Abstract Background and aims While national epidemiological data provide estimates of overall stroke incidence, stroke risk may vary substantially across regions, and the drivers of geographic variation in Taiwan remain unclear. Methods Nationwide stroke incidence data (ischemic stroke, intracerebral hemorrhage, and subarachnoid hemorrhage) from 349 administrative districts in Taiwan were analyzed using the National Health Insurance Research Database for 2021. To account for unstable coefficients from administrative districts with small populations, we merge adjacent small districts such that all districts have a population 10, 000. Standardized stroke ratios (SSRs) were calculated by dividing the observed and expected cases for the region. SSR 1. 5 was defined as hotspot. Bayesian spatial regression models using the integrated nested Laplace approximation (INLA) were adopted, with covariates including the proportion of the population aged ≥ 65 years, sex, hypertension, and diabetes. Results Significant spatial clustering was observed (Moran’s I = 0. 38, p 0. 001), and spatial models substantially outperformed non-spatial models. Among them, the intrinsic conditional autoregressive (ICAR) Model demonstrated the best fit and an interactive visualization of the results is available at: https: //albert-l-phan. github. io/TaiwanStroke/. Among covariates, hypertension and diabetes mellitus were strongly associated with higher stroke burden (coefficients 4. 72 and 4. 05, respectively with P values 0. 05). After accounting for spatial dependence and chronic disease prevalence, the proportion of the population aged ≥ 65 years and sex were no longer statistically significant. Conclusions The stroke incidence in Taiwan shows marked geographic clustering, providing a quantitative basis for allocating public health expenditure and healthcare resources to support more equitable stroke care planning. Conflict of interest All authors: Nothing to disclose
Sung et al. (Fri,) studied this question.