Tracking Turnover Among Health Care Workers During the COVID-19 Pandemic: A Cross-sectional Study.

Importance: The health care sector lost millions of workers during the COVID-19 pandemic and job recovery has been slow, particularly in long-term care. Objective: To identify which health care workers were at highest risk of exiting the health care workforce during the COVID-19 pandemic. Design Setting and Participants: This was an observational cross-sectional study conducted among individuals employed full-time in health care jobs from 2019 to 2021 in the US. Using the data from the Current Population Survey (CPS), we compared turnover rates before the pandemic (preperiod, January 2019-March 2020; 71 843 observations from CPS) with the first 9 months (postperiod 1, April 2020-December 2020; 38 556 observations) and latter 8 months of the pandemic (postperiod 2, January 2021-October 2021; 44 389 observations). Main Outcomes and Measures: Health care workforce exits (also referred to as turnover) defined as a health care worker's response to the CPS as being unemployed or out of the labor force in a month subsequent to a month when they reported being actively employed in the health care workforce. The probability of exiting the health care workforce was estimated using a logistic regression model controlling for health care occupation, health care setting, being female, having a child younger than 5 years old in the household, race and ethnicity, age and age squared, citizenship status, being married, having less than a bachelor's degree, living in a metropolitan area, identifier for those reporting employment status at the first peak of COVID-19, and select interaction terms with time periods (postperiods 1 and 2). Data analyses were conducted from March 1, 2021, to January 31, 2022. Results: The study population comprised 125 717 unique health care workers with a mean (SD) age of 42.3 (12.1) years; 96 802 (77.0%) were women; 84 733 (67.4%) were White individuals. Estimated health care turnover rates peaked in postperiod 1, but largely recovered by postperiod 2, except for among long-term care workers and physicians. We found a 4-fold difference in turnover rates between physicians and health aides or assistants. Rates were also higher for health workers with young children (<5 years), for both sexes and highest among women. By race and ethnicity, persistently higher turnover rates were found among American Indian/Alaska Native/Pacific Islander workers; White workers had persistently lower rates; and Black and Latino workers experienced the slowest job recovery rates. Conclusions and Relevance: The findings of this observational cross-sectional study suggest that although much of the health care workforce is on track to recover to prepandemic turnover rates, these rates have been persistently high and slow to recover among long-term care workers, health aides and assistants, workers of minoritized racial and ethnic groups, and women with young children. Given the high demand for long-term care workers, targeted attention is needed to recruit job-seeking health care workers and to retain those currently in these jobs to lessen turnover.