the Activity score, for both people with ILD and in the subgroup of IPF than those observed in people with COPD. These significant findings support the international pulmonary rehabilitation statement suggesting that people with ILD should be included in pulmonary rehabilitation programmes (Spruit 2013) and strongly indicates that pulmonary rehabilitation should be part of the standard of care in ILD, just as it has become in COPD. Compared with the earlier version of this Cochrane Review (Dowman 2014), the three‐fold increase in number of trials led to smaller CIs, providing a more precise estimate of treatment effect.

Authors' conclusions

Implications for practice.

This review indicates that pulmonary rehabilitation can be performed safely with no evidence of adverse events in people with interstitial lung disease (ILD). Moderate‐certainty evidence suggests improvements in functional exercise capacity and health‐related quality of life are probable immediately following pulmonary rehabilitation in people with ILD and idiopathic pulmonary fibrosis (IPF). Low‐certainty evidence suggests pulmonary rehabilitation may improve maximum exercise capacity and dyspnoea in people with ILD and IPF. It is appropriate to include people with ILD of all types in a standard pulmonary rehabilitation programme. Benefits of pulmonary rehabilitation are probably sustained in the longer term in people with ILD, and may be sustained in people with IPF.

Implications for research.

The optimum exercise training method for people with ILD has not been established. Large studies are required to determine the optimal exercise training strategy that provides the greatest benefits of pulmonary rehabilitation, and to investigate ways to ensure that the benefits of exercise training can be sustained longer term, particularly for people with IPF. Future trials should ensure that assessors are blinded to the intervention and that appropriate methods are used to account for dropouts.

What's new

Date	Event	Description
16 April 2020	New citation required and conclusions have changed	Conclusions updated to incorporate data from new studies. Methods updated to reflect current requirements. Elements of background, results and discussion redrafted. Risk of bias blinding domain split into participants and personnel and outcomes assessors.
16 April 2020	New search has been performed	New literature search run 16 April 2020. 13 new studies identified.

History

Protocol first published: Issue 1, 2007
Review first published: Issue 4, 2008

Date	Event	Description
27 June 2014	New citation required but conclusions have not changed	Four new studies were identified Background was modified to describe pulmonary rehabilitation Summary of findings table was added
27 June 2014	Amended	Title changed from 'physical training' to 'pulmonary rehabilitation' to reflect current practice and terminology
27 June 2014	New search has been performed	New literature search run
2 February 2010	New search has been performed	Updated search re‐run; no new studies were identified.
28 January 2009	Amended	Contact details changed
10 April 2008	Amended	Converted to new review format.
11 October 2006	New citation required and major changes	Substantive amendment

Appendices

Appendix 1. Cochrane Airways Trials Register (Cochrane Register of Studies)/CENTRAL search strategy

#1 MESH DESCRIPTOR Lung Diseases, Interstitial EXPLODE ALL
#2 MESH DESCRIPTOR Pulmonary Fibrosis EXPLODE ALL
#3 MESH DESCRIPTOR Sarcoidosis, Pulmonary EXPLODE ALL
#4 interstitial* NEAR3 (lung* or disease* or pneumon*)
#5 (pulmonary* or lung* or alveoli*) NEAR3 (fibros* or fibrot*)
#6 (pulmonary* or lung*) NEAR3 (sarcoid* or granulom*)
#7 ILD:MISC1
#8 {OR #1‐#7}
#9 rehabilitat* or fitness* or exercis* or physical* or train* or activ* or physiotherap* or kinesiotherap* or exert*
#10 MeSH DESCRIPTOR Physical Therapy Modalities Explode All
#11 MeSH DESCRIPTOR Exercise Explode All
#12 MeSH DESCRIPTOR Physical Fitness
#13 MeSH DESCRIPTOR Physical Exertion
#14 MeSH DESCRIPTOR Rehabilitation
#15 {OR #9‐#14}
#16 #15 AND #8

Appendix 2. MEDLINE (Ovid) search strategy

1. exp Lung Diseases, Interstitial/ 
2. (interstitial$ adj3 lung$ adj3 disease$).tw. 
3. (interstitial$ adj3 (fibros$ or pneumonitis or pneumonia or pneumopathy)).tw. 
4. alveolitis.tw. 
5. exp Bronchiolitis Obliterans/ or (bronchiolitis adj3 obliterans).tw. 
6. (goodpasture$ adj3 syndrome$).tw. 
7. granulomatosis.tw. 
8. exp Histiocytosis/ or histiocytosis$.tw. 
9. exp Pneumoconiosis/ or pneumoconiosis.mp. or pneumokoniosis.mp. or pneumonoconiosis.mp. 
10. bagassosis.tw. 
11. (pulmonar$ adj3 sarcoid$).tw. 
12. (pulmonar$ adj3 fibros$).tw. 
13. (wegener$ adj3 granuloma$).tw. 
14. (lung$ adj3 purpura).tw. 
15. ((bird$ or farmer$ or pigeon$ or avian$ or budgerigar$) adj3 (lung$ or disease$)).tw. 
16. (asbestosis or byssinosis or siderosis or silicosis or berylliosis or anthracosilicosis or silicotuberculosis).mp. 
17. or/1‐16 
18. exp Scleroderma, Systemic/ 
19. scleroderma.tw. 
20. exp Rheumatic Diseases/ 
21. rheumatic$.tw. 
22. or/18‐21 
23. 22 and (lung$ or pulmonary$ or respiratory$).tw. 
24. 17 or 23 
25. exp Exercise Therapy/ 
26. exp exercise/ 
27. exp Physical Fitness/ 
28. rehabilitation/ 
29. exp Physical Therapy Modalities/ 
30. exp Physical Exertion/ 
31. (rehabilitat$ or fitness$ or exercis$ or physical$ or train$ or active$ or activit$ or physiotherap$ or kinesiotherap$ or exert$).tw. 
32. or/25‐31 
33. 32 and 24 
34. (controlled clinical trial or randomised controlled trial).pt. 
35. (randomised or randomised).ab,ti. 
36. placebo.ab,ti. 
37. dt.fs. 
38. randomly.ab,ti. 
39. trial.ab,ti. 
40. groups.ab,ti. 
41. or/34‐40 
42. Animals/ 
43. Humans/ 
44. 42 not (42 and 43) 
45. 41 not 44 
46. 33 and 45 
47. (2019$ or "2020").dt. 
48. 46 and 47

Appendix 3. Embase (Ovid) search strategy

1 exp Interstitial Lung Disease/
2 (interstitial$ adj3 lung$ adj3 disease$).tw.
3 Interstitial Pneumonia/ or exp lung fibrosis/ or (interstitial$ adj3 (fibros$ or pneumonitis or pneumonia or pneumopathy)).tw.
4 alveolitis.tw.
5 exp Bronchiolitis Obliterans/ or (bronchiolitis adj obliterans).tw.
6 (goodpasture$ adj3 syndrome$).tw.
7 granulomatosis.tw.
8 exp Histiocytosis/ or histiocytosis$.tw.
9 exp Pneumoconiosis/ or pneumoconiosis.tw. or pneumokoniosis.tw. or pneumonoconiosis.tw.
10 bagassosis.tw.
11 (pulmonar$ adj3 sarcoid$).tw.
12 (pulmonar$ adj3 fibros$).tw.
13 (wegener$ adj3 granuloma$).tw.
14 (lung$ adj3 purpura).tw.
15 ((bird$ or farmer$ or pigeon$ or avian$ or budgerigar$) adj3 (lung$ or disease$)).tw.
16 (asbestosis or byssinosis or siderosis or silicosis or berylliosis or anthracosilicosis or silicotuberculosis).tw.
17 or/1‐16
18 exp systemic sclerosis/
19 scleroderma.tw.
20 exp Rheumatic Disease/
21 rheumatic$.tw.
22 or/18‐21
23 22 and (lung$ or pulmonary$ or respiratory$).tw.
24 17 or 23
25 exp kinesiotherapy/
26 exp exercise/
27 exp fitness/
28 rehabilitation/
29 exp physiotherapy/
30 exertion.mp.
31 (rehabilitat$ or fitness$ or exercis$ or physical$ or train$ or active$ or activit$ or physiotherap$ or kinesiotherap$ or exert$).tw.
32 or/25‐31
33 32 and 24
34 Randomized Controlled Trial/
35 randomization/
36 controlled clinical trial/
37 Double Blind Procedure/
38 Single Blind Procedure/
39 Crossover Procedure/
40 (clinica$ adj3 trial$).tw.
41 ((singl$ or doubl$ or trebl$ or tripl$) adj3 (mask$ or blind$ or method$)).tw.
42 exp Placebo/
43 placebo$.ti,ab.
44 random$.ti,ab.
45 ((control$ or prospectiv$) adj3 (trial$ or method$ or stud$)).tw.
46 (crossover$ or cross‐over$).ti,ab.
47 or/34‐46
48 exp animals/ or exp invertebrate/ or animal experiment/ or animal model/ or animal tissue/ or animal cell/ or nonhuman/
49 human/ or normal human/ or human cell/
50 48 and 49
51 48 not 50
52 47 not 51
53 33 and 52
54 (2018$ or 2019$).em.
55 53 and 54

Appendix 4. CINAHL (EBSCO) search strategy

S1 (MH "Lung Diseases, Interstitial+")
S2 interstitial* N3 lung* N3 disease*
S3 interstitial* N3 (fibros* or pneumonitis or pneumonia or pneumopathy)
S4 alveolitis
S5 (MH "Bronchiolitis Obliterans+")
S6 bronchiolitis N3 obliterans
S7 goodpasture* N3 syndrome*
S8 granulomatosis
S9 (MH "Histiocytosis+")
S10 histiocytosis
S11 (MH "Pneumoconiosis+")
S12 pneumoconiosis or pneumokoniosis or pneumonoconiosis
S13 bagassosis
S14 pulmonary* N3 (sarcoid* or fibros*)
S15 wegener* N3 granuloma*
S16 lung* N3 purpura
S17 S1 OR S2 OR S3 OR S4 OR S5 OR S6 OR S7 OR S8 OR S9 OR S10 OR S11 OR S12 OR S13 OR S14 OR S15 OR S16
S18 (MH "Scleroderma, Systemic+")
S19 scleroderma
S20 (MH "Rheumatic Diseases+")
S21 rheumatic*
S22 S18 OR S19 OR S20 OR S21
S23 lung* or pulmonary* or respiratory*
S24 S22 AND S23
S25 S17 OR S24
S26 (MH "Physical Therapy+")
S27 (MH "Exercise")
S28 (MH "Physical Fitness")
S29 (MH "Rehabilitation")
S30 (MH "Rehabilitation, Pulmonary+")
S31 (MH "Exertion")
S32 rehabilitat* or fitness* or exercis* or physical* or train* or active* or activit* or physiotherap* or kinesiotherap* or exert*
S33 S26 OR S27 OR S28 OR S29 OR S30 OR S31 OR S32
S34 S25 AND S33
S35 (MH "Randomized Controlled Trials")
S36 (MH "Double‐Blind Studies")
S37 (MH "Random Assignment")
S38 (MH "Placebos")
S39 placebo*
S40 random*
S41 crossover* or cross‐over*
S42 clinical* N3 (trial* or study or studies)
S43 (single* or double* or triple*) N3 blind*
S44 S35 OR S36 OR S37 OR S38 OR S39 OR S40 OR S41 OR S42 OR S43
S45 S34 AND S44

Appendix 5. PEDro search strategy

interstitial 
scleroderma
rheumatic disease*
lung disease*

(all limited to clinical trial)
ILD (not limited to trial)

Data and analyses

Comparison 1. Pulmonary rehabilitation versus no pulmonary rehabilitation.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1.1 Change in 6‐minute walk distance immediately following pulmonary rehabilitation. Mean change from baseline, metres	13		Mean Difference (IV, Fixed, 95% CI)	Subtotals only
1.1.1 All participants	13	585	Mean Difference (IV, Fixed, 95% CI)	40.07 [32.70, 47.44]
1.1.2 Idiopathic pulmonary fibrosis only	8	278	Mean Difference (IV, Fixed, 95% CI)	37.25 [26.16, 48.33]
1.1.3 Severe lung disease	2	84	Mean Difference (IV, Fixed, 95% CI)	15.37 [‐10.70, 41.43]
1.1.4 Desaturators	2	103	Mean Difference (IV, Fixed, 95% CI)	20.12 [‐2.62, 42.87]
1.2 Change in 6‐minute walk test at long‐term follow‐up. Mean change from baseline, metres	5		Mean Difference (IV, Fixed, 95% CI)	Subtotals only
1.2.1 All participants	5	297	Mean Difference (IV, Fixed, 95% CI)	32.43 [15.58, 49.28]
1.2.2 Idiopathic pulmonary fibrosis only	3	123	Mean Difference (IV, Fixed, 95% CI)	1.64 [‐24.89, 28.17]
1.2.3 Severe lung disease	2	84	Mean Difference (IV, Fixed, 95% CI)	4.20 [‐28.99, 37.40]
1.2.4 Desaturators	2	103	Mean Difference (IV, Fixed, 95% CI)	1.76 [‐28.95, 32.47]
1.3 Change in peak work rate immediately following pulmonary rehabilitation, watts	4	274	Mean Difference (IV, Fixed, 95% CI)	7.55 [5.66, 9.44]
1.3.1 All participants	4	159	Mean Difference (IV, Fixed, 95% CI)	9.04 [6.07, 12.00]
1.3.2 Idiopathic pulmonary fibrosis only	2	62	Mean Difference (IV, Fixed, 95% CI)	9.94 [6.39, 13.49]
1.3.3 Severe lung disease	1	23	Mean Difference (IV, Fixed, 95% CI)	2.10 [‐2.29, 6.49]
1.3.4 Desaturators	1	30	Mean Difference (IV, Fixed, 95% CI)	5.40 [0.07, 10.73]
1.4 Change in VO 2 peak immediately following pulmonary rehabilitation, mL/kg/minute	3		Mean Difference (IV, Fixed, 95% CI)	Subtotals only
1.4.1 All participants	3	94	Mean Difference (IV, Fixed, 95% CI)	1.28 [0.51, 2.05]
1.4.2 Idiopathic pulmonary fibrosis only	2	62	Mean Difference (IV, Fixed, 95% CI)	1.45 [0.51, 2.40]
1.4.3 Severe lung disease	1	18	Mean Difference (IV, Fixed, 95% CI)	‐0.03 [‐1.36, 1.30]
1.4.4 Desaturators	1	27	Mean Difference (IV, Fixed, 95% CI)	0.84 [‐0.31, 1.99]
1.5 Change in maximum ventilation (Ve max ) immediately following pulmonary rehabilitation, L/minute	3		Mean Difference (IV, Fixed, 95% CI)	Subtotals only
1.5.1 All participants	3	94	Mean Difference (IV, Fixed, 95% CI)	7.21 [4.10, 10.32]
1.5.2 Idiopathic pulmonary fibrosis only	2	62	Mean Difference (IV, Fixed, 95% CI)	9.80 [6.06, 13.53]
1.5.3 Severe lung disease	1	20	Mean Difference (IV, Fixed, 95% CI)	4.16 [‐3.34, 11.66]
1.5.4 Desaturators	1	27	Mean Difference (IV, Fixed, 95% CI)	6.95 [0.03, 13.87]
1.6 Change in maximum heart rate immediately following pulmonary rehabilitation, beats/minute	3		Mean Difference (IV, Fixed, 95% CI)	Subtotals only
1.6.1 All participants	3	94	Mean Difference (IV, Fixed, 95% CI)	‐0.77 [‐4.25, 2.72]
1.6.2 Idiopathic pulmonary fibrosis only	2	62	Mean Difference (IV, Fixed, 95% CI)	‐0.38 [‐3.78, 3.01]
1.6.3 Severe lung disease	1	20	Mean Difference (IV, Fixed, 95% CI)	‐5.38 [‐11.46, 0.70]
1.6.4 Desaturators	1	27	Mean Difference (IV, Fixed, 95% CI)	‐0.45 [‐6.07, 5.17]
1.7 Change in dyspnoea score immediately following pulmonary rehabilitation	7		Std. Mean Difference (IV, Fixed, 95% CI)	Subtotals only
1.7.1 All participants	7	348	Std. Mean Difference (IV, Fixed, 95% CI)	‐0.36 [‐0.58, ‐0.14]
1.7.2 Idiopathic pulmonary fibrosis only	4	155	Std. Mean Difference (IV, Fixed, 95% CI)	‐0.41 [‐0.74, ‐0.09]
1.7.3 Severe lung disease	2	84	Std. Mean Difference (IV, Fixed, 95% CI)	‐0.25 [‐0.68, 0.19]
1.7.4 Desaturators	2	103	Std. Mean Difference (IV, Fixed, 95% CI)	‐0.39 [‐0.79, 0.00]
1.8 Change in dyspnoea score at long‐term follow‐up	6		Mean Difference (IV, Fixed, 95% CI)	Subtotals only
1.8.1 All participants	6	335	Mean Difference (IV, Fixed, 95% CI)	‐0.29 [‐0.49, ‐0.10]
1.8.2 Idiopathic pulmonary fibrosis only	3	123	Mean Difference (IV, Fixed, 95% CI)	‐0.38 [‐0.72, ‐0.05]
1.8.3 Severe lung disease	2	84	Mean Difference (IV, Fixed, 95% CI)	0.14 [‐0.36, 0.63]
1.8.4 Desaturators	2	103	Mean Difference (IV, Fixed, 95% CI)	‐0.03 [‐0.42, 0.35]
1.9 Change in quality of life (St George's Respiratory Questionnaire (SGRQ) Symptoms) immediately following pulmonary rehabilitation	7	588	Mean Difference (IV, Fixed, 95% CI)	‐13.68 [‐16.59, ‐10.77]
1.9.1 All participants	7	312	Mean Difference (IV, Fixed, 95% CI)	‐15.58 [‐19.54, ‐11.62]
1.9.2 Idiopathic pulmonary fibrosis only	4	142	Mean Difference (IV, Fixed, 95% CI)	‐13.92 [‐19.68, ‐8.17]
1.9.3 Severe lung disease	1	61	Mean Difference (IV, Fixed, 95% CI)	‐9.20 [‐19.17, 0.77]
1.9.4 Desaturators	1	73	Mean Difference (IV, Fixed, 95% CI)	‐7.70 [‐16.17, 0.77]
1.10 Change in quality of life (SGRQ Activity) immediately following pulmonary rehabilitation	7	588	Mean Difference (IV, Fixed, 95% CI)	‐2.30 [‐3.46, ‐1.14]
1.10.1 All participants	7	312	Mean Difference (IV, Fixed, 95% CI)	‐2.47 [‐4.11, ‐0.83]
1.10.2 Idiopathic pulmonary fibrosis only	4	142	Mean Difference (IV, Fixed, 95% CI)	‐1.71 [‐3.44, 0.01]
1.10.3 Severe lung disease	1	61	Mean Difference (IV, Fixed, 95% CI)	‐3.60 [‐11.51, 4.31]
1.10.4 Desaturators	1	73	Mean Difference (IV, Fixed, 95% CI)	‐8.20 [‐15.55, ‐0.85]
1.11 Change in quality of life (SGRQ Impact) immediately following pulmonary rehabilitation	7	588	Mean Difference (IV, Fixed, 95% CI)	‐8.66 [‐10.37, ‐6.94]
1.11.1 All participants	7	312	Mean Difference (IV, Fixed, 95% CI)	‐8.81 [‐11.17, ‐6.46]
1.11.2 Idiopathic pulmonary fibrosis only	4	142	Mean Difference (IV, Fixed, 95% CI)	‐8.94 [‐11.76, ‐6.13]
1.11.3 Severe lung disease	1	61	Mean Difference (IV, Fixed, 95% CI)	‐8.00 [‐16.18, 0.18]
1.11.4 Desaturators	1	73	Mean Difference (IV, Fixed, 95% CI)	‐5.90 [‐12.99, 1.19]
1.12 Change in quality of life (SGRQ Total) immediately following pulmonary rehabilitation	11		Mean Difference (IV, Fixed, 95% CI)	Subtotals only
1.12.1 All participants	11	478	Mean Difference (IV, Fixed, 95% CI)	‐9.29 [‐11.06, ‐7.52]
1.12.2 Idiopathic pulmonary fibrosis only	6	194	Mean Difference (IV, Fixed, 95% CI)	‐7.91 [‐10.55, ‐5.26]
1.12.3 Severe lung disease	1	61	Mean Difference (IV, Fixed, 95% CI)	‐6.40 [‐12.79, ‐0.01]
1.12.4 Desaturators	1	73	Mean Difference (IV, Fixed, 95% CI)	‐6.00 [‐11.56, ‐0.44]
1.13 Change in quality of life (SGRQ Symptoms) at long‐term follow‐up	4	463	Mean Difference (IV, Fixed, 95% CI)	‐9.14 [‐12.91, ‐5.37]
1.13.1 All participants	4	240	Mean Difference (IV, Fixed, 95% CI)	‐11.31 [‐16.58, ‐6.03]
1.13.2 Idiopathic pulmonary fibrosis only	2	89	Mean Difference (IV, Fixed, 95% CI)	‐6.84 [‐15.77, 2.10]
1.13.3 Severe lung disease	1	61	Mean Difference (IV, Fixed, 95% CI)	‐12.00 [‐22.41, ‐1.59]
1.13.4 Desaturators	1	73	Mean Difference (IV, Fixed, 95% CI)	‐3.20 [‐12.08, 5.68]
1.14 Change in quality of life (SGRQ Activity) at long‐term follow‐up	4	463	Mean Difference (IV, Fixed, 95% CI)	‐1.41 [‐2.51, ‐0.30]
1.14.1 All participants	4	240	Mean Difference (IV, Fixed, 95% CI)	‐1.54 [‐3.11, 0.02]
1.14.2 Idiopathic pulmonary fibrosis only	2	89	Mean Difference (IV, Fixed, 95% CI)	‐1.07 [‐2.70, 0.56]
1.14.3 Severe lung disease	1	61	Mean Difference (IV, Fixed, 95% CI)	‐1.80 [‐9.93, 6.33]
1.14.4 Desaturators	1	73	Mean Difference (IV, Fixed, 95% CI)	‐5.20 [‐12.82, 2.42]
1.15 Change in quality of life (SGRQ Impact) at long‐term follow‐up	4	463	Mean Difference (IV, Fixed, 95% CI)	‐3.57 [‐5.79, ‐1.35]
1.15.1 All participants	4	240	Mean Difference (IV, Fixed, 95% CI)	‐4.73 [‐7.76, ‐1.69]
1.15.2 Idiopathic pulmonary fibrosis only	2	89	Mean Difference (IV, Fixed, 95% CI)	‐4.59 [‐8.60, ‐0.57]
1.15.3 Severe lung disease	1	61	Mean Difference (IV, Fixed, 95% CI)	1.40 [‐7.05, 9.85]
1.15.4 Desaturators	1	73	Mean Difference (IV, Fixed, 95% CI)	2.90 [‐4.45, 10.25]
1.16 Change in quality of life (SGRQ Total) at long‐term follow‐up	4	463	Mean Difference (IV, Fixed, 95% CI)	‐3.60 [‐5.66, ‐1.55]
1.16.1 All participants	4	240	Mean Difference (IV, Fixed, 95% CI)	‐4.93 [‐7.81, ‐2.06]
1.16.2 Idiopathic pulmonary fibrosis only	2	89	Mean Difference (IV, Fixed, 95% CI)	‐3.45 [‐7.43, 0.52]
1.16.3 Severe lung disease	1	61	Mean Difference (IV, Fixed, 95% CI)	‐1.90 [‐8.57, 4.77]
1.16.4 Desaturators	1	73	Mean Difference (IV, Fixed, 95% CI)	0.10 [‐5.62, 5.82]
1.17 Change in quality of life (Chronic Respiratory Disease Questionnaire (CRQ) Dyspnoea) immediately following pulmonary rehabilitation	5	677	Mean Difference (IV, Fixed, 95% CI)	0.72 [0.55, 0.88]
1.17.1 All participants	5	321	Mean Difference (IV, Fixed, 95% CI)	0.68 [0.42, 0.93]
1.17.2 Idiopathic pulmona ry fibrosis only	3	169	Mean Difference (IV, Fixed, 95% CI)	0.81 [0.49, 1.14]
1.17.3 Severe lung disease	2	84	Mean Difference (IV, Fixed, 95% CI)	0.68 [0.21, 1.15]
1.17.4 Desaturators	2	103	Mean Difference (IV, Fixed, 95% CI)	0.69 [0.30, 1.08]
1.18 Change in quality of life (CRQ Fatigue) immediately following pulmonary rehabilitation.	5	677	Mean Difference (IV, Fixed, 95% CI)	0.66 [0.49, 0.82]
1.18.1 All participants	5	321	Mean Difference (IV, Fixed, 95% CI)	0.66 [0.43, 0.90]
1.18.2 Idiopathic pulmonary fibrosis only	3	169	Mean Difference (IV, Fixed, 95% CI)	0.67 [0.36, 0.98]
1.18.3 Severe lung disease	2	84	Mean Difference (IV, Fixed, 95% CI)	0.65 [0.17, 1.13]
1.18.4 Desaturators	2	103	Mean Difference (IV, Fixed, 95% CI)	0.60 [0.15, 1.06]
1.19 Change in quality of life (CRQ Emotion) immediately following pulmonary rehabilitation	5	677	Mean Difference (IV, Fixed, 95% CI)	0.55 [0.40, 0.70]
1.19.1 All participants	5	321	Mean Difference (IV, Fixed, 95% CI)	0.63 [0.42, 0.84]
1.19.2 Idiopathic pulmonary fibrosis only	3	169	Mean Difference (IV, Fixed, 95% CI)	0.64 [0.33, 0.95]
1.19.3 Severe lung disease	2	84	Mean Difference (IV, Fixed, 95% CI)	0.30 [‐0.15, 0.75]
1.19.4 Desaturators	2	103	Mean Difference (IV, Fixed, 95% CI)	0.30 [‐0.09, 0.70]
1.20 Change in quality of life (CRQ Mastery) immediately following pulmonary rehabilitation	5	677	Mean Difference (IV, Fixed, 95% CI)	0.62 [0.46, 0.79]
1.20.1 All participants	5	321	Mean Difference (IV, Fixed, 95% CI)	0.67 [0.44, 0.90]
1.20.2 Idiopathic pulmonary fibrosis only	3	169	Mean Difference (IV, Fixed, 95% CI)	0.63 [0.33, 0.94]
1.20.3 Severe lung disease	2	84	Mean Difference (IV, Fixed, 95% CI)	0.52 [‐0.04, 1.07]
1.20.4 Desaturators	2	103	Mean Difference (IV, Fixed, 95% CI)	0.49 [0.02, 0.96]
1.21 Change in quality of life (CRQ Dyspnoea) at long‐term follow‐up	4	551	Mean Difference (IV, Fixed, 95% CI)	0.25 [0.07, 0.44]
1.21.1 All participants	4	269	Mean Difference (IV, Fixed, 95% CI)	0.42 [0.17, 0.68]
1.21.2 Idiopathic pulmonary fibrosis only	2	95	Mean Difference (IV, Fixed, 95% CI)	0.23 [‐0.26, 0.72]
1.21.3 Severe lung disease	2	84	Mean Difference (IV, Fixed, 95% CI)	0.08 [‐0.42, 0.58]
1.21.4 Desaturators	2	103	Mean Difference (IV, Fixed, 95% CI)	‐0.10 [‐0.54, 0.34]
1.22 Change in quality of life (CRQ Fatigue) at long‐term follow‐up	4	551	Mean Difference (IV, Fixed, 95% CI)	0.26 [0.05, 0.48]
1.22.1 All participants	4	269	Mean Difference (IV, Fixed, 95% CI)	0.40 [0.09, 0.70]
1.22.2 Idiopathic pulmonary fibrosis only	2	95	Mean Difference (IV, Fixed, 95% CI)	0.31 [‐0.20, 0.83]
1.22.3 Severe lung disease	2	84	Mean Difference (IV, Fixed, 95% CI)	0.16 [‐0.37, 0.69]
1.22.4 Desaturators	2	103	Mean Difference (IV, Fixed, 95% CI)	‐0.04 [‐0.54, 0.46]
1.23 Change in quality of life (CRQ Emotion) at long‐term follow‐up	4	551	Mean Difference (IV, Fixed, 95% CI)	0.32 [0.13, 0.50]
1.23.1 All participants	4	269	Mean Difference (IV, Fixed, 95% CI)	0.51 [0.26, 0.77]
1.23.2 Idiopathic pulmonary fibrosis only	2	95	Mean Difference (IV, Fixed, 95% CI)	0.23 [‐0.23, 0.70]
1.23.3 Severe lung disease	2	84	Mean Difference (IV, Fixed, 95% CI)	0.16 [‐0.29, 0.61]
1.23.4 Desaturators	2	103	Mean Difference (IV, Fixed, 95% CI)	‐0.00 [‐0.42, 0.42]
1.24 Change in quality of life (CRQ Mastery) at long‐term follow‐up	4	551	Mean Difference (IV, Fixed, 95% CI)	0.27 [0.05, 0.50]
1.24.1 All participants	4	269	Mean Difference (IV, Fixed, 95% CI)	0.47 [0.17, 0.78]
1.24.2 Idiopathic pulmonary fibrosis only	2	95	Mean Difference (IV, Fixed, 95% CI)	0.10 [‐0.47, 0.67]
1.24.3 Severe lung disease	2	84	Mean Difference (IV, Fixed, 95% CI)	0.23 [‐0.37, 0.83]
1.24.4 Desaturators	2	103	Mean Difference (IV, Fixed, 95% CI)	‐0.15 [‐0.68, 0.38]
1.25 Long‐term survival	4		Odds Ratio (M‐H, Fixed, 95% CI)	Subtotals only
1.25.1 All participants	4	291	Odds Ratio (M‐H, Fixed, 95% CI)	0.40 [0.14, 1.12]
1.25.2 Idiopathic pulmonary fibrosis only	3	127	Odds Ratio (M‐H, Fixed, 95% CI)	0.32 [0.08, 1.19]
1.25.3 Severe lung disease	2	84	Odds Ratio (M‐H, Fixed, 95% CI)	0.53 [0.14, 2.05]
1.25.4 Desaturators	2	103	Odds Ratio (M‐H, Fixed, 95% CI)	0.59 [0.15, 2.35]

1.14. Analysis.

Comparison 1: Pulmonary rehabilitation versus no pulmonary rehabilitation, Outcome 14: Change in quality of life (SGRQ Activity) at long‐term follow‐up

Characteristics of studies

Characteristics of included studies [ordered by study ID]

Baradzina 2005.

Study characteristics
Methods	Randomised controlled trial
Participants	Sarcoidosis: n = 65 Pulmonary rehabilitation group: n = 30, 12 men, mean age 38 years Control group: n = 35, 14 men, mean age 36 years
Interventions	Pulmonary rehabilitation group: 5‐week multi‐disciplinary exercise programme. Included exercise training (5 times weekly for 40 minutes), physiotherapy, education, nutritional advice and stress management Control group: not specified
Outcomes	Walking test: type unspecified Dyspnoea: measure unspecified HRQoL: WHO questionnaire All measures were obtained before and after intervention period.
Notes	Abstract
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Not specified.
Allocation concealment (selection bias)	Unclear risk	Not specified.
Blinding of participants and personnel (performance bias) All outcomes	High risk	Due to the nature of the intervention, it was impossible for participants or personnel to be blinded.
Blinding of outcome assessment (detection bias) All outcomes	Unclear risk	Not specified.
Incomplete outcome data (attrition bias) All outcomes	Unclear risk	Not specified.
Selective reporting (reporting bias)	Unclear risk	Abstract only, insufficient details provided.
Other bias	Unclear risk	Abstract only, insufficient details provided.

Dale 2014.

Study characteristics
Methods	Randomised controlled trial
Participants	Dust‐related ILD: n = 11, 100% men Pulmonary rehabilitation group: n = 6, mean age 70 (SD 7) years, mean FVC 86 (SD 23) % predicted, mean TLCO 54 (SD 15) % predicted Control group: n = 4, mean age 72 (SD 6) years, mean FVC 86 (SD 18) % predicted, mean TLCO 57 (SD 13) % predicted
Interventions	Pulmonary rehabilitation group: 8‐week outpatient exercise programme, twice‐weekly supervised sessions consisting of 30 minutes of endurance exercise (cycling and walking). Initial intensity was at 80% of walking speed on initial 6MWT for walking and 60% of peak work achieved on CPET for cycling. Exercise training was progressed according to protocol. Control group: usual medical management
Outcomes	6MWT CPET Endurance cycle test SGRQ CRQ mMRC Dyspnoea Scale Physical activity (SenseWear Armband) Measured before and after intervention period. 6MWT and questionnaires repeated at 6‐month follow‐up.
Notes	Supported by Workers' Compensation Dust Diseases Board of New South Wales, Australia.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Computer‐generated randomisation sequence.
Allocation concealment (selection bias)	Low risk	Method of concealment not described within paper. Authors confirmed allocation was concealed and was marked down in the Pedro score (search.pedro.org.au/search-results/record-detail/41222).
Blinding of participants and personnel (performance bias) All outcomes	High risk	Due to the nature of the intervention, it was impossible for participants or personnel to be blinded.
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Data collector blinded to treatment allocation.
Incomplete outcome data (attrition bias) All outcomes	Low risk	Intention‐to‐treat analysis.
Selective reporting (reporting bias)	Low risk	All data available at all time points.
Other bias	Low risk	Study appeared free of other sources of bias.

De Las Heras 2019.

Study characteristics
Methods	Randomised controlled trial
Participants	IPF: n = 29, 21 men Pulmonary rehabilitation group: n = 15, mean age 70 (SD 9) years, mean FVC 77 (SD 16) % predicted, mean TLCO 47 (SD 11) % predicted Control group: n = 14, mean age 72 (SD 8) years, mean FVC 91 (SD 17) % predicted, mean TLCO 55 (SD 14) % predicted
Interventions	Pulmonary rehabilitation group: 12‐week tele‐rehabilitation exercise programme consisting of video and chat sessions with a physiotherapist (once weekly for 1 month, once every 2 months for second month and once a month for the remainder of the trial) and home‐based workout sessions with a virtual physiotherapist agent. Workout sessions involved 10–20 minutes of exercise daily using elastics, weights and a fitness step. Control group: usual control programme for people with IPF of outpatient visits approximately once every 3 months
Outcomes	6MWT King's Brief Interstitial Lung Disease questionnaire SGRQ‐I General Anxiety Disorder Score 7‐day physical activity (steps) Measured before and after intervention period, and at 3‐ and 6‐month follow‐up
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Randomisation allocated participants to intervention or control group; performed electronically.
Allocation concealment (selection bias)	Unclear risk	Not described.
Blinding of participants and personnel (performance bias) All outcomes	High risk	Due to the nature of the physical intervention, it was not possible for participants or personnel to be blinded.
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Data collector blinded to treatment allocation.
Incomplete outcome data (attrition bias) All outcomes	High risk	4 participants in the exercise group and 4 in the control group did not complete the intervention and were not included in the analysis.
Selective reporting (reporting bias)	High risk	Not all outcome measures from the clinical trial registry were reported in the abstract. Not all domains for the SGRQ were reported in the results.
Other bias	Low risk	Study appeared free of other sources of bias.

Dowman 2017.

Study characteristics
Methods	Randomised controlled trial, stratified for IPF, dust‐related ILD and CTD ILD
Participants	ILD: n = 142, including IPF: n = 61, asbestosis: n = 22, CTD ILD: n = 23, 87 men Pulmonary rehabilitation group: n = 74, mean age 69 (SD 11) years, mean FVC 76 (SD 18) % predicted, mean TLCO 50 (SD 16) % predicted Control group: n = 68, mean age 70 (SD 11) years, mean FVC 75 (SD 20) % predicted, mean TLCO 48 (SD 14) % predicted
Interventions	Pulmonary rehabilitation group: 8‐week outpatient exercise programme, twice‐weekly supervised sessions consisting of 30 minutes of endurance exercise (cycling and walking). Initial intensity was at 80% of walking speed on initial 6MWT for walking and 70% of peak work rate estimated from the 6MWT for cycling. Exercise training was progressed according to protocol. Upper limb endurance and functional strength training for lower limbs also performed. Supplemental oxygen provided for SpO2 < 85%. Unsupervised home exercise programme prescribed 3 times per week. Group education sessions were offered twice weekly and included understanding lung disease, medications, home oxygen therapy, self‐management, managing breathlessness, exercise and physical activity, stress and anxiety, nutrition, swallowing and airway clearance Control group: weekly telephone calls for general health advice and support
Outcomes	6MWT Knee extensor and elbow flexor strength via hand‐held dynamometry SGRQ‐I SGRQ UCSD SOBQ HADS mMRC Dyspnoea Scale Measured before and after intervention period and at 6‐month follow‐up
Notes	Supported by Pulmonary Fibrosis Foundation/American Thoracic Society Foundation, Institute for Breathing and Sleep and Eirene Lucas Foundation.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Computer‐generated random number sequence.
Allocation concealment (selection bias)	Low risk	Sealed opaque envelopes.
Blinding of participants and personnel (performance bias) All outcomes	High risk	Due to the nature of the intervention, it was impossible for participants or personnel to be blinded.
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Data collector blinded to treatment allocation.
Incomplete outcome data (attrition bias) All outcomes	Low risk	Intention‐to‐treat analysis, linear mixed models.
Selective reporting (reporting bias)	Low risk	All outcome measures identified in the clinical trials registry (anzctr.org.au) and previously published protocol paper were reported at the same time points.
Other bias	Low risk	Study appeared free of other sources of bias.

Gaunaurd 2014.

Study characteristics
Methods	Randomised controlled trial
Participants	IPF: n = 21 Pulmonary rehabilitation group: n = 11, mean age 71 (SD 6) years, mean FVC 60 (SD 11) % predicted, mean TLCO 44 (SD 11) % predicted Control group: n = 11, mean age 66 (SD 7) years, mean FVC 61 (SD 14) % predicted, mean TLCO 43 (SD 11) % predicted
Interventions	Pulmonary rehabilitation group: 12‐week outpatient exercise programme, twice‐weekly supervised sessions consisting of 30 minutes of endurance training (20 minutes of treadmill walking and 10 minutes of semi‐recumbent cycling) with an initial intensity of 80% HRmax. Strength training for upper and lower limbs using therabands for 15–30 minutes, and flexibility exercise for upper and lower body performed for 15 minutes. Supplemental oxygen was provided to maintain SpO2 > 88%. 10 education lectures were provided including medication use, breathing techniques, exercise training, nutrition, pulmonary physiology and psychological coping mechanisms Control group: no structured exercise. Handouts from education lectures were provided to the control participants
Outcomes	SGRQ‐I IPAQ Dyspnoea (Borg Index) Measured before and after intervention and at 3‐month follow‐up
Notes	Unpublished quality‐of‐life data from study by Jackson 2014, published separately
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Assigned by block randomisation according to random number programme.
Allocation concealment (selection bias)	Unclear risk	Not reported.
Blinding of participants and personnel (performance bias) All outcomes	High risk	Due to the nature of the intervention, it was impossible for participants or personnel to be blinded.
Blinding of outcome assessment (detection bias) All outcomes	High risk	Data collector not blinded to treatment allocation.
Incomplete outcome data (attrition bias) All outcomes	High risk	3 participants in exercise group and 1 in control group did not complete the intervention period; data were not included in analysis.
Selective reporting (reporting bias)	High risk	Borg Dyspnoea Index not reported. Not all domains for SGRQ‐I were reported in the results. Additional data sought from authors.
Other bias	Low risk	Study appeared free of other sources of bias.

He 2016.

Study characteristics
Methods	Randomised controlled trial
Participants	IPF n = 30, 17 men Pulmonary rehabilitation group: n = 15, mean age 65 (SD 7) years, mean FVC 65 (SD 11) % predicted, mean TLCO 43 (SD 10) % predicted Control group: n = 15, mean age 65 (SD 8) years, mean FVC 63 (SD 9) % predicted, mean TLCO 43 (SD 10) % predicted
Interventions	Pulmonary rehabilitation group: 12‐week outpatient exercise programme, 3–5 days per week consisting of 30–40 minutes of endurance exercise (cycling) Control group: prednisone‐ and azathioprine‐based anti‐inflammatory treatment and symptomatic supportive therapy as necessary
Outcomes	6MWT ATAQ‐IPF Questionnaire Measured before and after intervention period
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Not specified.
Allocation concealment (selection bias)	Unclear risk	Not specified.
Blinding of participants and personnel (performance bias) All outcomes	High risk	Due to the nature of the intervention, it was impossible for participants or personnel to be blinded.
Blinding of outcome assessment (detection bias) All outcomes	Unclear risk	Not specified.
Incomplete outcome data (attrition bias) All outcomes	Low risk	All 30 participants' results were included in the analysis. No mention of withdrawals or dropouts.
Selective reporting (reporting bias)	Low risk	All data available at all time points.
Other bias	Low risk	Study appeared free of other sources of bias.

Holland 2008.

Study characteristics
Methods	Randomised controlled trial Stratified for IPF
Participants	ILD: n = 57, including IPF: n = 34, 31 men Pulmonary rehabilitation group: n = 30, mean age 70 (SD 8) years, mean TLCO 50 (SD 19) % predicted Control group: n = 27, mean age 67 (SD 13) years, mean TLCO 49 (SD 18) % predicted
Interventions	Pulmonary rehabilitation group: 8‐week outpatient exercise programme, twice‐weekly supervised sessions consisting of 30 minutes of endurance exercise (cycling and walking) with initial intensity at 80% of walking speed on initial 6MWT and progressed according to protocol. Upper limb endurance and functional strength training for lower limbs also performed. Supplemental oxygen provided for SpO2 < 85%. Unsupervised home exercise programme prescribed 3 times per week Control group: weekly telephone calls for general health advice and support
Outcomes	6MWT CPET CRQ mMRC Dyspnoea Scale Measured before and after intervention period. 6MWT and questionnaires repeated at 6‐month follow‐up
Notes	Supported by the Victorian Tuberculosis and Lung Association.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Computer‐generated random number sequence.
Allocation concealment (selection bias)	Low risk	Central location, sealed opaque envelope.
Blinding of participants and personnel (performance bias) All outcomes	High risk	Due to the nature of the intervention, it was impossible for participants or personnel to be blinded.
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Data collector blinded to treatment allocation.
Incomplete outcome data (attrition bias) All outcomes	Low risk	Intention‐to‐treat analysis, last observation carried forward.
Selective reporting (reporting bias)	Low risk	All data available at all time points.
Other bias	Low risk	Study appeared free of other sources of bias.

Jackson 2014.

Study characteristics
Methods	Randomised controlled trial
Participants	IPF: n = 21 Pulmonary rehabilitation group: n = 11, mean age 71 (SD 6) years, mean FVC 60 (SD 11) % predicted, mean TLCO 44 (SD 11) % predicted Control group: n = 10, mean age 66 (SD 7) years, mean FVC 61 (SD 14) % predicted, mean TLCO 43 (SD 11) % predicted
Interventions	Pulmonary rehabilitation group: 12‐week outpatient exercise programme, twice‐weekly supervised sessions consisting of 30 minutes of endurance training (20 minutes of treadmill walking and 10 minutes of semi‐recumbent cycling) with an initial intensity of 80% HRmax. Strength training for upper and lower limbs using therabands for 15–30 minutes, and flexibility exercise for upper and lower body performed for 15 minutes. Supplemental oxygen provided to maintain SpO2 > 88%. Education component included PowerPoint presentations and handouts for 15 minutes a session (bi‐weekly) Control group: no structured exercise
Outcomes	6MWT Constant work rate cycle test Treadmill exercise (METs) Maximum inspiratory and expiratory pressures Dyspnoea (Borg Index) F2‐isoprostanes and lactate plasma levels Measured before and after intervention period
Notes	Supported by a Merit Review Award from the Veteran Affairs Research Service.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Not specified.
Allocation concealment (selection bias)	Low risk	Independent researcher provided group allocation.
Blinding of participants and personnel (performance bias) All outcomes	High risk	Due to the nature of the intervention, it was impossible for participants or personnel to be blinded.
Blinding of outcome assessment (detection bias) All outcomes	High risk	Data collector not blinded to treatment allocation.
Incomplete outcome data (attrition bias) All outcomes	High risk	3 participants in the exercise group and 1 in the control group did not complete the intervention period; data were not included in the analysis.
Selective reporting (reporting bias)	Low risk	All data available at all time points.
Other bias	Low risk	Study appeared free of other sources of bias.

Jarosch 2020.

Study characteristics
Methods	Randomised controlled trial
Participants	IPF: n = 545433 Pulmonary rehabilitation group: n = 36, mean age 68 (SD 9) years, mean FVC 74 (SD 19) % predicted, TLCO mean 44 (SD 15) % predicted Control group: n = 18, mean age 65 (SD 10) years, mean FVC 72 (SD 20) % predicted, TLCO mean 37 (SD 19) % predicted
Interventions	Pulmonary rehabilitation group: 3 weeks of inpatient pulmonary rehabilitation, performed 5–6 days per week consisting of endurance or interval cycle training at 60–100% peak work rate on CPET and resistance training for major muscle groups (3 sets of 15–20 repetitions). Psychological support, breathing therapy, education such as disease management, physical activity, nutritional counselling and motivation were also provided Control group: usual care
Outcomes	6MWT SF‐36 HRQoL Questionnaire CRQ HADS Physical activity (SenseWear Armband) Data measured before and after intervention and at 3‐month follow‐up
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Computer‐based randomisation performed with a 1:2 ratio (usual care: pulmonary rehabilitation) using random permutations.
Allocation concealment (selection bias)	Low risk	Stated in article on pg 4 that the group allocation sequence was concealed.
Blinding of participants and personnel (performance bias) All outcomes	High risk	Due to the nature of the intervention, it was impossible for participants or personnel to be blinded.
Blinding of outcome assessment (detection bias) All outcomes	High risk	Data collector not blinded to treatment allocation.
Incomplete outcome data (attrition bias) All outcomes	High risk	2 participants in the pulmonary rehabilitation group and 1 in the control group did not complete the study; data were not included in the analysis.
Selective reporting (reporting bias)	Low risk	All data available at all time points.
Other bias	Low risk	Study appeared free of other sources of bias.

Ku 2017.

Study characteristics
Methods	Randomised controlled trial
Participants	ILD: n = 40, 16 men Pulmonary rehabilitation group: n = 20, mean age 59 (SD 10) years, mean FVC 52 (SD 14) % predicted Control group: n = 20, mean age 62 (SD 14) years, mean FVC 59 (SD 14) % predicted
Interventions	Pulmonary rehabilitation group: 8‐week outpatient exercise programme, twice‐weekly supervised sessions consisting of endurance exercise (cycling and walking) and strength training. Supplemental oxygen provided to maintain SpO2 > 90% Unsupervised home exercise programme prescribed 2 times per week. Education provided on ILD, drugs used in the treatment, behavioural modification and non‐pharmacological treatment modalities Control group: conventional treatment alone
Outcomes	6MWT SGRQ mMRC Dyspnoea Scale Measured before and after intervention
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Randomised by permuted block randomisation.
Allocation concealment (selection bias)	Unclear risk	Not reported.
Blinding of participants and personnel (performance bias) All outcomes	High risk	Due to the nature of the intervention, it was impossible for participants or personnel to be blinded.
Blinding of outcome assessment (detection bias) All outcomes	High risk	Data collector not blinded to treatment allocation.
Incomplete outcome data (attrition bias) All outcomes	Low risk	All 40 participants completed intervention and all their results were analysed.
Selective reporting (reporting bias)	High risk	Incomplete reporting of SGRQ domains and mMRC Dyspnoea Scale.
Other bias	Low risk	Study appeared free of other sources of bias.

Lanza 2019.

Study characteristics
Methods	Randomised controlled trial
Participants	IPF: n = 24 Pulmonary rehabilitation group: n = 14 Control group: n = 10
Interventions	Pulmonary rehabilitation group: 3‐month outpatient exercise programme, twice‐weekly supervised sessions consisting of 90 minutes of exercise Control group: maintained its preceding normal physical activity
Outcomes	6MWT SGRQ Measured before and after intervention
Notes	Abstract
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Not specified.
Allocation concealment (selection bias)	Unclear risk	Not specified.
Blinding of participants and personnel (performance bias) All outcomes	High risk	Due to the nature of the intervention, it was impossible for participants or personnel to be blinded.
Blinding of outcome assessment (detection bias) All outcomes	Unclear risk	Not specified.
Incomplete outcome data (attrition bias) All outcomes	Unclear risk	Not specified.
Selective reporting (reporting bias)	Unclear risk	Abstract only, insufficient details provided.
Other bias	Unclear risk	Abstract only, insufficient details provided.

Mejia 2000.

Study characteristics
Methods	Randomised controlled trial
Participants	ILD: n = 22 Mean age 52 (SD 14) years, mean FVC 61 (SD 19) % predicted
Interventions	12‐week exercise programme, 3 times weekly, supervised sessions of 30–35 minutes each, interval training Pulmonary rehabilitation group: exercised at 60% of maximal power output on cycle ergometer Control group: sham exercise training at minimum workload achievable on cycle ergometer (no resistance)
Outcomes	CRQ 12‐Minute walk test Measured at baseline and 12 weeks
Notes	Abstract
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Not specified.
Allocation concealment (selection bias)	Unclear risk	Not specified.
Blinding of participants and personnel (performance bias) All outcomes	Unclear risk	Control group provided with sham training, no details on blinding provided.
Blinding of outcome assessment (detection bias) All outcomes	Unclear risk	Not specified.
Incomplete outcome data (attrition bias) All outcomes	Unclear risk	Not specified.
Selective reporting (reporting bias)	Unclear risk	Abstract only, insufficient details provided.
Other bias	Unclear risk	Abstract only, insufficient details provided.

Menon 2011.

Study characteristics
Methods	Randomised controlled trial
Participants	ILD: n = 28
Interventions	Pulmonary rehabilitation group: 8 weeks of supervised pulmonary rehabilitation Control group: standard medical care
Outcomes	6MWT Mid‐thigh cross‐sectional area on CT TLCO
Notes	Abstract
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Not specified.
Allocation concealment (selection bias)	Unclear risk	Not specified.
Blinding of participants and personnel (performance bias) All outcomes	High risk	Due to the nature of the intervention, it was impossible for participants or personnel to be blinded.
Blinding of outcome assessment (detection bias) All outcomes	Unclear risk	Not specified.
Incomplete outcome data (attrition bias) All outcomes	Unclear risk	All data on all 28 participants were reported.
Selective reporting (reporting bias)	Unclear risk	Abstract only, insufficient details provided.
Other bias	Unclear risk	Abstract only, insufficient details provided.

Naz 2018.

Study characteristics
Methods	Randomised controlled trial
Participants	Stage 3 and 4 sarcoidosis: n = 18 Pulmonary rehabilitation group: n = 9, 3 men, mean age 59 years, mean FVC 76% predicted Control group: n = 9, 3 men, mean age 51 years, mean FVC 69% predicted
Interventions	Pulmonary rehabilitation group: 12‐week outpatient exercise programme, twice‐weekly supervised sessions consisting of 30 minutes of endurance exercise (cycling and walking). Initial intensity at 80% of walking speed on initial 6MWT and 70% of peak work rate estimated from the 6MWT for cycling. Upper and lower limb strength training also performed. Supplemental oxygen provided for SpO2 < 90%. Unsupervised home exercise programme prescribed. Control group: routine medical treatment
Outcomes	6MWT SGRQ SF‐36 HRQoL Questionnaire Peripheral muscle force (Quadriceps) mMRC Dyspnoea Scale FSS HADS Measured before and after intervention
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	An individual unrelated to the study performed random allocation to the exercise and usual care groups.
Allocation concealment (selection bias)	Low risk	Sealed envelopes.
Blinding of participants and personnel (performance bias) All outcomes	High risk	Due to the nature of the intervention, it was impossible for participants or personnel to be blinded.
Blinding of outcome assessment (detection bias) All outcomes	Unclear risk	Not specified.
Incomplete outcome data (attrition bias) All outcomes	Low risk	All 18 participants completed intervention and all their results were analysed.
Selective reporting (reporting bias)	Low risk	All data available at all time points.
Other bias	Low risk	Study appeared free of other sources of bias.

Nishiyama 2008.

Study characteristics
Methods	Randomised controlled trial
Participants	IPF: n = 28 Pulmonary rehabilitation group: n = 13, 12 men, mean age 68 (SD 9) years, mean TLCO 59.4 (SD 16.7) % predicted Control group: n = 15, 9 men, mean age 65 (SD 9) years, mean TLCO 48.6 (SD 16.7) % predicted
Interventions	Pulmonary rehabilitation group: 9‐week outpatient exercise programme, twice‐weekly supervised sessions. Exercise on treadmill at 80% of walking speed on initial 6MWT, or on cycle ergometer at 80% of initial maximum workload. Strength training for limbs using elastic bands for approximately 20 minutes. Supplemental oxygen administered to achieve SpO2 > 90%. Some educational lectures were included (content unspecified) Control group: not specified
Outcomes	6MWT BDI SGRQ All measured at baseline and 10 weeks
Notes	Supported by the Japanese Ministry of Health, Labor and Welfare.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Not specified.
Allocation concealment (selection bias)	Low risk	Allocation concealed using sealed envelopes that had been prepared prior to the study.
Blinding of participants and personnel (performance bias) All outcomes	High risk	Due to the nature of the intervention, it was impossible for participants or personnel to be blinded.
Blinding of outcome assessment (detection bias) All outcomes	Unclear risk	Not specified.
Incomplete outcome data (attrition bias) All outcomes	Low risk	2 participants randomised to exercise training withdrew before baseline data collected.
Selective reporting (reporting bias)	Low risk	All data available at all time points.
Other bias	Low risk	Study appeared free of other sources of bias.

Perez Bogerd 2018.

Study characteristics
Methods	Randomised controlled trial
Participants	ILD: n = 60 Pulmonary rehabilitation group: n = 30, 22 men, mean age 64 (SD 13) years, mean TLCO 45 (SD 16) % predicted Control group: n = 30, 15 men, mean age 64 (SD 8) years, mean TLCO 41 (SD 13) % predicted
Interventions	Pulmonary rehabilitation group: 6 months of pulmonary rehabilitation with a total of 60 sessions, 3 times a week for first 3 months and twice weekly thereafter. Programme consisted of 90 minutes of exercise training (endurance and strength training) and 30 minutes rotated by session of patient education, occupational therapy, nutrition counselling or psychosocial support. Initial intensity was 75% of walking speed on 6MWT for walking and 60% of peak work rate achieved on CPET for cycling. All participants trained with supplemental oxygen Control group: usual medical care
Outcomes	6MWT Peak work rate (watts) on CPET SGRQ CRQ Peripheral muscle force (quadriceps) Hand grip muscle force mMRC Dyspnoea Scale Physical activity (SenseWear Armband) Measured before, mid‐way (3 months) and after intervention (6 months), and at 6‐month follow‐up (12 months)
Notes	Abstract in previous review, published as paper in updated review.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote: "Patients were randomly assigned to the rehabilitation or control group using sealed envelopes prepared and shuffled before the start of the study by an independent person unrelated to the study protocol." (pg 2)
Allocation concealment (selection bias)	Low risk	Sealed envelopes.
Blinding of participants and personnel (performance bias) All outcomes	High risk	Due to the nature of the intervention, it was impossible for participants or personnel to be blinded.
Blinding of outcome assessment (detection bias) All outcomes	High risk	Data collector not blinded to treatment allocation.
Incomplete outcome data (attrition bias) All outcomes	Low risk	Intention‐to‐treat analysis, linear mixed modelling.
Selective reporting (reporting bias)	Low risk	All outcomes listed in the controlled trial registry (ClinicalTrials.gov) and the paper were reported for all time points.
Other bias	Low risk	Study appeared free of other sources of bias.

Shen 2016.

Study characteristics
Methods	Randomised controlled trial
Participants	IPF: n = 31 Pulmonary rehabilitation group: n = 16 Control group: n = 15
Interventions	Pulmonary rehabilitation group: continuous exercise 3 times week for 3 months Control group: not specified
Outcomes	6MWT SGRQ
Notes	Abstract
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Not specified.
Allocation concealment (selection bias)	Unclear risk	Not specified.
Blinding of participants and personnel (performance bias) All outcomes	High risk	Due to the nature of the intervention, it was impossible for participants or personnel to be blinded.
Blinding of outcome assessment (detection bias) All outcomes	Unclear risk	Not specified.
Incomplete outcome data (attrition bias) All outcomes	Unclear risk	Not specified.
Selective reporting (reporting bias)	Unclear risk	Abstract only, insufficient details provided.
Other bias	Unclear risk	Abstract only, insufficient details provided.

Vainshelboim 2014.

Study characteristics
Methods	Randomised controlled trial
Participants	IPF: n = 32 Pulmonary rehabilitation group: n = 15, mean age 69 (SD 6) years, mean FVC 66 (SD 15) % predicted, mean TLCO 49 (SD 17) % predicted Control group: n = 17, mean age 66 (SD 9) years, mean FVC 70 (SD 17) % predicted, mean TLCO 53 (SD 12) % predicted
Interventions	Pulmonary rehabilitation group: 12‐week outpatient exercise programme, 2 × 6‐week blocks of twice‐weekly 60‐minute supervised sessions. First block consisted of 30 minutes of aerobic interval training with treadmill walking, cycling and step climbing + 10 minutes of strength training; second block consisted of longer periods (20 minutes) of continuous aerobic exercise with resistance training Control group: standard medical care
Outcomes	CPET 6MWT mMRC Dyspnoea Score SGRQ 30‐second chair stand test Pulmonary function Measured before and after intervention and at 11‐month follow‐up
Notes	Abstract in previous review, published as paper in updated review.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote: "Randomization was performed by a study coordinator uninvolved in patient assessment or treatment." (pg 379)
Allocation concealment (selection bias)	Low risk	Sealed opaque envelopes.
Blinding of participants and personnel (performance bias) All outcomes	High risk	Due to the nature of the intervention, it was impossible for participants or personnel to be blinded.
Blinding of outcome assessment (detection bias) All outcomes	High risk	Data collector not blinded to treatment allocation.
Incomplete outcome data (attrition bias) All outcomes	High risk	1 participant in the exercise group and 1 in the control group did not complete the intervention period; data were not included in the analysis.
Selective reporting (reporting bias)	Low risk	All data available at all time points.
Other bias	Low risk	Study appeared free of other sources of bias.

Wallaert 2020.

Study characteristics
Methods	Randomised controlled trial
Participants	Sarcoidosis: n = 38 Pulmonary rehabilitation group: n = 20, median age 57.5 (IQR 48–64) years, mean FVC 81 (SD 18) % predicted, mean TLCO 57 (SD 16) % predicted Control group: n = 18, median age 57.5 (IQR 49–65) years, mean FVC 81 (SD 18) % predicted, mean TLCO 63 (SD 19) % predicted
Interventions	Pulmonary rehabilitation group: 2‐month outpatient exercise programme, 3 times a week supervised sessions consisting of an individual‐ and group‐based strengthening exercises, upper/lower limb training and endurance training for minimum of 30 minutes. In addition, participants received therapeutic patient education, psychosocial support and motivational communication to facilitate health‐related behavioural changes and self‐management Control group: oral counselling to increase their physical activity at home
Outcomes	6‐minute step test mMRC Dyspnoea Score Fatigue Assessment Scale Visual Simplified Respiratory Questionnaire HADS Measured before and after intervention (pulmonary rehabilitation group only), and at 6‐month and 12‐month follow‐up
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Computer‐generated random number sequence.
Allocation concealment (selection bias)	Low risk	Sequentially numbered, sealed opaque envelope.
Blinding of participants and personnel (performance bias) All outcomes	High risk	Due to the nature of the intervention, it was impossible for participants or personnel to be blinded.
Blinding of outcome assessment (detection bias) All outcomes	High risk	Data collector not blinded to treatment allocation, open label.
Incomplete outcome data (attrition bias) All outcomes	Low risk	Intention‐to‐treat analysis, linear mixed models.
Selective reporting (reporting bias)	Low risk	All data available at all time points.
Other bias	Low risk	Study appeared free of other sources of bias.

Wewel 2005.

Study characteristics
Methods	Randomised controlled trial
Participants	ILD: n = 99 Usual interstitial pneumonia: n = 38; extrinsic allergic alveolitis: n = 8; non‐specific interstitial pneumonia: n = 30; sarcoidosis: n = 23 Pulmonary rehabilitation group: n = 49, mean age 59 years, mean TLCO 49% predicted Control group: n = 50, mean age 62 years, mean TLCO 44% predicted
Interventions	Pulmonary rehabilitation group: 6‐month home‐based walking training, twice‐daily walking for 15 minutes Control group: no scheduled walking
Outcomes	6MWT CPET Walking distance at home (pedometer) SGRQ Dyspnoea: measure unspecified Measured at baseline and 6 months
Notes	Abstract
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Not specified.
Allocation concealment (selection bias)	Unclear risk	Not specified.
Blinding of participants and personnel (performance bias) All outcomes	High risk	Due to the nature of the intervention, it was impossible for participants or personnel to be blinded.
Blinding of outcome assessment (detection bias) All outcomes	Unclear risk	Not specified.
Incomplete outcome data (attrition bias) All outcomes	Unclear risk	Not specified.
Selective reporting (reporting bias)	Unclear risk	Abstract only, insufficient details provided.
Other bias	Unclear risk	Abstract only, insufficient details provided.

Xiao 2019.

Study characteristics
Methods	Randomised controlled trial
Participants	Pneumoconiosis n = 80, 100% men Pulmonary rehabilitation group: n = 40, mean age 69 (SD 4) years Control group: n = 34, mean age 71 (SD 6) years
Interventions	Pulmonary rehabilitation group: comprehensive rehabilitation treatment with individualised exercise programme for 48 weeks consisting of whole‐body respiratory gymnastics (4 days per week, 15 minutes per day), lung function exercise (4 days per week, 20 minutes per day), upper limb dumbbell exercise (twice a day, 4 days per week), lower limb resistance bicycle exercise (4 days per week, 15 minutes per day) and walking (twice a day, 4 days per week, 15 minutes per day) Control group: routine treatment (oxygen inhalation, atomisation and other common drug therapy) and simple exercise (free movement and hospital‐led gymnastics)
Outcomes	6MWT SGRQ SF‐36 HRQoL Questionnaire HADS Measured before and after intervention
Notes	There was no information as to whether the comprehensive individualised rehabilitation programme was supervised, or where it was conducted. In addition, participants were encouraged to exercise on their own and could adjust the training intensity and time. Therefore, it was unclear how much of programme was adhered to.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Random number table used to randomly divide the pulmonary rehabilitation group and control group.
Allocation concealment (selection bias)	Unclear risk	Not specified.
Blinding of participants and personnel (performance bias) All outcomes	Unclear risk	Control group received hospital‐led gymnastics and simple exercise; therefore, possible participants could be blinded although study did not specify.
Blinding of outcome assessment (detection bias) All outcomes	Unclear risk	Not specified.
Incomplete outcome data (attrition bias) All outcomes	High risk	6 participants in the control group were lost to follow‐up; these data were not included in the analysis.
Selective reporting (reporting bias)	High risk	Data for HADS not reported.
Other bias	Unclear risk	Hospital‐led respiratory gymnastics was supervised for entire 48 weeks. In addition, it appeared this was the case for both the control and intervention groups. There was no information whether the comprehensive individualised rehabilitation programme was supervised, or where it was conducted. Participants were encouraged to exercise on their own and they could adjust the training intensity and time according to their own conditions while ensuring their own safety implying a home‐based model. This suggests the programme could have been quite varied across the individuals.

6MWT: six‐minute walk test; ATAQ‐IPF: A Tool to Assess Quality of life in IPF; BDI: Baseline Dyspnoea Index; CPET: Cardiopulmonary Exercise Test; CRQ: Chronic Respiratory Disease Questionnaire; CT: computed tomography; CTD: connective tissue disease; FSS: Fatigue Severity Scale; FVC: forced vital capacity; HRQoL: health‐related quality of life; HADS: Hospital Anxiety and Depression Scale; HRmax: maximum heart rate; ILD: interstitial lung disease; IPAQ: International Physical Activity Questionnaire; IPF: idiopathic pulmonary fibrosis; IQR: interquartile range; METs: metabolic equivalents; mMRC: Modified Medical Research Council Dyspnoea Scale; n: number of participants; SD: standard deviation; SF‐36: 36‐item Short Form; SGRQ: St George's Respiratory Questionnaire; SGRQ‐I: SGRQ‐I; St George's Respiratory Questionnaire IPF version; SpO₂: oxyhaemoglobin saturation; TLCO: transfer factor for carbon monoxide; UCSD SOBQ: University of California San Diego Shortness of Breath Questionnaire; WHO: World Health Organization.

Characteristics of excluded studies [ordered by study ID]

Study	Reason for exclusion
Arizono 2014	Not a randomised controlled trial.
Cockcroft 1981	Mixed disease group: participants had chronic obstructive pulmonary disease and coal worker's pneumoconiosis.
Cockcroft 1982	Mixed disease group: participants had chronic obstructive pulmonary disease and coal worker's pneumoconiosis.
Daltroy 1995	Participants did not have ILD.
Greening 2014	ILD represented 5% of total cohort, Intervention too short.
Igarashi 2018	Not a randomised controlled trial.
Jastrzebski 2006	Not a randomised controlled trial.
Maddali Bongi 2009	Rehabilitation programme did not qualify as pulmonary rehabilitation.
Naji 2006	Not a randomised controlled trial.
Nakazawa 2012	Not a randomised controlled trial.
Nikoletou 2016	No control group, compared 2 difference exercise training protocols.
Oh 2003	Mixed disease group: diagnoses not reported.
Ong 2001	Not a randomised controlled trial.
Sciriha 2019	Not a randomised controlled trial.
Senstrom 1996	Participants did not have ILD.
Senstrom 1997	Participants did not have ILD.
Stessel 2015	Efficacy of training modality not an intervention.
Tryfon 2003	Participants did not undergo pulmonary rehabilitation.
Yuen 2019	Rehabilitation programme did not qualify as pulmonary rehabilitation.

ILD: interstitial lung disease.

Characteristics of studies awaiting classification [ordered by study ID]

El‐Komy 2019.

Methods	Randomised controlled trial
Participants	ILD: n = 62 (50 completed study and were included in analysis) (50 completed study and were included in analysis), 18 men Pulmonary rehabilitation group: n = 25, mean age 47 (SD 13) years, mean FVC 57 (SD 2) % predicted Control group: n = 25, mean age 49 (SD 10) years, mean FVC 57 (SD 2) % predicted
Interventions	Pulmonary rehabilitation group: 8‐week outpatient exercise programme, twice‐weekly supervised sessions consisting of endurance interval exercise (cycling), with 30–180 seconds of exercise and equivalent rest intervals. Initial intensity was at 80–100% maximum heart rate. Duration was 15–20 minutes increasing to 45–60 minutes. Resistance training (2–4 sets of 6–12 repetitions) was also performed. Supplemental oxygen provided for SpO2 < 85%. Unsupervised home exercise programme prescribed 3 times per week. The pulmonary rehabilitation programme included patient health education Control group: received conventional pharmacological therapy for ILD
Outcomes	6MWT SF‐36 HRQoL Questionnaire mMRC Dyspnoea Scale Measured before and after intervention period and at 6‐month follow‐up
Notes	Accuracy of data unclear

6MWT: six‐minute walk test; FVC: forced vital capacity; HRQoL: health‐related quality of life; ILD: interstitial lung disease; mMRC: Modified Medical Research Council Dyspnoea Scale; SD: standard deviation; SF‐36: 36‐item Short Form; SpO₂: oxygen saturation.

Characteristics of ongoing studies [ordered by study ID]

Kondoh 2017.

Study name	Long‐term effect of pulmonary rehabilitation under nintedanib treatment in idiopathic pulmonary fibrosis
Methods	Randomised controlled trial
Participants	80 participants with IPF
Interventions	Pulmonary rehabilitation group: short‐term pulmonary rehabilitation programme (8 weeks of twice weekly aerobic cycle exercise training) followed by maintenance pulmonary rehabilitation programme (44 weeks of once weekly aerobic cycle exercise training) Control group: no pulmonary rehabilitation
Outcomes	6MWT SGRQ Dyspnoea‐12 Questionnaire TDI CAT Physical activity (triaxial accelerometer) Pulmonary function Rate of mortality and hospitalisation Measured before and at 12‐month follow‐up
Starting date	September 2017
Contact information	Ryo Kozu, Email: ryokozu@nagasaki‐u.ac.jp
Notes	upload.umin.ac.jp/cgi-open-bin/ctr_e/ctr_view.cgi?recptno=R000030312

6MWT: six‐minute walk distance; CAT: COPD Assessment Test; IPF: idiopathic pulmonary fibrosis; SGRQ: St George's respiratory questionnaire; TDI: Transition Dyspnoea Index.

Differences between protocol and review

We specified two subgroup analyses for this update. Subgroup analysis for exercise type could not be conducted, as we identified no trials on resistance training.

Contributions of authors

Initiated the protocol: AH.

Developed the protocol: AH and CH.

Undertook literature search for the original version: AH and CH; for the first update: AH and LD; for the current updated version: LD and AM.

Retrieved papers for the original version: AH; for the first and current updated version: LD.

Screened retrieved papers against eligibility criteria for the original version: AH and CH; for the first update: AH and LD; for the current updated version: LD and AM.

Appraised quality for the original version: AH and CH; for the first update: AH and LD; for the current updated version: LD and AM.

Extracted data for the original version: AH and CH; for the first update: AH and LD; for the current updated version: LD and AM.

Wrote to study authors for additional information for the original version: AH; for the 1st and current updated version: LD.

Entered data into Review Manager 5 for the original version: AH; for the first and current updated version: LD.

Performed analysis for the original version: AH and CH; for the first and current updated version: LD.

Wrote review for the original version: AH and CH; amended manuscript for the first and current updated version: LD; reviewed the first updated version of the manuscript: AH and CH; reviewed the current updated version of the manuscript: AH, CH and AM.

Served as guarantor of the review: AH.

Contributions of editorial team

Chris Cates (Co‐ordinating Editor) checked the data entry prior to the full write up of the review, edited the protocol; advised on methodology; approved the protocol prior to publication.

John White (Contact Editor): edited the review; advised on methodology, interpretation and content.

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; obtained translations; edited the reference and other sections of the review.

Elizabeth Stovold (Information Specialist): designed the search strategy; ran the searches; edited the search methods section.

Sources of support

Internal sources

• No sources of support supplied

External sources

• Victorian Tuberculosis and Lung Association, Australia

• National Health and Medical Research Council, Australia

  PhD stipend for Ms Dowman

Declarations of interest

AEH: none. Conducted two studies included in the review (Dowman 2017; Holland 2008).

LD: none. Conducted one study included in the review (Dowman 2017).

CJH: none. Conducted two studies included in the review (Dowman 2017; Holland 2008).

AM: none. Undertook the assessment of risk of bias for Holland 2008.

LD and AM undertook the assessment of risk of bias for all the studies except for Dowman 2017. AM undertook the assessment of risks of bias for this study.

New search for studies and content updated (conclusions changed)