Thirteen RCTs were included (278 participants, range four to 47): nine used adequate methods of sequence generation, one reported allocation concealment, 10 used some form of blinding, three reported withdrawals, eight clearly reported baseline comparability of groups and two reported use of intention-to-treat analysis. Jadad scores ranged from 3 to 5.
There was a significant benefit for the constraint-induced movement therapy group in measures of arm motor impairment (Fugl-Meyer Assessment MD 7.8, 95% CI 4.21 to 11.38; six RCTs, Ι²=12%), arm motor function (Action Research Arm Test MD 14.15, 95% CI 10.71 to 17.59; five RCTs, Ι²=20%) and perceived arm motor function (Motor Activity Log: amount of use MD 1.09, 95% CI 0.26 to 1.91; six RCTs, Ι²=90% and quality of use MD 1.02, 95% CI 0.55 to 1.49; six RCTs, Ι²=71%).
Sensitivity analysis excluded an outlying trial with a longer duration of training. Heterogeneity was reduced and the findings still significantly favoured constraint-induced movement therapy (Motor Activity Log amount of use: MD 0.78, 95% CI 0.37 to 1.19; five RCTs, Ι²=31% and quality of use: MD 0.84, 95% CI 0.42 to 1.25; five RCTs, Ι²=36%). There was no significant difference between the groups in focal disability level measured with the Functional Independence Measure (three RCTs) or the Barthel Index (one RCT).
For kinematic variables (three RCTs), the constraint-induced movement therapy group had a significantly shorter reaction time (MD -0.23, 95% CI -0.38 to -0.08; two RCTs, Ι²=0%) and a higher percentage of movement time where peak velocity occurs (MD 7.50, 95% CI 1.94 to 13.05; three RCTs, Ι²=36%), but did not differ significantly from the traditional rehabilitation group in other kinematic variables measured; detailed data were presented in the review.