Background Cardiovascular disease (CVD) event prediction (heart attack or stroke) is a classic time-to-event problem. Accurate prognostic models guide prevention and resource allocation. High-quality survival analysis requires attention to censoring/time dependence, competing risks, calibration, and generalizability. Primary aims Build and validate multivariable prognostic models predicting time to first major CVD event (composite: heart attack or stroke). Explore non-linear and time-dependent effects of predictors (age, sex, smoking, blood pressure, BMI, lung function). Account for competing risks (non-CVD death) using Fine–Gray model. Produce robust internal validation (bootstrap optimism correction) and temporal validation (train/test split by calendar time). Investigate causal effect of a binary “treatment” (e.g., antihypertensive therapy) using IPTW as sensitivity analysis. Provide clinical utility assessment (calibration, decision curves).