Motor recovery after stroke: A systematic review of the literature

Objective: To collect and integrate existing data concerning the occurrence, extent, time course, and prognostic determinants of motor recovery after stroke using a systematic methodologic approach. Data Sources: A computer-aided search in bibliographic databases was done of longitudinal cohort studies, original prognostic studies, and randomized controlled trials published in the period 1966 to November 2001, which was expanded by references from retrieved articles and narrative reviews. Study Selection: After a preliminary screening, internal, external, and statistical validity was assessed by a priori methodologic criteria, with special emphasis on the internal validity. Data Extraction: The studies finally selected were discussed, based on the quantitative analysis of the outcome measures and prognostic determinants. Meta-analysis was pursued, but was not possible because of substantial heterogeneity. Data Synthesis: The search resulted in 174 potentially relevant studies, of which 80 passed the preliminary screening and were subjected to further methodologic assessment; 14 studies were finally selected. Approximately 65% of the hospitalized stroke survivors with initial motor deficits of the lower extremity showed some degree of motor recovery. In the case of paralysis, complete motor recovery occurred in less than 15% of the patients, both for the upper and lower extremities. Hospitalized patients with small lacunar strokes showed relatively good motor recovery. The recovery period in patients with severe stroke was twice as long as in patients with mild stroke. The initial grade of paresis was the most important predictor for motor recovery (odds ratios [OR], >4). Objective analysis of the motor pathways by motor-evoked potentials (MEPs) showed even higher ORs (ORs, >20). Conclusions: Our knowledge of motor recovery after stroke in more accurate, quantitative, and qualitive terms is still limited. Nevertheless, our data synthesis and quantitative analysis comprises data from many methodologically robust studies, which may support the clinician in the management of stroke patients. With respect to early prognosis of motor recovery, our review confirms clinical experience that the initial grade of paresis (as measured on admission in the hospital) is the most important predictor, although the accuracy of prediction rapidly improves during the first few days after stroke. Initial paralysis implies the worst prognosis for subsequent motor recovery. Remarkably, the prognostic accuracy of MEPs appears much higher than that of clinical examination for different subgroups of patients.