Entry criteria and baseline characteristics predict outcome in acute stroke trials

Background and Purpose-We sought to study the range of entry criteria and b aseline characteristics in acute stroke trials and to understand their effe cts on patient outcomes. Methods-Randomized, placebo-controlled therapeutic trials in patients with acute ischemic stroke were identified. Entry criteria, baseline clinical ch aracteristics, and outcome were extracted for the placebo group of each tri al. The relationship between key variables was then determined. Results-Across 90 placebo groups identified, there was great variation in e ntry criteria and outcome measures. This was associated with divergent outc omes; for example, in some studies most placebo group patients died, while in other studies nearly all had no disability. Entry criteria were signific antly correlated with outcome; for example, higher age cutoff for study ent ry correlated with 3-month mortality. Entry criteria also predicted baselin e clinical characteristics; for example, wider time window for study entry correlated directly with time to treatment and inversely with stroke severi ty (initial National Institutes of Health Stroke Scale score). Baseline cha racteristics predicted outcome. Greater stroke severity predicted higher 3- month mortality rate; despite this, successful thrombolytic trials have enr olled more severe strokes than most trials. The mean age of enrollees also predicted 3-month mortality and was inversely related to percentage of pati ents with 3-month Barthel Index score greater than or equal to 95. The stro ngest predictors of 3-month mortality were obtained with multivariate model s. Conclusions-Acute stroke studies vary widely in entry criteria and outcome measures. Across multiple studies, differences in entry criteria, and the b aseline clinical characteristics they predict, influence patient outcomes a long a continuum. In some studies, enrolling a specific subset of patients may have improved the chances of identifying a treatment-related effect, wh ile in others, such chances may have been reduced. These findings may be us eful in the design of future stroke therapeutic trials.