Eighteen studies were included (n=736): 14 randomised controlled trials and four controlled trials. PEDro scores ranged from 3 to 7 out of 10 (median 5). Six studies reported allocation concealment, five reported blinding of assessors for some outcomes, two studies reported intention-to-treat analysis and three studies reported no withdrawals. Mean drop-out rate was 21% across 16 studies (range zero to 38%). Attendance at interventions sessions was about 88% (10 studies). None of the studies reported withdrawals due to adverse effects of the programme.
Follow-up duration ranged from 12 weeks to 12 months after completion of the intervention (five studies).
Body structure and function: Progressive resistance exercise was associated with a medium statistically significant improvement in knee extensor strength (d=0.52, 95% CI 0.30 to 0.74; 10 studies), a large significant effect of leg press strength (d=0.96, 95% CI 0.26 to 1.66; five studies) and a small significant increase in latissimus dorsi strength (d=0.43, 95% CI 0.07 to 0.8; four studies). There was no statistically significant effect on biceps strength, pectoral strength (among homogeneous studies) or quadriceps cross-sectional area.
Activity: Progressive resistance exercise was associated with a significant improvement in submaximal cycling endurance compared to no intervention (d=0.87, 95% CI 0.29 to 1.44; five studies) and large significant effects on timed stair climbing (d=1.31, 95% CI 0.48 to 2.13; two studies) and timed stand-to sit performance (d=1.07, 95% CI 0.22 to 1.92; three studies). Aerobic exercise was associated with a significant improvement in submaximal cycling endurance compared to progressive resistance exercise (d=-0.89, 95% CI -1.42 to -0.37; two studies). There was no statistically significant effect on walking distance when analysed by control or arm elevation tasks.
Participation: Intervention effects remained unclear (three studies).
Long-term effects: There were no statistically significant differences between interventions at follow-up (three studies).