Effect of Cardiac Rehabilitation Dose on Mortality and Morbidity: A Systematic Review and Meta-regression Analysis

Objective To ascertain the effect of cardiac rehabilitation (CR) dose (ie, duration × frequency/wk; categorized as low [<12 sessions], medium [12-35 sessions], or high [≥36 sessions]) on mortality and morbidity. Methods The Cochrane, CINAHL, EMBASE, PsycINFO, and MEDLINE databases were systematically searched from inception through November 30, 2015. Inclusion criteria included randomized or nonrandomized studies with a minimum CR dose of 4 or higher and presence of a control/comparison group. Citations were considered for inclusion, and data were extracted in included studies independently by 2 investigators. Studies were pooled using random-effects meta-analysis and meta-regression where warranted (covariates included study quality, country, publication year, and diagnosis). Results Of 4630 unique citations, 33 trials were included comparing CR to usual care (ie, no dose). In meta-regression, greater dose was significantly related to lower all-cause mortality (high: −0.77; SE, 0.22; P<.001; medium: −0.80; SE, 0.21; P<.001) when compared with low dose. With regard to morbidity, meta-analysis revealed that dose was significantly associated with fewer percutaneous coronary interventions (high: relative risk, 0.65; 95% CI, 0.50-0.84; medium/low: relative risk, 1.04; 95% CI, 0.74-1.48; between subgroup difference P=.03). This reduction was also significant in meta-regression (high vs medium/low: −0.73; SE, 0.20; P<.001). Publication bias was not evident. No dose-response association was found for cardiovascular mortality, all-cause hospitalization, coronary artery bypass graft surgery, or myocardial infarction. Conclusion A minimum of 36 CR sessions may be needed to reduce percutaneous coronary interventions. Future studies should examine the effect of actual dose of CR, and trials are needed comparing different doses. PROSPERO Registration CRD42016036029.