Introduction/Objective: To describe issues and problems arising when undertaking a systematic review of a rehabilitative intervention, and ways to address these.

Methods: Observations during a review of speech and language therapy for post stroke aphasia within the Cochrane Collaboration stroke group.

Results: This type of health care did not fit easily into the model of a review of a medical intervention. This applied particularly when: deciding study eligibility (definitions, outcomes, study design); developing the search strategy (where, how extensive and by whom); deriving data (few, poorly reported 'old' trials); performing the analysis (poor quality data, heterogeneity, and variable continuous data); and drawing conclusions (both for current practice and future research). More general problems for the reviewer, a non practising speech and language therapist, were the medical orientation of the Cochrane Collaboration, and tensions between this approach and the views of speech and language therapists. Ways these were addressed included: keeping eligibility criteria broad in the first instance; consulting experts in the field; working in a team; forging good links with others facing the same problems; and having confidence in non medical expertise.

Discussion: This study illustrates a number of issues that the Cochrane Collaboration may wish to address to improve support for reviewers of medical interventions; it also raises questions about the future research agenda for the speech and language therapy profession.