Eighty-one studies (2,492 participants) were included: 30 parallel RCTs (1,197 participants), 9 randomised crossover trials (287 participants), 5 non-randomised controlled trials (126 participants), and 37 longitudinal cohort studies (882 participants).
Events (including hospitalisation) causing discontinuation of the programmes (17 RCTs): there were 30 adverse events in the exercise groups compared with 34 in the control groups (OR 0.85, 95% CI: 0.51, 1.41, P=0.52).
Mortality (11 RCTS): there were 26 deaths in the exercise groups and 41 in the control groups (OR 0.61, 95% CI: 0.37, 1.02, P=0.06).
For the composite end point of adverse event or death, exercise appeared to be favourable (18 studies, OR 0.68, 95% CI: 0.46, 1.00, P=0.05)
Safety: no exercise-related deaths were reported in any of the studies.
Functional capacity (57 studies): the mean increase in maximal oxygen uptake was 16.8% (SD 8.0, 95% CI: 13.7, 17.9). The greatest increase was in the 40 studies involving continuous or intermittent aerobic exercise (mean increase 16.5%, SD 6.9, 95% CI: 14.3, 18.7).
Exercise mode: no correlation was found between session frequency, duration, intensity or programme duration and functional improvement.