• Regina Kunz, Ilona Autti-Rämö, Heidi Anttila, Antti Malmivaara, and Marjukka Mäkelä.
• Basel Institute for Clinical Epidemiology, University Hospital of Basel, Hebelstrasse 10, 4031 Basel, Switzerland. rkunz@uhbs.ch
• J Clin Epidemiol. 2006 Dec 1; 59 (12): 1239-48.

ObjectivesTo identify critical issues in performing randomized controlled trials (RCTs) on complex interventions such as physiotherapy in multifaceted disabilities like cerebral palsy (CP); to systematically assess how well trials handled patient characteristics, key components of complex interventions, and outcome assessments; to make suggestions for improving the effectiveness of physiotherapy research.Design And MethodsA methodological review. An ongoing metareview served as sampling frame. RCTs on physiotherapy in children with CP published after 1990 were eligible. The main outcome measures were sampling, recruitment, and comparability of groups; defined components of a complex intervention; outcome measures according to the World Health Organization International Classification of Functioning and clinical interpretation of score change.ResultsFifty-seven clinical trials were identified, of which 14 RCTs were included. Most studies used simple randomization, complemented by balancing techniques such as matching or stratification. Only three of 14 studies specified concealed allocation. Numerous studies provided sufficient details on components of a complex intervention (e.g., 'standardized intervention' (12/14); 'the active therapeutic ingredient' (8/15). Overlap of instruments across studies was poor, only instruments on activity endpoints were more homogeneous. The clinical interpretation of observed score changes was mostly missing.ConclusionWe found good to fair methodological quality in a considerable number of RCTs on physiotherapy in CP children. Nevertheless, improvement is indicated for certain areas in design and performance of future studies. This review shows that informative RCTs on complex interventions for multifaceted disabilities are feasible.

11 keywords...

hide…