Fourteen studies were included in the review: 11 cluster randomised controlled trials (RCTs) (approximately 3,164 participants), one crossover RCT (45 participants), one quasi-experimental study (approximately 117 participants) and a parallel group RCT (40 participants). Five studies were rated as having high risk of bias, eight as having moderate risk of bias and one as having a low risk of bias. Follow-up ranged from one to 156 weeks (with most occurring after 35 weeks) in the 10 studies that measured baseline MPVA levels.
All of the included studies reported a higher proportion of active learning time in the intervention group compared with the control group; most of these differences were reported as being statistically significant. The meta-analysis (13 studies) revealed an absolute difference of 10.37% (95% CI 6.33 to 14.41) of lesson time spent in MPVA in favour of the intervention over controls. The 10.37% difference related to 24% more active learning time in the intervention groups than in the control groups (SMD 0.62, 95% CI 0.39 to 0.84). Substantial heterogeneity was shown between the studies (Ι²=88%).
Results from subgroup analyses according to intervention type showed the same direction of effect (reported in paper). None of the factors in the other subgroup analyses (such as age and gender) appeared to moderate the intervention effects (data not shown). No evidence of publication bias was reportedly found but some outlier studies were shown on the funnel plot.