The effect of pulmonary rehabilitation design on outcomes: a component network meta-analysis

<bold>Background:</bold> The optimal programme design to maximise benefits from pulmonary rehabilitation (PR) is unclear. <bold>Aim:</bold> To investigate the effect of different components of PR design on outcomes. <bold>Methods:</bold> We conducted a systematic review and Bayesian component network meta-analysis to investigate the effect of exercise modality, non-exercise components and supervision on exercise capacity, dyspnoea and quality of life (QoL) in COPD. We included randomised controlled trials (RCTs) in which at least one arm performed an intervention involving exercise. We investigated for effects of cohort demographics. <bold>Results:</bold> We included 341 RCTs (n=19,165, across 44 countries) with 228 individual components. In a combined model, exercise training alone (without supervision) resulted in improvements in exercise capacity (Figure) and QoL (SMD 0.5, 95%CrI 0.1;1.0) but not dyspnoea (SMD 0.2, 95%CrI -0.2;0.5). Direct supervision enhanced gains in exercise capacity (Figure), QoL (SMD 0.4 95%CrI 0.0;0.8) and dyspnoea (SMD 0.5 9%%CrI 0.1;0.8). Remote supervision or inclusion of non-exercise components did not improve outcomes. Direct supervision was more effective in cohorts with lower baseline exercise capacity and worse QoL. In a split model, investigating exercise intensity and modality, no single exercise regimen appeared superior. <bold>Conclusion:</bold> Whilst unsupervised exercise appears effective in COPD, direct supervision may enhance outcomes. <fig><object-id>erj;64/suppl_68/OA3672/F1</object-id><object-id>F1</object-id><object-id>F1</object-id><graphic></graphic></fig>