Current issues of ACP Journal Club are published in Annals of Internal Medicine


Review: Programs for rehabilitation after stroke are most effective when given early and in younger patients

ACP J Club. 1993 July-Aug;119:8. doi:10.7326/ACPJC-1993-119-1-008

Source Citation

Ottenbacher KJ, Jannell S. The results of clinical trials in stroke rehabilitation research. Arch Neurol. 1993 Jan;50:37-44.



To determine the effects of rehabilitation programs on functional abilities of patients who survive a stroke.

Data sources

Computer and manual searches of Index Medicus, Cumulative Index to Nursing and Allied Health, Psychological Abstracts, Educational Resources Information Center, Dissertation Abstracts, and Science Citation Index were done for 1960 to 1990, using the terms rehabilitation, stroke, cerebrovascular accident, hemiplegia, hemiparesis, remedial therapy, function, activities of daily living (ADL), and functional outcome. Bibliography reviews and citation tracking were done, and information clearinghouses were checked.

Study selection

Studies were selected if they evaluated treatments or procedures designed to improve functional performance; the primary diagnosis was hemiplegia or hemiparesis secondary to a stroke; functional abilities studied included motor or reflex function, cognitive skill, language ability, visual-perception function, or ADL; a comparison group was included; and enough data were available to calculate effect size.

Data extraction

36 of the original 124 clinical investigations met the selection criteria. The effect sizes were calculated using the d-index, the difference between the group means in terms of their common (mean) standard deviation.

Main results

Diagnosis was right hemiplegia (48%), left hemiplegia (41%), and bilateral or undifferentiated hemiplegia (11%) in 3717 patients (mean age 69 y, 52% men). The mean d-index was 0.40 (95% CI 0.35 to 0.45), which shows that the average person in the treatment groups functioned better than 66% (CI 64% to 67%) of the participants in the control groups. The groups did not differ for service provider (psychologist, therapist, team, or other) or length of treatment. Outcomes were better for early interventions after stroke and for younger patients. ADL and visual-perception function showed the most improvement. More rigorous experimental designs and blinding of outcomes were associated with less treatment effect.


Rehabilitation programs designed to improve physical function in patients after a stroke are effective. Younger patients benefit more, and early intervention after stroke enhances outcomes, especially for activities of daily living and visual-perception function. Length of program and type of caregiver do not affect outcomes.

Sources of funding: National Institute on Disability and Rehabilitation Research and Department of Education.

For article reprint: Dr. K.J. Ottenbacher, School of Allied Health Sciences, University of Texas Medical Branch, 301 University Boulevard, Galveston, Texas 77555-1028, USA. FAX 409-747-1623.


Stroke is the most common serious neurologic disorder. Many persons are left with substantial impairment and disabilities. Despite a large body of medical research, the effectiveness of rehabilitation is unclear. The review by Ottenbacher and Jannell is unique in that it is the first time the techniques of meta-analysis have been applied to answer whether rehabilitation is effective, providing quantification of effect size and more objective analysis than traditional narrative reviews.

The major difficulty with applying the techniques of meta-analysis to answer the question of the effectiveness of rehabilitation after stroke is that the published trials are heterogeneous in patient entry characteristics, services provided, and outcomes measured. The authors use a concept termed the "d-index" to estimate effect size, and they discuss the rationale for its use to overcome these limitations. Unfortunately, most readers may still find this a difficult concept to understand.

The analysis indicated that rehabilitation was effective, but the difference between the treated and untreated groups was small and of arguable clinical significance. Concern should have been raised about this conclusion when it was found that it did not matter what type of treatment was provided or for how long, only that treatment was given early and to younger patients. Has the analysis measured some difference not related to treatment? Was the measured effect from rehabilitation even present given that randomized trials and blinding of outcome tended to reduce it? I believe, having worked as a stroke rehabilitation program director, that we do reduce patient disabilities and handicaps, but I cannot prove it and am not convinced by the evidence summarized in this review.

This review outlines many of the difficulties in stroke rehabilitation outcome research. It supports the need for definitive randomized trials with clearly defined patient entry characteristics, treatment modalities, and clinically important outcome measures. Only through the accumulation of proper trials will we know for whom, with what, and to what end rehabilitation after stroke is indicated.

Wieslaw Oczkowski, MD
McMaster UniversityHamilton, Ontario, Canada