Identifying inconsistencies in intervention descriptors and outcome reporting within systematic reviews of physical activity interventions in COPD

Abstract

Introduction: This review of reviews aimed to identify inconsistencies in intervention descriptors and outcome reporting within systematic reviews (SR) of physical activity (PA) interventions for people with COPD. Methods: Six SRs directed towards PA published since 2012 and a European Respiratory Society Statement (ERS) on PA were compared. Data were extracted on study characteristics: year, design, intervention, control and PA measure; and outcomes: direction of intervention effect reported by review. Results: 85 separate studies were assessed (SR1-2012: n=7; SR2-2014: n=31; SR3-2015: n=20; SR4-2016: n=60; SR5-2016: n=39; SR6-2017: n=10; ERS-2014: n=20), of which 53 (62%) were assessed in more than one review. In 26 (49%) of these, the descriptor of the intervention was different between reviews (e.g. “exercise training” vs “walking program”; “counselling” vs “PA advice”). Across reviews there were discordant judgements on study outcomes for 16 (30%) of the 53 studies appearing in more than one review, due to reviews variously reporting: 1) within-group vs between-group change (n=5, 31%); 2) between-group difference for pre- to post-intervention change vs post-intervention scores (n=3, 19%); 3) sub-group analyses (n=2, 13%); 4) different outcomes (n=2, 13%); 5) different end-points (n=1, 6%); and 6) identifying the control as an intervention group (n=1, 6%). For two studies, it was unclear how the discordance in outcomes occurred. Conclusion: When the same studies are included within different SRs, inconsistencies in intervention descriptors and outcome reporting occur, this may lead to misinterpretation of results by readers.