Spatially autocorrelated and clustered associations between short-term exposure to PM2.5 and daily reported confirmed COVID-19 cases across the contiguous United States, 2020–2022

Background Coronavirus disease 2019 (COVID-19) is a highly transmissible respiratory infection that has caused a global pandemic. Evidence for the effect of short-term PM 2.5 exposure on COVID-19 incidence varied across regions, and few studies have used advanced spatial analysis methods to examine spatial autocorrelation or clustering in these associations to improve estimation accuracy and identify highly susceptible geographic regions. Methods In this study, we used a novel three-stage strategy, including generalized additive model (GAM) and the newly developed estimation-error-based scan statistic (EESS) and multivariate conditional meta-autoregression (MCMAR), to fill the gap across the contiguous United States. Results We found there was significant spatial heterogeneity ( I 2 = 67.416 % , P < 0.001 ), autocorrelation ( ρ = 0.972 , P < 0.001 ), and clustering (four clusters were discovered, P ≤ 0.011 ) in the associations between short-term exposure to PM 2.5 and daily reported confirmed COVID-19 cases. Considering the spatial autocorrelation and the spatial clustering effectively improved the fitting results of meta-regression. The pooled relative risk (RR) and 95% confidence intervals (95% CI) for per 10 μg/m 3 increase in PM 2.5 were 1.0341 and (1.0308, 1.0375). The RRs (95% CIs) for per 10 μg/m 3 increase in PM 2.5 were 1.0107 (1.0074, 1.0140), 1.1770 (1.1551, 1.1992), 1.0005 (0.9984, 1.0026), 1.0719 (1.0566, 1.0875), and 1.1032 (1.0798, 1.1270) in baseline and clusters 1–4, respectively. By 30 June 2022, a total of 1,761,726 (95% CI: 1,583,970–1,916,779; accounting for 2.06%, 95% CI: 1.85–2.24%) COVID-19 cases were found to be related to elevated PM 2.5 exposure relative to the WHO annual mean PM 2.5 guideline threshold of 5 μg/m 3 across the contiguous United States. The eastern region and central region exhibited greater RR, greater attributable risk percentage, and more attributable cases than the western coastal region. Conclusion There was significant spatial heterogeneity, autocorrelation, and clustering in the associations between short-term exposure to PM 2.5 and daily reported confirmed COVID-19 cases across the contiguous United States.