Interventions in adult patients with multimorbidity in low-income and middle-income countries: protocol for a mixed-methods systematic review

Abstract

Introduction

Multimorbidity, the coexistence of two or more chronic conditions in the same individual, is a major public health problem in low-income and middle-income countries (LMICs). The use of single-disease guidelines contributes to polypharmacy, fragmented care and increased treatment burden. Health systems in LMICs are very different from those in high-income countries, and adapting interventions from one to the other may not be feasible. This review aims to systematically present the current evidence for interventions for multimorbidity in the LMIC setting.

Methods and analysis

In this mixed-methods systematic review, we will include all studies of interventions for the care of adults (>18 years of age) with multimorbidity (defined as the presence of two or more chronic illnesses in an individual) in any healthcare organisation (primary, secondary or tertiary care) in an LMIC (as defined by the World Bank), published between 2000 and March 2023. All primary study designs will be included. Studies reported in languages other than English and those describing interventions classified as ‘financial’ or ‘governance arrangement’ according to the Cochrane Effective Practice and Organisation of Care classification will be excluded. MEDLINE, PubMed, Cochrane Library, TRIP, SCOPUS and the 3ie databases will be searched. The titles will be screened by one author, and two authors will independently screen all included abstracts and full texts. A third author will resolve conflicts at every stage. Studies will be reviewed for quality of evidence using appropriate tools. Epidemiological, intervention and outcome data will be extracted and summarised. Outcomes of interest for LMICs defined by the Global Alliance for Chronic Diseases research group will be analysed. Subgroup analysis according to study types and study settings will be done.

Ethics and dissemination

No ethics approval is required for this systematic review. Results will be disseminated through publication in an open-access journal and presentation at conferences.

PROSPERO registration number

CRD42023391897.

STRENGTHS AND LIMITATIONS OF THIS STUDY.

• A systematic search for literature specific to low-income and middle-income countries (LMICs), including the search of an LMIC-specific database.

• Use of Cochrane Effective Practice and Organisation of Care predefined intervention classification.

• Reporting on outcomes of interests for LMICs, as defined by the Global Alliance for Chronic Diseases research group.

• Restricted to articles in the English language.

Introduction

The WHO defines multimorbidity as the coexistence of two or more chronic diseases in the same individual, including infectious diseases and mental health disorders.^{1 2} The global prevalence of multimorbidity is 42.4% and is projected to increase as a result of increasing life expectancy.^{3 4}

The prevalence of multimorbidity in low-income and middle-income countries (LMICs) is estimated to range from 3.2% to 90.5%, with non-communicable diseases accounting for most of it.^{5 6} The common clusters noted are cardiometabolic, respiratory, mental health and musculoskeletal.⁵ There is a wide variation in the definitions used for multimorbidity across these countries and in the diagnosis of conditions which makes it difficult to adequately quantify and describe this problem in some LMICs.²

Patients with multimorbidity may experience complex challenges, including polypharmacy, adverse drug interactions, financial burden due to their illnesses, loss of pay, delays in medical procedures and poor quality of life.^7–10 Healthcare providers and healthcare systems experience challenges due to: a lack of information and guidance on managing coexisting conditions, lack of training, reduced numbers of general physicians and general practitioners, inadequate time for patients and systemic issues with coordination of care. These barriers contribute to burnout for healthcare professionals and suboptimal patient outcomes.^11–15

Multimorbidity is a global health issue and a problem in both high-income countries (HICs) and LMICs. However, the scenario in LMICs is more complex. Many HICs have transitioned from a period where mortality is caused by pestilence, infectious diseases and famines to a period of longer life expectancy and prevalent chronic disease. In many LMICs, both periods are coexistent, reflecting a prolonged period of transition.^16–18 Healthcare systems in many LMICs face a shortage of finances, trained professionals, drugs and infrastructure, which may be further complicated by poor governance and fragmented financing.^{19 20} In addition, health systems in LMICs are arranged vertically, integrating programmes around a single disease. This leaves little space for integrated patient-centred care which is required for managing patients with multimorbidity.²⁰ Thus, multimorbidity poses a challenge at patient, healthcare worker and healthcare system levels in many LMICs.²¹

In the last two decades, there has been increasing interest in research on multimorbidity.²² There is recognition that single-disease guidelines are not always appropriate for patients with multiple chronic diseases and that they often require complex interventions and integration of healthcare services. The vast differences between the healthcare systems in LMICs and HICs make success in translating interventions between these settings unlikely.

Most models of care for multimorbidity are derived from the Chronic Care Model and suggest integrated care for multimorbidity. These include the SELFIE (Sustainable intEgrated chronic care modeLs for multi-morbidity, delivery, FInancing, and performancE) framework, developed for multimorbidity in primary care in Europe. Developed with key stakeholders including patients, professionals and policymakers, the framework has the following concepts: ‘service delivery’, ‘leadership and governance’, ‘workforce’, ‘financing’, ‘technologies and medical products’ and ‘information and research’. Each concept is stratified into the micro, meso and macro levels which represent patient-level, organisational-level and policy-level strategies for integration of care.²³ Other models of care for multimorbidity include the 3-D model developed in the UK, the TrueBlue Model in Australia and the CHRODIS framework in the UK, to name a few.^24–26 Evidence from LMICs about interventions and models of care for multimorbidity is lacking and based on studies with poor methodology. A systematic review of randomised controlled trials (RCTs) of integrated care for multimorbidity in children and adults in primary care in LMICs identified five RCTs, with only low-certainty evidence found for improvement in systolic blood pressure.²⁷ In sub-Saharan Africa, there was moderate-grade evidence for integrated care models in primary care settings for people with cardiometabolic morbidity.²⁸ A Cochrane review of the interventions for multimorbidity published in 2021 could not make strong recommendations for the care of people with multimorbidity but suggested that organisational changes that target specific problems and align with outcomes of interest specific to the patient could be successful and that more research was needed. This review included 17 RCTs, all from HICs.²⁹ Thus, there is a lack of clarity on strategies and specific interventions to manage multimorbidity in LMICs.²

It is becoming widely recognised that LMICs need to develop context-specific interventions to meet the needs of their patients with multimorbidity and those of the settings in which these interventions are delivered. The Academy of Sciences has highlighted the research gaps in multimorbidity with a focus on LMICs, identifying that there is a need for research in interventions in multimorbidity care such as in the organisation of care and in understanding patient perspectives of care.³⁰

This protocol is for a systematic review aiming to understand and synthesise evidence regarding healthcare delivery interventions for patients with multimorbidity and their implementation in LMICs using Cochrane’s Effective Practice and Organisation of Care (EPOC) taxonomy of interventions. The EPOC divides complex health systems’ interventions into four categories: ‘delivery arrangement’, ‘financial arrangement’, ‘governance arrangement’ and ‘implementation strategies’. We have chosen to focus on studies that can be categorised as ‘delivery arrangement’ or ‘implementation strategies’.

The Global Alliance of Chronic Diseases (GACD) recommends outcome measures such as healthcare utilisation, economic implications and treatment burden as important for multimorbidity interventions in LMICs in addition to those which are usually included for interventions in all settings.³¹ Our review has incorporated this recommendation with modifications.

Objective

Through this review, we aim to consolidate the current evidence for interventions in the organisation of healthcare delivery and implementation strategies that improve outcomes for patients with multimorbidity in LMICs who access primary, secondary or tertiary care facilities.

Methods and analysis

This study aims to consolidate the evidence for interventions in healthcare delivery and implementation strategies for patients with multimorbidity in LMICs, with an interest in a broad range of outcomes. We will include primary qualitative and quantitative studies which may address different outcomes about a variety of different interventions; hence, this mixed-methods systematic review will use a convergent segregated approach for data synthesis.³² This protocol was registered in PROSPERO (CRD42023391897; online supplemental file 1).

Eligibility criteria

The eligibility criteria in table 1 will be used to guide the selection of studies.

Table 1.

Eligibility criteria for studies included and excluded from the systematic review

	Inclusion criteria	Exclusion criteria
Type of participants and time period	Studies with adult participants aged ≥18 years who are diagnosed (self-assessed or medically assessed) with two or more chronic, physical, mental or infectious diseases. This is consistent with the definition of multimorbidity by the WHO and the Academy of Medical Sciences.1 2	Studies involving children or paediatric setting.
	Studies including patients who attended a healthcare facility for medical care (primary, secondary or tertiary care).	Studies where the intervention focuses on a single condition.
	Studies with patients living in an LMIC as defined by the World Bank criteria 2021 (EPOC LMIC filter).38	Studies involving patients with chronic conditions where the data for people with multimorbidity cannot be extracted.
	Only studies published from the year 2000, to incorporate most recent evidence.
Type of study designs, interventions and control arm	Primary studies, qualitative, quantitative and mixed, which describe outcomes of interventions designed to bring about changes in: healthcare organisations, the behaviour of healthcare professionals, the use of health services by healthcare recipients or the delivery arrangements; which include changes in how, when and where healthcare is organised and delivered, and who delivers healthcare.	Studies where the intervention is delivered in the community without the involvement of a healthcare facility.
	Studies describing more than one intervention, and if the data regarding the outcomes for the intervention of interest can be extracted.	Data regarding outcomes of interest cannot be extracted.
	Studies with the relevant study designs, routine care (care practised routinely at the time the study was done) or an alternative intervention could be the control arm.	Interventions in financial and governance categories of interventions as defined in the EPOC taxonomy.39
		Protocol papers will be excluded, and secondary papers in the form of systematic reviews will be hand searched for primary studies of relevance.
Type of outcome measures	Studies which have outcome measures of interest for LMICs as described below.	Studies that measure only disease-specific outcomes.
Other eligibility criteria		Non-English-language studies due to a lack of translation facilities.

EPOC, Effective Practice and Organisation of Care; LMIC, low-income and middle-income country.

Outcome measures of interest

The GACD has defined the outcome measures in multimorbidity suitable for LMICs. The outcome measures of interest for our review are guided but not restricted by the GACD-defined outcome measures. We have, in addition, planned to include studies that report on implementation and health service outcomes, as listed below.³¹

These outcomes can be divided into the following categories:

Patient outcomes

1. Health-related quality of life based on any scoring system such as EuroQol-5D, 36-Item Short Form Health Survey, 12-Item Short Form Health Survey and WHO Quality of Life Brief Version.

2. Mortality

3. Readmission rates

4. Self-efficacy

5. Treatment burden as measured by tools such as the Treatment Burden Questionnaire, Healthcare Task Difficulty Questionnaire and Multimorbidity Illness Perception Scale

Health services outcomes

1. Health economic measures, including incremental cost-effectiveness ratio and out-of-pocket costs

2. Patient and healthcare professionals’ perspectives on interventions

Implementation outcomes

1. Acceptability, including barriers and facilitators

2. Sustainability, including barriers and facilitators

Information sources

A systematic search of electronic databases (Cochrane Library, PubMed, MEDLINE, Ovid, TRIP database, LMIC database-3ie, SCOPUS) will be undertaken for original primary qualitative and quantitative studies, both controlled and uncontrolled, from 2000 up to March 2023. Observational studies are planned for inclusion so that important interventions from LMICs, which describe an intervention as an exposure and provide information towards its efficacy, are not missed.

References from systematic reviews will be hand searched and any additional relevant articles identified. If there is a significant time between the running of the search and the completion of the study, the search will be rerun.

Search strategy

A comprehensive search strategy was developed with the assistance of the University of Melbourne Library services using two main concepts:

1. Multimorbidity

2. LMIC

Both subject headings and keyword searches were undertaken on the online databases. On the other hand, grey literature was not searched.

The search strategy for Ovid Medline was created and translated for Cochrane Library and Scopus. For the PubMed database, the search strategy needed modifications due to the restriction on the number of characters allowed while searching and the lack of an ‘adjacent function’ at that time. The TRIP database had a filter for LMICs which was based on World Bank classification, and this was used instead of the search terms for ‘LMIC. 3ie is a database of LMICs and the search strategy only included the search terms for ‘multimorbidity’. The full search strategy can be found in online supplemental file 2.

Screening and data extraction

The titles of the papers identified will be extracted and screened for duplicates. A single author will screen the included titles to exclude irrelevant studies. Ten per cent of the titles will also be screened by a second author independently and a concordance of screening will be reported. If there is a lack of concordance, the inclusion and exclusion criteria will be further refined. The titles and abstracts of the included studies will be uploaded into the Covidence systematic review management platform.³³

The included titles, with their abstracts, will be screened by two independent authors, and relevant articles included. The full texts will also be assessed by two independent authors. Conflicts at these levels will be resolved by a third author.

Data extraction and the risk of bias assessment will be conducted by two authors independently, and any disagreements will be discussed and agreed upon or resolved by a third author.

The number of included and excluded studies, together with reasons for exclusion, will be represented using a Preferred Reporting Items for Systematic Reviews and Meta-Analyses diagram. Details of both included and excluded studies will be summarised. The data from the included studies will be extracted, using a proforma in the EpiData software (Box 1).

Box 1. Data fields to be extracted for the systematic review.

• Year

• Journal

• Title

• Type of study

• Definition of multimorbidity used

• Number of participants

• Mean age of study participants

• Country where the data originate

• Epidemiological data regarding subjects: age, sex distribution, socioeconomic state

• Whether the intervention targets concordant/discordant diseases

• What were the diseases

• Number of conditions

• Setting the intervention was implemented in:

1. Primary care

2. Hospital: inpatient care

3. Hospital: outpatient care

4. Perioperative care

• Context of the intervention

• Intervention summary

• Intervention category as per EPOC taxonomy

• Control arm if any

• Outcome category

• Outcome measure used

• Treatment effect (ratio/mean/median)

• Barriers to intervention

• Theory used in the study

• Findings relating to barriers to sustainability and acceptability of the intervention, illustration and level of plausibility

• Findings relating to facilitators to sustainability and acceptability of the intervention, illustration and level of plausibility

• Findings relating to patient perspectives of the intervention, illustration and level of plausibility

• Findings relating to healthcare worker perspective of the intervention, illustration and level of plausibility

• Risk of bias assessment

For quantitative studies, dichotomous outcomes will be recorded as risk ratios or ORs, and as mean difference or risk difference for continuous variables. For studies where these outcomes are measured differently, they will be converted to a common measure before combining, in line with guidance offered by the Cochrane Handbook for Systematic Reviews and Interventions.³⁴ For qualitative evidence pertaining to the outcomes of interest, findings as written by the author in the results section will be extracted verbatim.

The data extraction form will be piloted and iteratively modified. In addition, authors will be contacted to access missing data via email.

Risk of bias assessment

The risk of bias assessment for the different study designs will be assessed in duplicate, using the tools indicated in table 2. Consensus will be reached by discussion or by a third author if required, and the results of the risk of bias for both quantitative and qualitative studies will be presented in a ‘risk of bias table’.

Table 2.

Risk of bias assessment tools that will be used

Type of study	Risk of bias tool	Outcome of assessment
Randomised controlled trial	Cochrane Risk of Bias tool 2.040	High, low or unclear risk of bias
Non-randomised studies (case–control, cohort, etc)	Newcastle–Ottawa Scale41	Good-quality, fair-quality, poor-quality study
Qualitative	CASP checklist42	High-quality, moderate-quality, low-quality study

CASP, Critical Appraisal Skills Programme.

For quantitative studies, summary statistics will be provided for high-quality studies, if numbers allow, and will be presented in addition to overall summary statistics. The quality of the studies will be reflected in the summary of the evidence table as well.

Data synthesis and confidence in evidence

Data synthesis

The extracted data will be collated and summarised. The available interventions will be categorised as per the EPOC taxonomy, and the evidence for each will be discussed.

As this review has a broad scope, the plans for data synthesis will largely depend on the included studies and the interventions they describe. However, an overarching plan for synthesis is presented below.

As per Joanna Briggs Institute (JBI) guidance, there are two predominant methods for data synthesis in a mixed-methods review. The convergent integrated approach is where quantitative and qualitative data from all the studies are converted, either to qualitative or quantitative data-using methods of ‘qualitising’ data or ‘quantitising’ data, and then integrated.

The second approach is the convergent segregated approach. Here, the quantitative and qualitative syntheses are carried out independently and then if feasible (if the synthesised data from the qualitative and quantitative studies pertain to similar interventions and contexts), the data are integrated using a narrative summary. Data are not transformed in this form of synthesis. For our data synthesis, we will adopt the convergent segregated approach.³²

Quantitative study data synthesis

If studies addressing the same research question are found, then the studies will be checked for heterogeneity of clinical context, type of study, intervention, comparator, baseline characteristics and method of outcome measure. If the studies are heterogeneous, their results will be presented independently and if they are similar, then the statistical heterogeneity will be calculated using the I² statistic. If statistical heterogeneity is acceptable (less than 75%), the data will be pooled using the random-effects model to present a summary statistic and presented as a forest plot.³⁵ Outliers will be explored and if sufficient studies are available, a sensitivity analysis will be done.

Qualitative data synthesis

For qualitative studies, a meta-aggregative strategy will be used to present findings. In this approach, the results, as described by the authors of the primary studies, are extracted verbatim and based on whether adequate illustrations (quotes which explain the finding) are provided in the paper to support that result, it is marked as an unequivocal (illustration present and is beyond reasonable doubt), equivocal (illustration present but unclear association) or unsupported finding (not supported by an illustration). Findings relating to similar aspects of a particular intervention will then be grouped into categories by discussion among three authors to synthesise indicatory statements. This is in keeping with the methodology proposed by the JBI approach to qualitative synthesis.^{32 36}

A narrative synthesis of the results of the studies which are not eligible for pooling will be presented.

Subgroup analysis

The following subgroup analyses will be presented if relevant studies are found.

1. Randomised trials

2. Non-randomised trials

3. Qualitative studies

4. Studies performed in tertiary and teaching hospital settings

5. Studies performed in the perioperative setting

6. Studies performed in the primary care setting

Further subgroup analysis based on age, sex and type of disease cluster will be undertaken if a sufficient number of similar studies are found.

Confidence in the evidence

The Grading of Recommendations, Assessment, Development and Evaluations methodology will be used to mark the certainty of findings synthesised from quantitative studies as high, moderate or low, and Confidence in Evidence from Reviews of Qualitative research methodology will be used for qualitative studies. This will be done independently by two sets of authors, and conflicts will be resolved by discussion. The results will be presented with the summary of the evidence table along with the quality of the evidence informing the finding.³⁷

Publication bias

Publication bias will be assessed using Egger’s and Begg’s test, for quantitative studies.³²

Patient and public involvement

None.

Ethics and dissemination

No ethics approval is required for this systematic review. Results will be disseminated through publication in an open-access journal and presentation at medical conferences. The findings will also be disseminated through a summary report to general physicians, nurses, doctors in training and patients at the participating institutions.

Discussion

This is the first review in the authors’ knowledge to address this question in an LMIC setting. Previous systematic reviews have only examined RCTs and have included studies only from HICs.²⁹ Critical inputs regarding acceptability and sustainability can be gleaned from non-randomised and qualitative studies. This may be of particular importance in LMICs which are resource-poor and where social norms and structures are different from HICs.

We acknowledge that this study protocol has some limitations. One of these includes the restriction to studies published in English. Moreover, the heterogeneity in studies could be high, as interventions in this field could be varied and summarising them may not be feasible.

Multimorbidity is increasing in LMICs. Through conducting this systematic review, we hope to identify interventions and understand their effectiveness to inform future models of care for multimorbidity management in LMICs.

Ethics statements

Patient consent for publication

Not applicable.