Long-Term Coexposure to Multiple Air Pollutants and Stroke Incidence: A Large Prospective Cohort Study Based on a Shape-Constrained Nonlinear Weighted-Sum Model

No prior study has examined the flexible association between stroke incidence and long-term coexposure to multiple air pollutants within an interpretable framework or assessed coexposure attributable burden and intervention benefits. In this study, based on a large prospective cohort of 469,174 Chinese adults (2013-2018) and the annual exposures to PM_2.5, SO₂, NO₂, CO, and O₃, we developed a shape-constrained nonlinear weighted-sum model within a time-varying survival framework to characterize multipollutant-stroke associations, visualized via nonlinear nomograms. We calculated attributable cases, population attributable fraction (PAF), and benefits of 10% pollutant reductions across China in 2013-2022. Our model identified PM_2.5, SO₂, NO₂, and CO as risk factors for ischemic stroke (decreasing importance), showing an S-shaped coexposure-response association. No adverse effect was observed for hemorrhagic stroke. Ischemic stroke PAF declined from 60.8% (50.0-69.3%) in 2013 to 36.7% (24.5-46.9%) in 2022, with stable annual attributable cases (∼1080.6 thousand). Intervention benefits varied by regions and years, e.g., traditional high-pollution areas showed significant population benefits in 2013 but limited benefits in 2019, while South China exhibited the opposite trend. Our results suggest that reducing air pollution remains the cornerstone of stroke prevention, and region-specific precise strategies─guided by our nomogram─are now needed.