Illness-Related Work Absence in Mid-April Was Highest on Record

Abstract

This survey study analyzes data from the Current Population Survey to compare trends in work absence over the first 4 months of 2020 relative to 2019 to shed light on the health and ecenomic effects of the coronavirus disease 2019 pandemic.

Data on diagnosed cases and deaths have been used to delineate the course of the coronavirus disease 2019 (COVID-19) pandemic. Information from population employment surveys could shed additional light on the pandemic's effect on the health and behavior of the nation’s workforce.

Methods

We analyzed the Current Population Survey (CPS), a monthly survey of approximately 115 000 persons that collects information on employment.¹ We classified as “out sick” respondents who reported having a job but being absent the previous week due to their “own illness/injury/medical problems.” We compared trends over the first 4 months of 2020 relative to 2019 using multivariable linear regressions adjusted for age, sex, race/ethnicity, education, year, month, and a month×year interaction term. We assessed the demographic characteristics of out-sick jobholders, and compared out-sick rates in “high–” vs “nonhigh–COVID-19” states,² and states with and without laws mandating paid sick leave.³ We also assessed the number of persons out sick each month since January 1976.

We used STATA/SE statistical software (version16.1, STATA Corp), SAS statistical software (version 9.4; SAS Institute, Inc), survey weights to produce national estimates, and Davern's method to calculate approximate standard errors.⁴ This analysis of publicly available, deidentified data is not considered human subjects research per the Cambridge Health Alliance institutional review board, and is hence exempt from review or informed consent requirements. Analyses were performed May 15 to 18, 2020.

Results

Employment was stable at approximately 156 to 158 million from January 2019 to March 2020, but fell to 133.7 million in April 2020. Of jobholders, 1.1 million were out sick in January and February of both 2019 and 2020 (Figure). In 2019, the numbers decreased steadily to 0.92 million (0.58% of jobholders) in April. Trends differed in 2020, rising in March and hitting 2.02 million (1.51% of jobholders) in April, when 1.10 million more workers (an adjusted increase of 0.95% of the workforce) were out sick than in April 2019. More workers were out sick in April 2020 than in any month since January 1976, the earliest month for which such data were available.

Figure. Number of Persons With Jobs Who Were Absent From Work Due to Illness or Injury, January to April, 2019 and 2020.

All demographic groups experienced increased rates of sickness-related work absence (Table). However, the increments, relative to 2019, were significantly larger for immigrants (adjusted difference 1.28% vs 0.87% among nonimmigrants), workers aged 55 years or older (1.69% vs 0.71% in younger workers), and workers without college education (1.46% vs 0.41% among those with a bachelor’s degree or higher). Increments were similar in states with and without paid sick-leave laws.

Table. Persons With Jobs Who Were Absent From Work Due to Illness or Injury, Mid-April 2019, and Mid-April, 2020.

Characteristic	Unadjusted, No. (%)			Adjusted change: January to April, 2020 vs 2019a		Adjusted difference: April 2020 vs April 2019 by subgroupb
	2019	2020	Difference	Adjusted difference (95% CI), %	P value	Adjusted difference (95% CI), %	P value
All persons with jobs	917 911 (0.58)	2 017 105 (1.51)	1 099 194 (0.93)	0.95 (0.75 to 1.14)	<.001	NA
Sex
Male	444 642 (0.53)	964 612 (1.34)	519 970 (0.81)	0.85 (0.61 to 1.10)	<.001	1 [Reference]
Female	473 269 (0.64)	1 052 492 (1.70)	579 223 (1.06)	1.05 (0.76 to 1.34)	<.001	0.27 (−0.03 to 0.57)	.08
Age, y
<55	594 950 (0.50)	1 211 291 (1.20)	616 341 (0.70)	0.71 (0.51 to 0.92)	<.001	1 [Reference]
≥55	322 961 (0.86)	805 814 (2.45)	482 853 (1.59)	1.69 (1.22 to 2.15)	<.001	0.87 (0.45 to 1.28)	<.001
Immigration status
Not immigrant	798 552 (0.63)	1 571 707 (1.43)	773 155 (0.81)	0.87 (0.67 to 1.08)	<.001	1 [Reference]
Immigrant	119 359 (0.40)	445 398 (1.86)	326 039 (1.46)	1.28 (0.79 to 1.77)	<.001	0.68 (0.22 to 1.13)	.004
Race
Non-Hispanic
White	567 819 (0.58)	1 224 419 (1.45)	656 600 (0.87)	0.85 (0.62 to 1.08)	<.001	1 [Reference]
Black	159 998 (0.88)	294 113 (1.94)	134 115 (1.06)	1.31 (0.58 to 2.04)	<.001	0.18 (−0.47 to 0.83)	.59
Hispanic	129 649 (0.47)	333 546 (1.48)	203 897 (1.01)	1.02 (0.53 to 1.52)	<.001	0.14 (−0.33 to 0.60)	.56
Other	60 445 (0.45)	165 027 (1.41)	104 582 (0.97)	0.97 (0.42 to 1.51)	<.001	0.13 (−0.39 to 0.64)	.63
Education
High school or less	429 893 (0.81)	912 340 (2.25)	482 447 (1.45)	1.46 (1.05 to 1.86)	<.001	0.94 (0.56 to 1.32)	<.001
Some college	294 305 (0.67)	629 749 (1.78)	335 444 (1.11)	1.20 (0.81 to 1.59)	<.001	0.59 (0.22 to 0.96)	.002
College or more	193 714 (0.32)	475 015 (0.82)	281 301 (0.50)	0.41 (0.19 to 0.63)	<.001	1 [Reference]
Paid leave statec
No	642 823 (0.56)	1 525 292 (1.56)	882 469 (1.00)	0.96 (0.73 to 1.19)	<.001	1 [Reference]
Yes	275 089 (0.66)	491 812 (1.38)	216 723 (0.72)	0.90 (0.53 to 1.27)	<.001	−0.28 (−0.64 to 0.08)	.13
High-COVID stated
No	730 002 (0.56)	1 588 813 (1.42)	858 811 (0.86)	0.87 (0.66 to 1.07)	<.001	1 [Reference]
Yes	187 909 (0.70)	428 291 (1.98)	240 382 (1.28)	1.36 (0.83 to 1.90)	<.001	0.44 (−0.07 to 0.95)	.09
Health care worker
No	830 300 (0.59)	1 742 431 (1.47)	912 131 (0.88)	0.89 (0.69 to 1.10)	<.001	1 [Reference]
Yes	87 611 (0.51)	274 674 (1.79)	187 063 (1.27)	1.36 (0.78 to 1.95)	<.001	0.39 (−0.11 to 0.88)	.13

^a Adjusted for age (years), sex, race-ethnicity (White, Black, Hispanic, other), education (high school or less, some college; bachelors or more), year, month, and year×month interaction term. The estimates are the coefficients of the interaction term comparing April to January, in 2020 relative to 2019.

^b Adjusted for age (years), sex, race/ethnicity (White, Black, Hispanic, other), education (high school or less, some college; bachelors or more), year, and an interaction term between year and the subgroup indicator. Analyses confined to April data. The estimates are the coefficients of the interaction term comparing out sick in April 2020 relative to April 2019 for that subgroup relative to the referent group.

^c Paid leave states include: Arizona, California, Connecticut, DC, Maryland, Massachusetts, New Jersey, New York, Oregon, Rhode Island, Vermont, and Washington. New York is excluded because benefits were not effective until January 2020.³

^d High COVID-19 states included Connecticut, DC, Louisiana, Massachusetts, Michigan, New Jersey, New York, and Rhode Island, the 8 states with the highest per capita COVID-19 cases. Other states had substantially lower incidence.²

Discussion

Work absence due to illness rose to record levels in mid-April 2020 coincident with the peak of COVID-19 hospitalizations and deaths. The excess of persons out sick compared with 2019 was about 5-fold greater than the number of cases of COVID-19 diagnosed that week.⁴ Although nationwide data on the demographic characteristics of hospitalized patients with COVID-19 are not available, our findings are consonant with data from some locales indicating high illness rates among racial minorities. The inability of low-income workers to telecommute could also have contributed to the apparent disparities we observed.

We had no data on what illnesses caused absences. Although the confidential survey offered respondents 13 non–illnesses-related options—including child care—as the reason for work absence, some jobholders who stayed home to care for children or others may have attributed their absenteeism to their own illness. Publicity around COVID-19 may have caused workers with non–COVID-19–related symptoms, or anxiety, to stay home. Increasing options for telecommuting and economic duress, however, might have the opposite effect. Similarly, reductions in other viral illnesses due to social distancing, and decreased injuries and air-pollution-related illnesses, might have cut work absences from non–COVID-19 illnesses.

Our findings shed light on the combined health and economic effects of the COVID-19 pandemic, particularly for immigrant, older, and less-educated workers. Finally, our study suggests that routinely collected CPS data on work absence may provide a rapidly available tool for surveillance of the effect of public health crises on the workforce.