328-OR: County-Level Diabetes Incidence among U.S. Adults in 2019: Timely Small-Area Estimates from a New Model

Abstract

Incident cases of diabetes among survey participants are defined as those diagnosed within the proceeding12 months. Because few cases may be observed each year, annual county-level estimates of incidence have required multiple years of data, delaying their release. We developed a new model, Bayesian weighted Binomial Zero-inflated (BBZ), to deliver timely estimates based on the most recent year of data while incorporating sampling weights and prior-year data and accounting for counties with “excess” zero cases. BBZ models were validated with American Community Survey county-level 1-year reports. We used 2019 Behavioral Risk Factor Surveillance System and US Census projections to estimate county-level incidence of diagnosed diabetes among US adults ≥20 years in 2019. In 2019, age-adjusted diabetes incidence ranged from 2.7 (95% CI: 2.6-2.8) per 1000 persons to 15.8 (95% CI: 15.7-1.60) per 1000 persons among 3121 counties (Figure). The national rate, calculated by aggregating county-level estimates, was 6.8 (95% CI: 6.5-7.0) per 1000 persons. National incidence among adults ≥65 years was over 4.5 times higher than those aged 20 to 44; there was no statistical difference by sex. The new BBZ modeling approach facilitates timely release of annual county-level estimates and prompt evaluation of disparities in the incidence of diagnosed diabetes. Disclosure H. Xie: None. D. B. Rolka: None. S. R. Benoit: None. K. M. Bullard: None.