Effect of home-based pulmonary rehabilitation on functional capacity in people with idiopathic pulmonary fibrosis—a systematic review protocol

Individuals with IPF diagnosed using high-resolution computed tomography findings and by registered physician will be included. We have considered these inclusion criteria as per the definition provided by the American Thoracic Society/European Respiratory Society guidelines [1]. IPF could be of any severity, but the individual should be in a stable clinical state. Individuals with IPF having dyspnoea with modified Medical Research Council score of grade 5 would be excluded from the review. Studies on surgically treated lung transplanted IPF individuals will be excluded as their care and rehabilitation pathways differ significantly from those with IPF individuals without lung transplant. Studies with mixed population will be excluded; however, we will include the study if it has carried out subgroup analysis of population of our interest. We will exclude other forms of interstitial lung disease other than IPF. There would be no restriction on the duration of diagnosis of IPF and the age of the population.

The intervention eligible to be included in this review is a comprehensive home-based PR, based on exercise training. Home-based PR is a self-care-based intervention, in which the physiotherapist/professional provides training and the participant perform the PR by themselves at home. This intervention is mostly unsupervised, but caregiver-supervised PR are eligible to be included. We will include studies with physiotherapist/any professionally trained individual periodically (minimum once in 15 days) supervising the participant, either at the participant’s home, community or hospital setup. This supervision could be web- or tele-based. Home-based PR may be performed in a group or individually in the community. To be included, studies must consider home-based PR programmes that have a component of aerobic exercise, resistance exercise, or both, with or without health education. Nevertheless, education programmes without home-based PR will be excluded. Minimum intervention duration should be 4 weeks but could be of any frequency per week. We will exclude studies that provided single exercise programmes. If the first and single training PR session has been delivered at centre/hospital, but remaining sessions were carried out at home/community setup, we will include the study. We will consider studies with individuals undergoing pharmacotherapy or any other standard care, but the co-intervention should have been equally distributed in both the groups. If the study included mixture of home-based and centre-based PR, we would include it provided there is subgroup available for unsupervised home-based PR.

Conventional supervised, centre-based PR treatment, no treatment or standard care is included in comparisons. We will include studies that compare unsupervised home-based PR with traditional/conventional PR or no rehabilitation. Comparison could be drawn with centre-based PR or between the providers/supervisors. Within home-based PR, comparison could be between two different forms (e.g. aerobic home-based PR compared to strength training home-based PR compared to strength training home-based PR), duration or intensity.

Primary outcome measures are functional capacity, as measured by 6-min walk distance, shuttle walk test or incremental shuttle walk test. Secondary outcome measures are condition specific quality of life measured using scales such as St. George respiratory questionnaire [26], Chronic Respiratory Distress Questionnaire [17] and King’s Brief Interstitial Lung Disease Questionnaire [28, 29]. Studies assessing outcome measures at time period of at least 4 weeks post intervention will be considered. We will exclude other generic tools for assessing quality of life such as WHO BREF and SF-36 (Short form- 36). We will also measure side-effects of PR such as fatigue and muscle weakness but not limited to desaturation and exacerbation of dyspnoea.

Randomised controlled trials (RCTs) with a parallel, cluster or cross-over design and quasi-randomised studies with at least two groups (intervention and control), followed-up for at least 4-week duration, will be included. Non-randomised studies (with single group), observational studies, letters to editor and reviews will be excluded.

Two review authors (RA and VK) will independently review titles and abstracts retrieved from the search and identify all potentially eligible studies. We will follow a piloted uniform screening protocol. Full text of included studies will be obtained, and the same team members will review independently according to the inclusion criteria. We will address disputes, if any, through discussion before consensus, and have a third review author (AM or SSP) for final decision when consensus cannot be achieved. The review authors shall document the rationale for excluding all full texts which do not meet the criteria for inclusion. The study selection process will be documented using PRISMA 2020 flow diagram (Additional data).

We will customise and use Cochrane EPOC data collection form [30] for extracting relevant information from included studies. We will follow an iterative process for piloting the data extraction form wherein all the authors will first extract the data from one included study and discuss the discrepancies in extracted data until consensus. This step will be repeated for more number of studies until authors have a thorough understanding in data extraction and are satisfied with the type of data to be extracted that would suffice for analysis. Furthermore, template for intervention description and replication (TIDieR) checklist and guide will be used to describe the intervention in detail [31]. From each included study, we will extract information on title, year of publication, author, study design, type of analysis, number of included participants, country, population (e.g. age, gender and other contextual information), severity of IPF, comorbidity, intervention (type, duration, intensity, frequency, provider etc.), comparison and outcome (follow-up, tool used, outcome assessor etc.) (Additional data). Corresponding authors of the included studies will be contacted in case of limited or lack of information in the studies. However, studies will be excluded if we do not receive a reply from the corresponding author within a fortnight. We will calculate disagreement (kappa) between reviewers for both screening and data extraction.

RA and VK, independently, will assess methodological quality of included studies using ‘Cochrane Risk of Bias 2’ tool [32]. Both the reviewers will clear the dissent, if any, by discussion. ROB 2 tool will assess bias at selection, blinding of participants, outcome assessment, reporting and other bias. We will judge each possible source of bias as high, low or some concern and provide a quote from the study report along with a justification. We will overcome disagreements by discussion until consensus for our decision on the ‘Risk of bias’.

Studies will be examined for methodological, clinical and statistical heterogeneity. The statistical heterogeneity between trials will be evaluated using I² statistics. The values will be categorised as ‘no’ (< 25%), ‘low’ (25–49%), ‘moderate’ (50–75%) or ‘substantial’ (> 75%) heterogeneity. Depending on the statistical heterogeneity, we will apply fixed or random effects model. If the statistical heterogeneity is 50 or higher, we will apply random effects model. Using RevMan, if possible, we will individually combine studies into a single meta-analysis using the generic method of inverse variance and will estimate the treatment effects reported by individual studies. Effect estimates such as odds ratios or risk ratios (for categorical data) and weighted/standardised mean differences (for continuous data) and their 95% confidence intervals will be computed. In case of substantial statistical heterogeneity or methodological and clinical heterogeneity, we will undertake narrative synthesis and subgroup analysis. Subgroups could be based on severity of IPF (or clinical variables), age, other contextual factors and study designs. Sensitivity analysis will be carried out to investigate the robustness of meta-analysis findings. We will consider variables such as risk of bias and sample size for conducting the sensitivity analysis. To assess publication bias, we will generate funnel plot and run statistical test such as Egger test [33].

We will assess certainty of evidence (high, moderate, low and very low) for each outcome measure through Grading of Recommendations Assessment, Development and Evaluation (GRADE) approach. GRADEpro GDT software will be used to generate summary of findings table [34].