Evidence for pulmonary rehabilitation in chronic respiratory diseases in sub-Saharan Africa: a systematic review.

BACKGROUND: Pulmonary rehabilitation (PR) is a highly effective non-pharmacological treatment for patients with chronic respiratory diseases.OBJECTIVE: To synthesise the evidence for PR practice and efficacy in sub-Saharan Africa.METHODS: We searched in PubMed and Scopus for relevant studies and scanned reference lists of relevant studies from these databases for additional studies. Articles meeting the inclusion criteria were included. Pre-determined data were extracted independently by two reviewers. A narrative synthesis approach was used in the interpretation of findings.RESULTS: Six studies were included, totalling 275 participants. Indications for PR were chronic obstructive pulmonary disease, asthma, pulmonary tuberculosis and post-tuberculosis lung disease. Programmes ran for 6-12 weeks, universally incorporated exercise, and variously used home-based and hospital-based delivery models. All were interventional studies, of which two were randomised controlled trials, and primarily reported pulmonary function and exercise tolerance endpoints. Evidence for individualising the exercise regimen was available in three studies.CONCLUSIONS: There is limited evidence on PR design and efficacy in sub-Saharan Africa, but available data support its use in a variety of chronic respiratory conditions. Future studies should report core outcome sets and their individualised exercise and education regimens.

post-tuberculosis lung disease. Programmes ran for 6-12 weeks, universally incorporated exercise, and variously used home-based and hospital-based delivery models. All were interventional studies, of which two were randomised controlled trials, and primarily reported pulmonary function and exercise tolerance endpoints. Evidence for individualising the exercise regimen was available in three studies. C O N C L U S I O N S : There is limited evidence on PR design and efficacy in sub-Saharan Africa, but available data support its use in a variety of chronic respiratory conditions. Future studies should report core outcome sets and their individualised exercise and education regimens. K E Y W O R D S : pulmonary rehabilitation; chronic respiratory diseases; sub-Saharan Africa; systematic review CHRONIC RESPIRATORY DISEASES (CRDs) are a major and increasingly important global health issue; chronic obstructive pulmonary disease (COPD) and asthma are the most common among these, 1 with combined annual death rates of approximately 3.15 million. 2,3 COPD is increasing in incidence, with agestandardised mortality being highest in low-and middle-income countries (LMICs), particularly South Asia and sub-Saharan Africa (SSA). 4,5 Known risk factors for CRDs include smoking, outdoor air pollution, household smoke exposure, occupational dust exposure, ozone and secondhand smoke. 6 Pulmonary tuberculosis (PTB) can also lead to irreversible lung damage which is most frequently radiologically apparent as bronchiectasis, fibrosis and cavitation. 6,7 Progressive loss of lung function is associated with long-term respiratory symptoms and clinical presentation as CRDs, including COPD, bronchiectasis and aspergillosis. 6 These conditions can jointly be referred to as post-TB chronic lung disorders (p-TBLD). 8 As TB remains one of the top 10 causes of mortality worldwide, with an estimated 10.4 million new cases and 1.3 million deaths in 2016, the problem of p-TBLD is likely to be substantial but is poorly measured. 6 CRDs can lead to significant chronic morbidity and loss of economic productivity, which burden patients, families and health systems alike. For conditions such as COPD, pharmacotherapy such as bronchodilators is currently the mainstay of the treatment. However, none of the drugs currently available for COPD can reduce the progressive decline in lung function which is the hallmark of this disease. 9 In addition, drugs for CRDs can be less available and affordable, especially in low-income settings, including SSA, where the availability and affordability of medicines and diagnostics recommended for the management of asthma and COPD is a big challenge. 10 Fortunately, the rational use of pharmacotherapy for CRDs could be complemented by non-pharmacological treatments. For COPD, pulmonary rehabilitation (PR) is well established as a highly effective intervention that improves symptoms, quality of life and survival 11 , and a recent trial has shown a substantial reduction in the risk of readmission in patients who undertake PR immediately after admissions for COPD. 12 Recent evidence also supports PR for other CRDs, including p-TBLD, asthma, and interstitial (fibrotic) lung disease (ILD). 13 In addition, while there is still a cost in providing it, PR is relatively more cost-effective than the pharmacotherapy for CRDs, as it may be delivered using existing local staff with minimal equipment. It would thus require little investment and be sustainable and scalable in resource-limited settings such as SSA. 14 PR is a comprehensive package of interventions, which can include exercise training, education and behaviour management. 15 Exercise training is a core component which should be prioritised; 16 however, this may require the modification of traditional training paradigms, and the need to take into consideration patient strength and endurance. 17 It has been suggested that PR is now sufficiently understood to obviate further randomized trials in high-income countries (HICs). 18 However, given that the design and delivery of PR programmes need to be individualised to patient groups and context, 19 highquality data are needed outside HICs.
While the need for PR in SSA is clear, there is currently no systematic review that has investigated the research evidence for its practice and effectiveness in the region. This study is aimed to close this research gap. Using the PICOS (participants, interventions, comparisons, outcomes and study design) approach, we systemically reviewed randomised controlled trials (RCTs) and non-RCTs to answer the following questions: 1) In which CRDs has PR been tested in SSA? 2) What designs and modes of delivery have been used in PR programmes tested in people with CRDs in SSA? 3) What is the evidence for individualising the exercise regimens of the PR programmes tested in people with CRDs in SSA? 4) What is the effectiveness of PR in people with CRDs in SSA?

METHODS
A protocol for this review was registered in the International Prospective Register of Ongoing Systematic Reviews (PROSPERO) 20 on 1 July 2019 (registration number: CRD42019128929). The conduct and reporting of the review were guided by the Preferred Reporting Items for Systematic Reviews and Meta-Analyses Protocol (PRISMA-P). 21,22 Ethics approval was not required, as only publicly available data were used.
Study eligibility criteria and identification JR and FB independently determined eligibility, and discrepancies were resolved by consensus. With reference to the same PICOS approach that informed our review questions, our eligibility criteria were as follows: participants with CRDs, attending a PR programme. The comparison or control domain was not necessary since additional RCTs comparing PR and conventional care in CRDs are not warranted as there is now enough evidence on the superiority of PR. 18 We therefore included both RCTs and non-RCTs. In the absence of standardised PR core outcomes, we captured the broadest range of outcomes possible, including symptoms, functional limitations, spirometry, and quality of life.
We selected peer-reviewed articles of studies conducted within SSA between 1997 and 2019. The start year (1997) was chosen to coincide with the publication of the first Pulmonary Rehabilitation Evidence-Based Guidelines. 23 Articles must have been accessible in full text in English (or available translation) and describe a study on PR for a defined respiratory disease or syndrome. The searches were performed in April 2019, using keywords and MeSH terms in PubMed and Scopus (see Supplementary  Table S1 for search terms used in PubMed); reference lists were also scanned for relevant secondary references.
Data extraction, synthesis and analysis Data were independently extracted by two reviewers (FB and CF) using a data extraction form which was based on our pre-review of identified manuscripts, with consensus-resolution of disagreement. The following data were sought and either coded or maintained in narrative form: study characteristics (country, year of publication and design), participant characteristics (sample size, condition and mean age), intervention characteristics (setting of delivery, duration of session and programme, frequency, components, team composition and evidence for individualising the exercise regimen), endpoints (primary and secondary) and summary measures of effect. We extracted all reported outcomes and synthesised and analysed the extracted data using a narrative description and tabular summary of key characteristics of the studies.

Assessment of reporting and methodological quality
We employed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) checklist, 24 and assessed potential for bias using the Newcastle-Ottawa Scale (NOS) 25 for non-randomised cohort studies. For randomised controlled studies, we employed the 2017 Consolidated Standards of Reporting Trials (CONSORT) checklist of information to include when reporting a RCT assessing non-pharmacological treatments (NPTs) 26 (for 2-arm parallel-group trials) and a CONSORT checklist for the reporting of multi-arm parallelgroup RCTs. 27 For methodological quality, we incorporated a revised Cochrane risk-of-bias tool for RCTs (RoB 2) 28,29 in tandem with the CONSORT checklist.

Study selection
Literature search terms identified 86 articles, of which six met the inclusion criteria ( Figure). We excluded 72 non-relevant articles during title and abstract review. Of the remaining 14 articles, we further excluded one duplicate, 30 one with abstract only and no author contact details, 31 one nonresearch guideline 32 and five that did not meet inclusion criteria. [33][34][35][36][37] Study characteristics Of the six finally included, three studies were conducted in South Africa, [38][39][40] two in Uganda 41,42 and one in Nigeria, 43 and all were published between 2010 and 2018. The studies were focused on the feasibility and acceptability of PR programmes. All 6 were experimental or intervention studies, with 2 being RCTs and 4 being pre-post (before-and-after) non-RCTs (Table).

Participant characteristics
A total of 275 participants were represented, all aged 18 years. Of these, COPD was the indication for inclusion in 59 participants across three studies. 38,42,43 A further 88 participants had asthma within a single study, 40 61 participants had p-TBLD in the two Uganda studies 41,42 and 67 participants were included in a separate study on PTB during antituberculous treatment 39 (Table).

Interventions
The PR programmes used were based either in referral-level hospitals (n ¼ 3) or at participants' homes (n ¼ 3). Their durations ranged from 6 to 12 weeks, with variations in composition as follows: exercise, education, nutrition, self-management activities and psychosocial support. The staff who delivered and supervised the programmes were variously composed across the studies, with at least a physiotherapist being present in all programmes to deliver and supervise the exercise training component. The other components of the programmes, such as education, were delivered and supervised by various professionals, including physicians, nurses, physiotherapists, counsellor, home-based caregivers, students (medical, physiotherapy and nutrition), assistant researchers (with unspecified expertise or profession) or a qualified practitioner in exercise    therapy and rehabilitation. More detailed information about each study is shown in the Table. Methodological quality A standard NOS criterion for what constitutes a highquality study has not yet been universally established, but we considered scores 7 to reflect high quality. 44 Based on this criterion, all four non-RCTs were found to be at high risk of bias (score 5). The issues included inadequate and/or lack of information on the three domains (selection, comparability, and outcome) on which quality assessment was based. Inadequate and/or lack of information in important areas were also an issue in the RCTs. Detailed results from the methodological quality assessment of all studies are given in Supplementary Table S2.

Exercise individualisation and education
Evidence for individualising the exercise regimens of PR programmes was available in three studies, [40][41][42] including the evidence for progression of exercise training loads. For example: The exercise regime was individually prescribed, monitored, and increased as the programme progressed. . . 41 The intervention was performed in groups, but the exercise training was individually prescribed and intensified as the programme progressed. 42 Individual age predicted maximum heart rate (HR max ) was utilised. . . to develop a suitable programme for all subjects without the need for expensive equipment. . . 40 The remaining three studies had no explicit evidence of regimen individualisation. Details on other programme components, including education and self-management aspects, were very limited (either completely missing or inadequate).

Effectiveness
The Table shows the effects of PR programmes. Multiple primary outcomes were the norm in all the studies reviewed, some of which were inconsistently used, which led to the reporting various results, including physical, economic, and psychosocial effects, and self-esteem, health literacy, exercise tolerance, pulmonary function and quality of life. Pulmonary function and exercise tolerance effects were predominantly assessed and reported. Forced expiratory volume in 1 sec (FEV 1 ) and forced expiratory volume (FVC) were common outcome measures for pulmonary function, while the 6-min walk test (6MWT) and the incremental shuttle walk test (ISWT) were used as outcome measures for exercise tolerance. Pulmonary function was reported in 5 studies, [38][39][40][41]43 of which 3 (including the 2 RCTs 37,38 ) demonstrated improvement after the programme, 38-40 1 suggested a non-significant but potentially clinically important increase, 41 and 1 showed no significant change. 43 Exercise tolerance was reported in 4 studies, 38,39,41,43 of which 1 reported significant improvement, 43 1 reported a non-significant but potentially clinically important increase, 41 and 2 showed no significant change 38,39 (including an RCT 39 ). Dyspnoea, as the most frequently reported symptom in CRD, 45 was assessed in four studies. 38,39,41,43 Of these, 2 reported significant improvements (including an RCT 39 ), 38,39 1 reported a non-significant but potentially clinically important difference, 41 and 1 showed no significant change. 43

DISCUSSION
We found evidence of clinically significant improvements in symptoms and lung function in patients receiving PR for various CRDs in SSA. However, data were limited and only three countries were represented. Studies predominantly examined participants with COPD (three out of the six studies). While the evidence base for PR outside of Africa is dominated by this patient group, recent evidence suggests that similar benefits could be gained in other CRDs, including interstitial lung disease, bronchiectasis, asthma, cystic fibrosis, lung transplantation, lung cancer and pulmonary hypertension. 15,46,47 For example, a new study has shown that patients with bronchiectasis benefit from PR to broadly the same extent as patients with COPD. 48 PR is also clearly indicated in the management of patients with p-TBLD, 15,49 which is infrequently found in highincome countries, but extremely common in TBendemic regions, 6 particularly LMICs. 50 It is reported that, in these settings, TB increases the odds of p-TBLD by more than threefold on average. 6 PR can be home-or hospital-/centre-based; 11 each was equally represented in this systematic review. Although it is recognised that hospital-/centre-based programmes are a standard mode of PR delivery, 51 both modes were effective in various study outcomes of interest in this systematic review. In a multicentre RCT, 52 home-based PR was found to be a useful, equivalent alternative to hospital-based PR in patients with COPD. In addition, another recent study, which aimed to compare the cost-effectiveness and cost-utility of home and centre-based PR for adults with stable COPD, concluded that home-based PR provided a cost-effective alternative model for people with COPD who cannot access traditional centrebased programmes. 53 However, to date, home-based PR has still not been widely adopted into clinical practice, possibly because a sound economic model has not been developed. 17 In line with recommendations, 54 each PR pro-gramme ran for at least 6 weeks and focussed on exercise training, which seems reasonable, as exercise is a component with the strongest level of evidence for benefit. 16 However, the health gains attributable to a broader package are important and may be underrepresented in the existing African literature. In addition, to optimise the physical and physiological benefits, it is necessary to individualise exercise training for each patient, including exercise intensity. 55 In this systematic review, there was no evidence for individualising the PR exercise regimens in three studies, which made it difficult to assess the quality of these programmes. Moreover, the issues of methodological quality may limit the validity of the study results when using these as a basis for recommending PR with absolute certainty in the management of CRDs in SSA. There was variable reporting of outcomes of interest, of which pulmonary function and exercise tolerance indices predominated. Studies in our review noted improvements in lung function which might not be typical of the broader literature. This raises the possibility that, perhaps due to limited access to pharmacological therapy, patients in SSA may benefit differently from those for whom prior pharmacotherapy was optimal. We note that the choice of reported metrics was also variable, leading to heterogeneity, which limits the potential for meta-analysis or a broader assessment. 56,57 Our review would support the suggestion of a consistently implemented ''core outcome set''. 58,59 This could reduce outcome reporting bias and broaden the relevance to different stakeholders, including patients, healthcare workers and those overseeing the wider healthcare system. 60 In this systematic review, both exercise tolerance and dyspnoea were outcomes in four of six studies; 38,39,41,43 these are well-recognised and stakeholder-valued positive PR outcomes in other settings. 61 Selection of core outcomes should include reaching a consensus on the key measures which best represent them.
Our systematic review used a broad search strategy to identify relevant evidence for PR in SSA and rigorous implementation of PRISMA guidelines. However, only studies published in English were included. This may have limited our sample, but we found no examples of this during article screening. The small number of studies, their heterogeneity and the lack of consistent outcome variables did not allow meta-analysis.

CONCLUSIONS
Despite limitations, all the included studies in this review suggest that extending PR programmes in SSA could be beneficial to patients. Efforts should be made to individualise the exercise regimens of PR programmes in the region. Larger studies with high methodological quality that employ consistent outcome variables are warranted to allow careful metaanalysis of both effectiveness and cost-effectiveness. Se realizó una búsqueda de artículos pertinentes en las bases de datos PubMed y Scopus y se examinaron las listas de referencias de los artículos escogidos para encontrar otros estudios. Se incluyeron los artículos que cumplían los criterios de inclusión. Dos revisores extrajeron de manera independiente los datos predeterminados. Los hallazgos se interpretaron mediante una estrategia de síntesis narrativa. R E S U LTA D O S: Se incluyeron seis estudios con un total de 275 participantes. Las indicaciones de la PR eran enfermedad pulmonar obstructiva crónica, asma, tuberculosis pulmonar y enfermedad pulmonar postuberculosis. Los programas duraban de 6-12 semanas, todos comportaban ejercicios y utilizaban diversamente un modelo de ejecución en los hogares o en los hospitales. Todos los estudios eran intervencionistas y dos de ellos eran ensayos aleatorizados y notificaban como principales criterios de evaluación, esencialmente la función respiratoria y la tolerancia al ejercicio. Tres estudios contenían información sobre la individualización del esquema de ejercicios.