Systematic review of prognostic models in patients with acute stroke

Prognostic models in stroke may be useful in clinical practice and research. We systematically reviewed the methodology and results of studies that have identified independent predictors of survival, independence in activities of daily living, and getting home in patients with acute stroke. Eligible studies (published in full in English) included at least 100 patients in whom at least 3 predictor variables were assessed within 30 days of stroke onset and who were followed up for at least 30 days. We recorded 25 indicators of the validity and practicality of each model and identified variables that were consistent independent predictors of each outcome. Eighty-three separate prognostic models were found but most had potentially serious deficiencies in internal and statistical validity, many had limited generalisability, and none had been adequately validated. Only 4 studies met 8 simple quality criteria. Over 150 different predictor variables have been analysed but most were assessed in only 1 or 2 models. None of the existing prognostic models have been sufficiently well developed and validated to be useful in either clinical practice or research. Better quality models must be produced to enable, for example, adequate case-mix correction when comparing outcome among different groups of stroke patients.