To calculate how often non-valvular atrial fibrillation should be considered as the cause of ischaemic stroke rather than an incidental finding as in primary cerebral haemorrhage, the frequency of non-valvular atrial fibrillation in 75 consecutive patients with primary intracerebral haemorrhage was compared with that in 63 patients with cortical infarcts and 44 patients with lacunar infarcts who were taken at random from 1487 consecutive patients with cerebral infarcts admitted during the same period. The frequency of non-valvular atrial fibrillation in lacunar infarcts did not differ from that in primary intracerebral haemorrhage (odds ratio: 1.28; 95% confidence interval: 0.32-5.1) (P = 1.0), whereas it was significantly higher in cortical infarcts (odds ratio: 5.57; 95% confidence interval 2.00-15.4) (P = 0.0001). Our findings suggest that non-valvular atrial fibrillation in lacunar infarcts, as in primary intracerebral haemorrhage, is unlikely to be the cause of this type of stroke, whereas perhaps 30% of all cortical infarcts are not caused by non-valvular atrial fibrillation when present. These data are important for the planning and explanatory evaluation of trials on the preventive effect of anticoagulants in stroke patients with non-valvular atrial fibrillation.

We analyzed the outcome of 61 consecutive patients with presumed basilar artery thrombosis. Patients received anticoagulation unless a possibly favorable effect of such a treatment was not expected. Eight of 23 patients receiving anticoagulation died, against 30 of 38 patients not receiving this treatment (odds ratio 0.14; 95% CI: 0.04–0.47; p < 0.05). However, after adjusting for differences in age distribution and degree of initial deficit, no statistically significant difference remained. Our data do not argue in favor of or against anticoagulation treatment in patients with basilar artery thrombosis.