Using computers, the researchers searched the medical literature to identify all studies of antithrombotic medications to prevent stroke for patients with atrial fibrillation. They included only randomized, controlled trials where patients received warfarin, aspirin, or no treatment. In a randomized, controlled trial, each patient has an equal chance of receiving one of the treatments. Studies are designed like this when it really is not clear that one treatment is better than another or better than no treatment at all. Using a statistical method called meta-analysis, the researchers combined the findings of the 16 separate studies.