Predictors of exercise capacity following exercise-based rehabilitation in patients with coronary heart disease and heart failure: A meta-regression analysis

Background The aim of this study was to undertake a comprehensive assessment of the patient, intervention and trial-level factors that may predict exercise capacity following exercise-based rehabilitation in patients with coronary heart disease and heart failure. Design Meta-analysis and meta-regression analysis. Methods Randomized controlled trials of exercise-based rehabilitation were identified from three published systematic reviews. Exercise capacity was pooled across trials using random effects meta-analysis, and meta-regression used to examine the association between exercise capacity and a range of patient (e.g. age), intervention (e.g. exercise frequency) and trial (e.g. risk of bias) factors. Results 55 trials (61 exercise-control comparisons, 7553 patients) were included. Following exercise-based rehabilitation compared to control, overall exercise capacity was on average 0.95 (95% CI: 0.76-1.41) standard deviation units higher, and in trials reporting maximum oxygen uptake (VO(2)max) was 3.3ml/kg.min(-1) (95% CI: 2.6-4.0) higher. There was evidence of a high level of statistical heterogeneity across trials (I-2 statistic>50%). In multivariable meta-regression analysis, only exercise intervention intensity was found to be significantly associated with VO(2)max (P=0.04); those trials with the highest average exercise intensity had the largest mean post-rehabilitation VO(2)max compared to control. Conclusions We found considerable heterogeneity across randomized controlled trials in the magnitude of improvement in exercise capacity following exercise-based rehabilitation compared to control among patients with coronary heart disease or heart failure. Whilst higher exercise intensities were associated with a greater level of post-rehabilitation exercise capacity, there was no strong evidence to support other intervention, patient or trial factors to be predictive.