HOW GOOD ARE VOLUNTEERS AT SEARCHING FOR PUBLISHED RANDOMIZED CONTROLLED TRIALS

The Ottawa Stroke Trials Registry (OSTR) collects information on all r andomized controlled trials (RCTs), in all languages, in which the pri mary focus is stroke. Published trials are added to the registry datab ase once they have been identified by volunteers hand searching journa ls. We undertook this study to assess volunteer sensitivity, specifici ty, positive and negative predictive values, accuracy and reliability in identifying RCTs. Three volunteers participated in this study. The complete contents of five years of one journal were searched (n = 240) as were ten research reports from five journals. Volunteer sensitivit y, specificity, positive and negative predictive values, and accuracy was 89.5%, 100%, 100%, 99.1%, and 99% respectively. Inter volunteer re liability was 0.87. Searching the published literature to identify RCT s is an enormous task. These results indicate that volunteers, with mi nimal training, can effectively contribute to this endeavour.