Seven RCTs were included (597 patients): 5 RCTs were unconfounded and 2 RCTs were confounded.
The trials were heterogeneous in terms of the aims and objectives of the intervention, the physiotherapy techniques employed, patient selection, rehabilitation setting and timing of intervention. Methodological problems found included the lack of clear definition of physiotherapy interventions and the frequent use of a mixture of physiotherapy and occupational therapy, use of different disability and impairment scores and incomplete data in two trials.
Intensity of physiotherapy: control groups received 20 to 40 minutes per day of formal physiotherapy; intervention patients received modest increase in therapy input (usually 1.5 to 2 times the control level).
Death or deterioration by the end of follow-up: OR = 0.54 (95%CI: 0.34, 0.85; P < 0.01).
Impairment score: at randomisation and at 12 months (P > 0.05); at 3 months (P < 0.05).
Disability score: at randomisation and at 12 months (P > 0.05); at 3 months (P < 0.05).
The standard motor and ADL scores: at initial assessment (P < 0.05); at final review (P > 0.05).
Results were similar after exclusion of 2 trials which were confounded by the experimental groups being managed in different settings and after limiting the analysis to those trials using blinded outcomes.