A Novel Approach for Estimating Prevalence and Incidence for 100+ Health Conditions by Race/Ethnicity and County in the United States

Background: In the United States (US), some race and ethnicity (R/E) groups have suffered from lower-quality health access and from under-ascertainment of their health burden. Meanwhile White patients have been over-represented in clinical trials. New US Federal legislation requires clinical trial sponsors to submit diversity plans for population-representative enrollment. However, universal high-quality health burden enumeration by R/E in the US only exists for deaths. Data sources for prevalence and incidence are fragmentary.

Objectives: Obtain reliable approximations of local US prevalence and incidence by R/E.

Methods: We combined county-level condition-specific mortality estimates and population counts by R/E, generated by the US Health Disparities team (USHD) of the Institute for Health Metrics and Evaluation (IHME), together with national condition-specific estimates of incidence and prevalence generated by IHME’s ongoing Global Burden of Disease study. This enables indirect prevalence/incidence estimates with uncertainty bounds by R/E, age, and sex, for five standard R/E groups (US OMB 1977) and 3,000+ US counties.
This approach assumes similar case-fatality ratios (CFRs) between different R/E groups, holding all other variables equal. This is a proactive assumption: it leads to higher prevalence/incidence estimates for R/E groups that tend to be missed by the health system, compared to raw reported rates.
We included all fatal health conditions modeled at the R/E and county level by IHME, except broad condition categories composed of heterogeneous disorders with disparate CFRs and demographic patterns.

Results: A total of 112 and 108 conditions were compatible with our estimation approach for prevalence and incidence, respectively. Nationally in 2019, sickle cell disorders had the greatest age-standardized disparities: Blacks experienced ~40x higher burden than Asian and Pacific-Islanders (A+PI). The non-genetic condition with greatest disparities was environmental heat and cold exposure: Native American and Alaskan Native (AIAN) males suffered ~30x the burden of A+PI males, and ~8x the all-groups average. For lung cancer, AIAN females had the highest burden at ~3x the lowest-burden group (Hispanic/Latine), while among males Blacks had the highest burden, ~2.4x higher than Hispanic/Latine. Maps visualizing geographic disparities will also be presented.

Conclusions: This approach helps to quantify disease incidence and prevalence in the US for a wide variety of health conditions, at an unprecedented demographic and geographic resolution. It is useful for clinical trial planning and can serve as a benchmark for specific disparity-estimation efforts. It can be adapted to other countries with high-quality mortality data.