A COVID-19 Risk Assessment for the US Labor Force

Sammantha Maher; Alexandra E Hill; Peter Britton; Eli P Fenichel; Peter Daszak; Carlos Zambrana-Torrelio; Jude Bayham

Microdata is aggregated into summary statistics for each region, industry, and occupation using the srvyr package in R (24). First, we sum the number of survey respondents (n) in each 5 region-industry-occupation category. We do the same for the number of people in each regionindustry-occupation category for each social and health risk factor. Next, we use the sampling weight variable (sampweight) provided by IPUMs to map each survey respondent to the number of people in the United States workforce that their response represents, based on their demographic characteristics (6) . The region, occupation, and industry categories in IPUMS are 10 unique and mutually exclusive for each respondent, so they can be summed to get the number of people in different permutations of region, industry, and occupation. However, each survey respondent can have multiple health and social risk factors present, and those categories are not additive. The srvyr package provides variance estimates based on the stratified sample design of the NHIS (24). 15 The risk data were then merged with data on county-level employment from the Quarterly Census of Earnings and Wages (QCEW) (7). We use the 2018 annual averages QCEW NAICS-based data files to obtain annual estimates of employment by industry (at the 3-digit NAICS level) for all reporting counties in the U.S. These data represent the number of workers who are covered by Unemployment Insurance for each employer in the county. This does not 20 count self-employed workers and unpaid family workers, and might double-count workers who are employed by multiple firms within the year. Importantly, these data do not estimate the number of workers in the workforce, but rather the average number of jobs in each industry . CC-BY-NC-ND 4.0 International license It is made available under a author/funder, who has granted medRxiv a license to display the preprint in perpetuity.