INTEGRATIVE ECOLOGICAL REGRESSION ANALYSIS OF US COUNTY AND STATE LEVEL COVID-19 DEATH DATA FOR STUDYING HEALTH DISPARITY ASSOCIATIONS

• 成果类型: Article
• 署名作者: Li, Daniel; Lin, Xihong
• 署名单位: Harvard University; Harvard T.H. Chan School of Public Health
• 刊物名称: ANNALS OF APPLIED STATISTICS
• ISSN/ISSBN: 1932-6157; 1941-7330
• DOI: 10.1214/25-AOAS2055
• 发表日期: 2025-09
• 页码: 2320-2338
• 关键词: covid-19; hybrid ecological data; integrative analyses; infectious disease; longitudinal data; Poisson mixed models; penalized composite log-likelihood; racial disparities; inference; mortality; models

摘要

It is of substantial interest to study health disparity associations withCOVID-19 death rates. Although high-quality individual-level COVID-19epidemiological data have been difficult to collect on a national scale, allUnited States (U.S.) counties have reported total COVID-19 death counts.A standard ecological analysis would then regress county total death countsby county-level covariates such as age, sex, and race percentages. However,such an analysis is limited by ecological bias and fallacy in which esti-mated county-level associations are different from individual-level associ-ations. Fortunately, state-level age, sex, and race specific COVID-19 deathcounts are also available for all U.S. states, so this information can be inte-grated with county-level data for more informative ecological analyses. Wepropose an approximate log-linear random effects model to jointly modelcounty-level total death counts and state-level age, sex, and race specificdeath counts. We then develop a penalized composite log-likelihood methodfor parameter estimation and perform simulation studies to evaluate our pro-posed approach. Lastly, we analyze COVID-19 death data from the entireU.S., show how incorporating state-level counts can prevent ecological biasand fallacy, and illustrate the heterogeneity in health disparity associationsacross different U.S. states.