Short- and long-term exposure to ambient PM2.5 and risk of cardiovascular disease hospitalisation: a nationwide cohort study in the Netherlands

Background Fine particulate matter (PM 2.5 ) is an established cardiovascular risk factor, but uncertainties remain regarding the timing of risk across short‑ and long‑term exposure windows and specific cardiovascular diseases (CVDs). We evaluated associations between short‑ and long‑term PM 2.5 exposure and CVD hospitalisation in a nationwide population‑based cohort in the Netherlands. Methods All residents aged ≥30 years registered in the Netherlands on 1 January 2012 (N = 10,156,661) were followed through 2019. Daily PM 2.5 concentrations for the study period and the preceding six years were estimated at ∼2 × 2 km resolution and linked to residential histories via postal codes. Using a nested case‑control design (four controls per case, matched on age and sex) and conditional logistic regression, we estimated odds ratios (ORs) per 10 µg/m³ increase in PM 2.5 for moving‑average windows (short‑term lags 0-6 days; long‑term lags 0-6 years) and explored lag patterns with distributed lag models. Results Annual PM 2.5 concentrations in the Netherlands declined over time, from an average of 20 µg/m³ (SD=2.29) in 2006 to 13 µg/m³ (SD=0.89) in 2019 . We identified 592,259 first CVD hospitalisations, including 437,089 ischaemic heart disease (IHD), 205,663 stroke, and 151,345 myocardial infarction (MI) events. Short‑term PM 2.5 exposure (lag 0-6 days) was associated with increased risks of IHD (OR 1.058; 95% CI 1.041-1.075), MI (1.053; 1.025-1.083), and first CVD hospitalisation (1.047; 1.033-1.062), with no statistically significant association observed for stroke (1.013; 0.989-1.038). Long‑term exposure showed associations with IHD (1.338; 1.291-1.387), stroke (1.201; 1.138-1.267), MI (1.112; 1.046-1.182), and first CVD hospitalisation (1.278; 1.239-1.319). Conclusions Higher PM 2.5 exposure was consistently associated with increased CVD hospitalisation risk across both short‑ and long‑term windows, with outcome‑specific lag patterns and generally larger estimated associations for long- than short-term exposure.