BACKGROUND: Patients with stroke due to spontaneous (non-traumatic) intracerebral haemorrhage (ICH) are at risk of recurrent ICH, ischaemic stroke, and other serious vascular events. We aimed to analyse these risks in population-based studies and compare them with the risks in RESTART, which assessed antiplatelet therapy after ICH. METHODS: We pooled individual patient data from two prospective, population-based inception cohort studies of all patients with an incident firs-in-a-lifetime ICH in Oxfordshire, England (Oxford Vascular Study; April 1, 2002, to Sept 28, 2018) and Lothian, Scotland, UK (Lothian Audit of the Treatment of Cerebral Haemorrhage; June 1, 2010, to May 31, 2013). We quantified the absolute and relative risks of recurrent ICH, ischaemic stroke, or any serious vascular event (non-fatal stroke, non-fatal myocardial infarction, or vascular death), stratified by ICH location (lobar vs non-lobar) and comorbid atrial fibrillation (AF). We compared pooled event rates with those after allocation to avoid antiplatelet therapy in RESTART. FINDINGS: Among 674 patients (mean age 74·7 years [SD 12·6], 320 [47%] men) with 1553 person-years of follow-up, 46 recurrent ICHs (event rate 3·2 per 100 patient-years, 95% CI 2·0-5·1) and 25 ischaemic strokes (1·7 per 100 patient-years, 0·8-3·3) were reported. Patients with lobar ICH (n=317) had higher risk of recurrent ICH (5·1 per 100 patient-years, 95% CI 3·6-7·2) than patients with non-lobar ICH (n=355; 1·8 per 100 patient-years, 1·0-3·3; hazard ratio [HR] 3·2, 95% CI 1·6-6·3; p=0·0010), but there was no evidence of a difference in the risk of ischaemic stroke (1·8 per 100 patient-years, 1·0-3·2, vs 1·6 per 100 patient-years, 0·6-4·4; HR 1·1, 95% CI 0·5-2·8). Conversely, there was no evidence of a difference in recurrent ICH rate in patients with AF (n=147; 3·3 per 100 patient-years, 95% CI 1·0-10·7) compared with those without (n=526; 3·2 per 100 patient-years, 2·2-4·7; HR 0·9, 95% CI 0·4-2·1), but the risk of ischaemic stroke was higher with AF (6·3 per 100 patient-years, 3·7-10·9, vs 0·7 per 100 patient-years, 0·1-5·6; HR 8·2, 3·3-20·3; p<0·0001), resulting in patients with AF having a higher risk of all serious vascular events than patients without AF (15·5 per 100 patient-years, 10·0-24·1, vs 6·8 per 100 patient-years, 3·6-12·5; HR 1·78, 95% CI 1·16-2·74; p=0·0090). Only for patients with lobar ICH without comorbid AF was the risk of recurrent ICH greater than the risk of ischaemic stroke (5·2 per 100 patient-years, 95% CI 3·6-7·5, vs 0·9 per 100 patient-years, 0·2-4·8; p=0·00034). Comparing data from the pooled population-based studies with that from patients allocated to not receive antiplatelet therapy in RESTART, there was no evidence of a difference in the rate of recurrent ICH (3·5 per 100 patient-years, 95% CI 1·9-6·0, vs 4·4 per 100 patient-years, 2·6-6·1) or ischaemic stroke (3·4 per 100 patient-years, 1·9-5·9, vs 5·3 per 100 patient-years, 3·3-7·2). INTERPRETATION: The risks of recurrent ICH, ischaemic stroke, and all serious vascular events after ICH differ by ICH location and comorbid AF. These data enable risk stratification of patients in clinical practice and ongoing randomised trials. FUNDING: UK Medical Research Council, Stroke Association, British Heart Foundation, Wellcome Trust, and the National Institute for Health Research Oxford Biomedical Research Centre.
437 - 447