The influence of randomized trials on the use of anticoagulants for atrial fibrillation.

INTRODUCTION: Anticoagulants and anti-platelet drugs have been shown in randomized trials to reduce the risk of stroke in patients with atrial fibrillation (AF). We therefore investigated their use in patients known to be in AF before a stroke, transient ischaemic attack (either cerebral or ocular) or retinal artery occlusion to assess the influence of trials on clinical practice. METHODS: Inpatients and outpatients with acute stroke, transient ischaemic attack or retinal artery occlusion were prospectively identified by a stroke physician from 1990 to 1997. The presence or absence of AF before the vascular event, and prior use of anticoagulant and anti-platelet drugs were recorded at the time of the assessment and verified using information from general practitioner and hospital case notes. RESULTS: Of 1934 patients with stroke or retinal artery occlusion, 191 (10%) were in AF before their ischaemic event. Anticoagulants had been used in 40 (21%) of these, but only in 32 (2%) of the 1743 patients in sinus rhythm [odds ratio (OR) 14.2, 95% confidence interval (CI) 8.6-23.2]. Anti-platelet drugs had been used in 62 (32%) of those with AF compared with 500 (30%) of those in sinus rhythm (OR 1.2, 95% CI 0.9-1.64). Of the 161 patients in AF without contraindications to anticoagulants, only 36 (22%) were taking them. Although there was a statistically significant increase in anticoagulant use from 8% in 1990 to 23% in 1996, this could be explained solely by a fall in the age of the patients referred to our hospital. CONCLUSION: Anticoagulation is probably under-used in AF. We found no conclusive evidence that anticoagulation trials have influenced clinical practice. This raises issues about the dissemination and implementation of trial results.