Functional outcome measures in contemporary stroke trials

Abstract

Various instruments are used to describe poststroke functional outcome, with limited consensus as to optimal end-point for clinical trial use. Many of the popular assessment tools are administered with little formal guidance on best practice. Thus there is potential for substantial heterogeneity in functional outcome assessment poststroke, with consequent effects on trial quality. We examined functional assessment methodology in recent stroke trials. We reviewed six journals representing high-impact international publications in the fields of: stroke (Stroke); neurology (Neurology, Lancet Neurology) and internal medicine (Lancet, New England Journal Medicine; Journal of the American Medical Association). Journals were hand searched for all interventional studies in stroke patients between 2001 and 2006 inclusive. Chosen manuscripts were then analyzed for outcome assessment methodology. We identified 126 trials, comprising a mix of early hypothesis generating studies through to multi-centre trials (phase I: four trials; phase II: 46 trials; phase III: 20 trials; noninvestigational medicinal product studies: 56 trials). The median number of patients assessed per trial was 100. Across the trials, 47 different outcome measures were used. One hundred trials had functional outcome assessment as the primary study end-point. The median number of outcome measures was two per trial (range 1-9). The modified Rankin scale was the most prevalent outcome assessment (64.3%); followed by Barthel index (40.5%). A minority of trials (33.3%) provided full details on outcome assessment methodology. Among these trials there was substantial heterogeneity in data collection procedures. There is heterogeneity in the use of functional outcome measures in stroke trials. This compromises comparison and meta-analysis. Trialists continue to use poorly validated approaches to outcome assessment. Given the potential effects on data quality, explicit description of methodology should be mandatory for all trials and rigour is desirable.