Influence of task-oriented training content on skilled arm-hand performance in stroke: a systematic review.

OBJECTIVE This review evaluates the underlying training components currently used in task-oriented training and assesses the effects of these components on skilled arm-hand performance in patients after a stroke. METHODS A computerized systematic literature search in 5 databases (PubMed, CINAHL, EMBASE, PEDro, and Cochrane) identified randomized clinical trials, published through March 2009, evaluating the effects of task-oriented training. Relevant article references listed in publications included were also screened. The methodological quality of the selected studies was assessed with the Van Tulder Checklist. For each functional outcome measure used, the effect size (bias corrected Hedges's g) was calculated. RESULTS The intervention results in 528 patients (16 studies) were studied. From these, 15 components were identified to characterize task-oriented training. An average of 7.8 (standard deviation = 2.1) components were used in the included trials. There was no correlation between the number of task-oriented training components used in a study and the treatment effect size. "Distributed practice" and "feedback" were associated with the largest postintervention effect sizes. "Random practice" and "use of clear functional goals" were associated with the largest follow-up effect sizes. CONCLUSION The task-oriented training was operationalized with 15 components. The number of components used in an intervention aimed at improving arm-hand performance after stroke was not associated with the posttreatment effect size. Certain components, which optimize storage of learned motor performance in the long-term memory, occurred more in studies with larger treatment effects.