(P-317) Experimenting with County-level COVID-19 Data as a Substitute for Demographic Information

Background: Two major challenges during the COVID-19 pandemic have been (1) the lack of unified and accessible national and regional data on demographics for cases and deaths and (2) missing data on race and ethnicity in state-level databases. Public county-level COVID-19 case data could be combined with county-level demographic information, but it is unclear whether such an approach accurately estimates the demographic distribution of COVID-19 cases and deaths.

Objectives: Assess whether combining county-level census data with public county-level data on COVID-19 cases and deaths produces similar estimates of the race, ethnicity, sex, and age distribution of COVID-19 cases and deaths obtained from North Carolina state-level data.

Methods: We merged the census-estimated July 2019 county-level demographics by age, sex, race, and ethnicity from the United States Census Bureau with the county-level COVID-19 case and death counts from the databases administered by Johns Hopkins University on January 6, 2021. We used this merged dataset to estimate demographic distributions among COVID-19 cases and deaths in North Carolina, assuming cases and deaths were randomly allocated at the county level (method 1, M1). We compared these distributions with state-level data from the North Carolina Department of Health and Human Service (NCDHHS).

Results: 18.6% of cases and 4.7% of deaths in the NCDHHS data were missing race while 35.8% and 13.7% were missing ethnicity, respectively. When comparing the proportion of COVID-19 cases by various demographics, estimates from M1 were similar to those from NCDHHS for female sex (51.4% with M1 vs 53.3% of cases with NCDHHS), the proportion over 75 years old (6.7% vs 6.9%), White race (71.8% vs 72.2%), and Black race (23.1% vs 23.8%). Several groups differed, however: these included those aged 0-24 (31.4% vs 25%), Asian and Pacific Islander race (3.2% vs 2.1%), and those reporting Hispanic ethnicity (9.9% vs 24%). For COVID-19 related deaths, estimates from M1 were similar to NCDHHS for a few groups, including White race (71.5% vs 69.3%) and Hispanic ethnicity (9.4% vs 8.1%). As expected, deaths were not similarly distributed by age (e.g. age-group 0-24: 31% vs 0.14%), Black race (23.8% vs 28.1%), or female sex (51.4% vs 47.9%).

Conclusions: Our investigation provides further evidence of COVID-19 disparities within demographic subgroups after accounting for geographic location. That said, combining public data appears to appropriately track cases for some demographic subgroups. By harmonizing public county-level data, we may be able to reasonably track some demographic trends in the distribution of COVID-19 at the national and regional levels.