Balancing the risk of major bleeding against vascular disease risk in people without atherosclerotic disease

Background and aims In the primary prevention setting, low-dose aspirin reduces major vascular events (MVEs) by approximately 11% but increases major bleeding (MB) by 40-50%, implying that net benefit will be most evident when the MVE-to-MB ratio is >4. This study aimed to derive cross-validated risk scores for MB and MVE and use the MVE-to-MB ratio to identify groups who may derive differing net benefits from treatment. Methods 431 167 UK Biobank participants without known atherosclerotic cardiovascular disease at baseline were followed through record linkage for incident MVEs (myocardial infarction, non-haemorrhagic stroke, transient ischaemic attack, arterial revascularisation or vascular death) and MB (gastrointestinal and intracranial bleeds with hospital admission for >= 2 days). Risk scores were derived for MVE and MB using Cox proportional hazards models with cross-validation. Ratios of observed MVE-to-MB rates were calculated across risk categories. Results During a median follow-up of 12 years, 18 310 participants suffered an MVE and 5352 an MB. MB risk was highest among participants with frailty, prior bleeds, cancer, liver disease or renal dysfunction, with a 4.3-fold difference in risk between the highest and lowest fifths of MB risk (HR 4.3, 95% CI 3.87 to 4.77). The MVE-to-MB ratio was <= 2.6 in the highest MB risk groups and >= 4 in lower MB risk categories. Conclusions The derived models using routinely available disease history and laboratory measurements improved distinction of the MVE-to-MB ratio compared with using conventional models for MB risk including vascular risk factors. Such models can help identify those with moderate MVE risk but low MB risk who may benefit from low-dose aspirin.