Trends in Labor Unionization Among US Health Care Workers, 2009-2021

Key Points

Question

What was the prevalence of labor unionization among US health care workers over the past 12 years, and was it associated with pay and benefits?

Findings

In this cross-sectional study of 14 298 US health care workers, the prevalence of reported labor unionization was 13.2%, with no significant change from 2009 through 2021. Reported membership or coverage by a labor union was significantly associated with higher weekly earnings and better noncash benefits but greater number of weekly work hours.

Meaning

From 2009 through 2021, labor unionization among US health care workers remained low.

Abstract

Importance

Labor unionization efforts have resurged in the US, and union membership has been shown to improve worker conditions in some industries. However, little is known about labor unionization membership and its economic effects across the health care workforce.

Objectives

To examine the prevalence of labor unionization among health care workers and its associations with pay, noncash benefits, and work hours.

Design, Setting, and Participants

This cross-sectional study was conducted using data from the Current Population Survey and Annual Social and Economic Supplement from 2009 through 2021. The US nationally representative, population-based household survey allowed for a sample of 14 298 self-identified health care workers (physicians and dentists, advanced practitioners, nurses, therapists, and technicians and support staff).

Exposures

Self-reported membership status or coverage in a labor union.

Main Outcomes and Measures

Prevalence and trend in labor unionization. Further comparisons included mean weekly pay, noncash benefits (pension or other retirement benefits; employer-sponsored, full premium–covered health insurance; and employer’s contribution to the worker’s health insurance plan), and work hours.

Results

The 14 298 respondents (81.5% women; 7.1% Asian, 12.0% Black, 8.5% Hispanic, 70.4% White individuals; mean [SD] age, 41.6 [13.4] years) included 1072 physicians and dentists, 981 advanced practitioners, 4931 nurses, 964 therapists, and 6350 technicians and support staff. After weighting, 13.2% (95% CI, 12.5% to 13.8%) of respondents reported union membership or coverage, with no significant trend from 2009 through 2021 (P = .75). Among health care workers, those who were members of a racial or ethnic minority group (Asian, Black, or Hispanic individuals compared with White individuals) and those living in metropolitan areas were more likely to report being labor unionized. Reported unionization was associated with significantly higher reported weekly earnings ($1165 vs $1042; mean difference, $123 [95% CI, $88 to $157]; P < .001) and higher likelihood of having a pension or other retirement benefits at work (57.9% vs 43.4%; risk ratio [RR], 1.33 [95% CI, 1.26 to 1.41]; P < .001) and having employer-sponsored, full premium–covered health insurance (22.2% vs 16.5%; RR, 1.35 [95% CI, 1.17 to 1.53]; P < .001). Union members reported more work hours (37.4 vs 36.3; mean differences, 1.11 [95% CI, 0.46 to 1.75]; P < .001) per week. White workers reported mean weekly earnings that were significantly more than members of racial and ethnic minority groups among nonunionized workers ($1066 vs $1001; mean difference, $65 [95% CI, $40 to $91]; P < .001), but there was no significant difference between the 2 groups among unionized workers ($1157 vs $1170; mean difference, −$13 [95% CI, −$78 to $52]; P = .70).

Conclusions and Relevance

From 2009 through 2021, labor unionization among US health care workers remained low. Reported union membership or coverage was significantly associated with higher weekly earnings and better noncash benefits but greater number of weekly work hours.

This cross-sectional study examines the effect of unionization on pay, noncash benefit, and working hours among health care workers.

Introduction

Labor unionization efforts have recently resurged throughout the US, with the National Labor Relations Board receiving a 57% increase in union election petitions in the first half of 2022.¹ Workers in multiple industries, including those in health care, are unionizing to bargain for better pay, better noncash benefits, and safer work conditions.^2,3 For health care workers, the toll of the COVID-19 pandemic—including struggles obtaining personal protective equipment, inconsistent testing and notification of COVID-19–positive exposures, and inadequate pay with increased work hours—against the backdrop of increasing burnout prior to the pandemic has amplified calls for labor unionization to improve working conditions in the US health care system.⁴

Although labor unions have been shown to improve working conditions in other industries,⁵ empirical evidence about their role in the health care workforce is limited. Previous studies reported improved workplace safety and little effect on worker well-being; however, the studies were limited to specific populations (eg, surgical residents) or care settings (eg, nursing homes).^6,7 No study, to our knowledge, has systematically investigated labor unions and their economic effects across the health care workforce. It remains unclear how labor unionization in health care has changed over the years and what benefits, if any, health care workers gain from unionizing. To bridge this gap, the prevalence of labor unionization among health care workers and its associations with employee pay, noncash benefits, and work hours across the health care workforce were examined.

Methods

This cross-sectional study was exempt from review and informed consent by the institutional review board at Harvard Pilgrim Health Care Institute because of the use of a publicly available, deidentified data set.

Study Population and Data Source

Data from the US Census Bureau–sponsored Current Population Survey (CPS) outgoing rotation group and Annual Social and Economic Supplement (ASEC) were used.⁸ The CPS is a nationally representative survey administered to 60 000 US households monthly. Households are surveyed for 4 consecutive months, given a break for 8 months and then sampled for another 4 months before leaving the sample permanently.⁹ Households in the 4th (before break) or 16th month (before leaving the sample) are considered the “outgoing rotation group.” Within this group, those aged 15 years or older who are currently employed as a wage or salaried worker (ie, not self-employed or practice owner) are asked additional labor questions, including their union membership. In the analysis, unionized workers were defined as those who reported labor union membership or coverage (ie, who reported being covered by a union but not being a member); nonunionized workers were defined as those who reported no union membership or coverage. The CPS survey collected information on sociodemographic characteristics, including age, sex, race and ethnicity, education, occupation, residence, and region. Race and ethnicity were included in the analyses because differences in compensation and union membership exist between racial and ethnic groups.^10,11 Race and ethnicity data were self-reported by participants choosing from fixed categories, which were further categorized into 5 groups in this study: Asian, Hispanic, non-Hispanic Black, non-Hispanic White, and other (including American Indian, multiracial, and other unspecified). Five groups of health care workers (physicians and dentists, advanced practitioners, nurses, therapists, and technicians and support staff) were categorized based on occupational codes (eTable 1 in Supplement 1).

The ASEC is administered to CPS participants during the months of February, March, or April each year, and collects supplementary data on income (eg, mean weekly earnings), noncash benefits (eg, pension or other retirement benefits, employer-sponsored full premium–covered health insurance, and employer’s contribution to the worker’s health insurance plan), and work hours (eg, mean hours worked per week),¹² which were the outcomes of interest. CPS outgoing rotation group members who self-reported as health care workers and also answered the ASEC questions from 2009 through 2021 were included in this study. ASEC response rates in the study period ranged from 61.1% to 85.9%.¹³

Statistical Analysis

The overall and annual prevalence of self-reported labor union coverage among US health care workers was examined throughout the 2009-2021 study period. To assess whether there was an increasing or decreasing trend of labor unionization across the last decade, trend analysis was conducted using the Cox-Stuart trend test.¹⁴ The prevalence of unionization was estimated by state and types of health care workers. The prevalence odds ratios and their 95% CIs for being a unionized vs nonunionized worker by sociodemographic groups were estimated using multivariable logistic regression.

To evaluate associations between unionization with pay, noncash benefits, and work hours, each outcome was regressed on unionization status in separate models, adjusting for sociodemographic characteristics (age, sex, race and ethnicity, state of residence, metropolitan or rural area, central city status, occupation, education, mean weekly work hours, public or private sector employee). Linear regression was used for pay (ie, mean weekly earnings), employers’ contributions to workers’ health insurance plan, and work hours outcomes, and logistic regression for noncash benefit outcomes to estimate multivariable models and marginal means with 95% CIs. All dollars were inflation-adjusted to 2020 US dollars using the consumer price index.

Disparities in pay among sex and racial and ethnic subgroups have been documented in the literature.^10,11,15 Two-sided t tests were used to test for interactions between union membership and mean weekly earnings by sex and by race and ethnicity groups. Because of the paucity of studies in diverse populations, stratified analyses were prespecified and conducted by sex (male; female) and by racial and ethnic groups (Asian, Hispanic, or non-Hispanic Black; non-Hispanic White).

Several sensitivity analyses were conducted. To examine the potential different effect of union membership and union coverage, individuals who reported being covered by a labor union but not being a member were excluded from the unionized group. To provide a more generalized assessment of health insurance benefits, analyses were conducted to examine the likelihood of having employer-sponsored partial or full premium–covered health insurance plans compared with none. To determine whether there remains residual influence of the cost of living on the association between unionization and pay (ie, mean weekly earnings) besides state of residence and central city status, analyses were conducted in the subsample of 2000 respondents living in major metropolitan areas.

All estimates were weighted to be nationally representative and correct for nonresponse bias. Less than 0.5% of data were missing for weekly earnings; therefore, missing data were classified as unknown in our model for the pay outcome. Throughout, a 2-sided P < .05 was considered statistically significant. Because of the potential for type I error due to multiple comparisons, findings should be interpreted as exploratory. All analyses were performed using R version 3.6.3 (R Foundation).

Results

The analytical sample comprised 14 298 US health care workers (81.5% women; 1021 Asian [7.1%], 1222 Hispanic [8.5%], 1719 Non-Hispanic Black [12.0%], and 10 066 Non-Hispanic White [70.4%]; mean [SD] age, 41.6 [13.4] years) who responded to the survey from 2009 through 2021: 1072 physicians and dentists (7.5%), 981 advanced practitioners (6.9%), 4931 nurses (34.5%), 964 therapists (6.7%), and 6350 technicians and support staff (44.4%). Of those respondents, 74.0% reported being full-time workers and 80.7% living in a metropolitan area (Table 1).

Table 1. Characteristics of Health Care Workers Who Participated in the Current Population Survey and Annual Social and Economic Supplement and the Prevalence of Reported Unionization, 2009-2021^a.

Characteristics	Participants, unweighted No. (weighted %)		Prevalence of unionization, weighted % (95% CI)b	Unadjustedc		Multivariable adjustedd
	Nonunion (n = 12 511)	Union (n = 1787)		Prevalence OR (95% CI)	P value	Prevalence OR (95% CI)	P value
Age, y
Mean (SD)e	41.3 (13.5)	43.7 (12.6)
15-29	2744 (24.0)	252 (14.9)	8.6 (7.4-9.8)	1 [Reference]		1 [Reference]
30-44	4515 (36.0)	661 (38.6)	14.0 (12.9-15.1)	1.73 (1.45-2.06)	<.001	1.52 (1.26-1.83)	<.001
45-59	3830 (29.3)	645 (34.6)	15.2 (14.0-16.4)	1.90 (1.60-2.27)	<.001	1.67 (1.39-2.01)	<.001
≥60	1422 (10.7)	229 (12.0)	14.5 (12.5-16.4)	1.80 (1.45-2.24)	<.001	1.57 (1.24-1.99)	<.001
Sex
Female	10 191 (80.9)	1462 (80.5)	13.1 (12.4-13.8)	1 [Reference]		1 [Reference]
Male	2320 (19.1)	325 (19.5)	13.4 (11.9-14.9)	1.02 (0.89-1.18)	.76	1.08 (0.92-1.27)	.72
Race and ethnicity
Asian	796 (7.5)	225 (13.8)	21.7 (18.9-24.4)	2.10 (1.76-2.51)	<.001	1.72 (1.40-2.10)	<.001
Hispanic	1063 (10.8)	159 (11.5)	13.9 (11.7-16.0)	1.22 (1.01-1.48)	.04	1.18 (0.95-1.47)	.14
Non-Hispanic Black	1473 (15.2)	246 (17.4)	14.8 (12.9-16.7)	1.32 (1.12-1.55)	<.001	1.60 (1.33-1.94)	<.001
Non-Hispanic White	8946 (64.6)	1120 (56.2)	11.6 (10.9-12.4)	1 [Reference]		1 [Reference]
Otherf	233 (1.9)	37 (1.1)
Regiong
South	4238 (38.6)	260 (14.0)	5.2 (4.5-5.9)	0.20 (0.17-0.24)	<.001	0.16 (0.13-0.19)	<.001
Midwest	3125 (24.5)	366 (20.7)	11.4 (10.1-12.6)	0.47 (0.40-0.55)	<.001	0.46 (0.39-0.55)	<.001
West	2704 (18.3)	602 (31.6)	20.7 (19.0-22.4)	0.95 (0.82-1.10)	.51	0.86 (0.73-1.00)	.06
Northeast	2444 (18.6)	559 (33.7)	21.5 (19.8-23.3)	1 [Reference]		1 [Reference]
Residenceh
Metropolitan area	9977 (85.2)	1560 (91.2)	14.0 (13.2-14.7)	1.80 (1.51-2.15)	<.001	1.56 (1.28-1.90)	<.001
Rural area	2534 (14.8)	227 (8.8)	8.3 (7.0-9.5)	1 [Reference]		1 [Reference]
Occupation
Technicians and support staff	5758 (46.0)	592 (33.1)	9.9 (9.1-10.8)	0.52 (0.46-0.59)	<.001	0.51 (0.44-0.59)	<.001
Nurses	4105 (32.8)	826 (46.2)	17.5 (16.3-18.8)	1 [Reference]		1 [Reference]
Physicians and dentists	970 (7.8)	102 (5.7)	9.8 (7.8-11.8)	0.51 (0.40-0.65)	<.001	0.59 (0.39-0.90)	.01
Advanced practitioners	852 (6.8)	129 (7.2)	14.7 (12.2-17.3)	0.81 (0.65-1.02)	.07	0.84 (0.65-1.08)	.17
Therapists	826 (6.6)	138 (7.7)	15.2 (12.6-17.8)	0.84 (0.68-1.05)	.13	0.74 (0.58-0.94)	.01
Education
≤High school diploma	2060 (16.5)	252 (14.1)	11.6 (10.1-13.1)	1 [Reference]		1 [Reference]
≤Bachelor’s degree	7961 (63.6)	1169 (65.4)	13.4 (12.6-14.3)	1.18 (1.00-1.39)	.04	0.97 (0.80-1.16)	.71
Master’s degree	1120 (9.0)	226 (12.6)	17.9 (15.6-20.3)	1.66 (1.34-2.06)	<.001	1.01 (0.78-1.31)	.92
Doctorate or professional degree	1370 (11.0)	140 (7.8)	9.5 (7.8-11.1)	0.79 (0.62-1.01)	.06	0.55 (0.37-0.81)	.003
Employment statusi
Full-time	9161 (73.2)	1426 (79.8)	14.1 (13.4-14.9)	1.44 (1.25-1.66)	<.001	1.31 (1.13-1.52)	<.001
Part-time	3134 (25.0)	340 (19.0)	10.3 (9.1-11.4)	1 [Reference]		1 [Reference]
Sector of employmentj
Private	11 411 (91.2)	1264 (70.7)	10.6 (9.9-11.1)	0.23 (0.21-0.27)	<.001	0.21 (0.18-0.24)	<.001
Public	1100 (8.8)	523 (29.3)	33.5 (30.8-36.1)	1 [Reference]		1 [Reference]

Abbreviation: OR, odds ratio.

^a Data were derived from the Integrated Public Use Microdata Series-Current Population Survey database maintained by the University of Minnesota. All of the variables listed in this Table had no missing values.

^b Includes those who reported labor union membership or coverage (ie, who reported being covered by a union but not being a member).

^c Models included only the variable under consideration.

^d Model adjusted for all variables listed in the table.

^e Weighted means and standard deviations (SDs).

^f Includes American Indians, multiracial individuals, and other unspecified racial and ethnic groups. This group was excluded from the analysis due to the heterogeneity with this group.

^g Defined by the US Census Bureau Regions classification system.

^h A city population of at least 50 000 and includes surrounding suburbs; rural, everything else.

ⁱ Full-time indicates working least 35 hours a week; 237 who reported “varying hours” per week were excluded due to the uncertainty of hours worked.

^j Private sector of employment includes private for-profit and nonprofit institutions. Public sector of employment includes government and armed forces health care institutions.

Prevalence of Reported Labor Unionization by Health Care Workers

Overall, 1787 health care workers (12.5%) reported being unionized during the study period; among them, 1577 (88.2%) reported being labor union members and 210 (11.8%) reported being covered by a labor union but not being a member. After survey weighting, 13.2% (95% CI, 12.5%-13.8%) of health care workers reported being unionized. There was no apparent trend in unionization over the last 12 years, and the reported prevalence remained unchanged (P = .75; eTable 2 in Supplement 1).

Unionization varied by sociodemographic characteristics. Compared with those aged 15 through 29 years (8.6% [95% CI, 7.4%-9.8%]), older health care workers were significantly more likely to report being unionized: 14.0% among those aged 30 through 44 years (95% CI, 12.9%-15.1%; P < .001); 15.2% among those aged 45 through 59 years (95% CI, 14.0%-16.4%; P < .001); and 14.5% among those aged 60 years or older (95% CI, 12.5%-16.4%; P < .001; Table 1). The reported prevalence of unionization was not significantly different between men and women: 13.4% of men (95% CI, 11.9%-14.9%) vs 13.1% of women (95% CI, 12.4%-13.8%; P = .76).

Unionization differed by race and ethnicity; compared with non-Hispanic White workers, in order of prevalence, 21.7% Asian workers were most likely to report being unionized (95% CI, 18.9%-24.4%; P < .001), 14.8% Non-Hispanic Black workers (95% CI, 12.9%-16.7%; P < .001), and 13.9% Hispanic workers (95% CI, 11.7%-16.0%; P = .04). Unionization also differed by state (Figure 1).

Figure 1. The Prevalence of Labor Unionization in US Health Care Workers by State, 2009-2021.

All values are survey weighted. Unionized health care workers were defined as those who reported labor union membership or coverage (ie, who reported being covered by a union but not being a member).

Compared with the 21.5% of health care workers living in the Northeast (95% CI, 19.8%-23.3%) who reported unionization, 20.7% in the West (95% CI, 19.0%-22.4%) reported unionization, which was not statistically significantly different (P = .51). But 11.4% of health care workers in the Midwest reported being unionized (95% CI, 10.1%-12.6%; P < .001) and 5.2% in the South (95% CI, 4.5-5.9; P < .001), both of which showed statistically significant differences from the Northeast region. Similarly, health care workers living in metropolitan areas (14.0% [95% CI, 13.2%-14.7%]) were also more likely to report being unionized than those in rural areas (8.3% [95% CI, 7.0%-9.5%]; P < .001).

Unionization also differed by occupation. Nurses had the highest prevalence of reporting being unionized (17.5% [95% CI, 16.3%-18.8%]). Compared with nurses, physicians and dentists (9.8% [95% CI, 7.8%-11.8%]; P < .001) and technicians and support staff (9.9% [95% CI, 9.1%-10.8%]; P < .001) had statistically significantly lower rates of reporting being unionized; whereas advanced practitioners (14.7% [95% CI, 12.2%-17.3%]; P = .07) and therapists (15.2% [95% CI, 12.6%-17.8%]; P = .13) had no significantly statistical difference from nurses in reporting being unionized.

Those working full-time (14.1% [95% CI, 13.4%-14.9%]) were more likely to report being unionized than those working part-time (10.3% [95% CI, 9.1%-11.4%]; P < .001). Those working in the private sector (10.6% [95% CI, 9.9%-11.1%]) were less likely to report being unionized than those working in the public sector (33.5% [95% CI, 30.8%-36.1%]; P < .001). In multivariable analyses adjusting for other factors, most of these sociodemographic factors remained associated with reported unionization (Table 1). Of the statistically significant variables associated with reported labor unionization, the strongest were race and ethnicity, region, age, occupation, metropolitan residence, and sector of employment.

Reported Labor Unionization and Pay, Benefits, and Equity

Being unionized was associated with better pay and better benefits. Unionized health care workers had significantly higher reported mean weekly earnings than nonunion workers ($1165 vs $1042; mean difference, $123 [95% CI, $88-$157]; P < .001) (Figure 2). Sensitivity analysis of individuals living within metropolitan areas found similar association ($1169 vs $1019; mean difference, $150 [95% CI, $63-$238]; P < .001; eTable 3 in Supplement 1). Unionized health care workers were more likely to report having a pension or other retirement benefits at work than nonunionized workers (57.9% vs 43.4%; risk ratio [RR], 1.33 [95% CI, 1.26-1.41]; P < .001). Unionized workers were also more likely to report having an employer paid-for, full premium–covered health insurance plan (22.2% vs 16.5%; RR, 1.35 [95% CI, 1.17-1.53]; P < .001). Unionized workers reported significantly higher annual employer contribution to their health insurance plans ($4561 vs $3455; mean difference, $1106 [95% CI, $843-$1369]; P < .001). However, compared with nonunionized workers, those who were unionized reported more weekly work hours (37.4 vs 36.3 hours; mean differences, 1.11 [95% CI, 0.46-1.75]; P < .001; Table 2).

Figure 2. Associations of Unionization and Mean Weekly Earnings at Current Job Among Health Care Workers .

Mean weekly earnings reports how much the survey respondent earned per week at their current job before deductions. Interviewers asked, “How much do you usually earn per week at this job before deductions?” For workers paid by the hour, they were also asked for their hourly wage rate and the number of hours they usually worked at their job. The higher of the values derived from these 2 sources is reported. All values are survey-weighted to be nationally representative. Multivariable linear regression models were used to compare reported mean weekly earnings, adjusting for sociodemographic factors associated with earnings (age, sex, race and ethnicity, education, occupation, public or private sector of employment, US state of residence, metropolitan or rural area of residence, and central city status). All dollars were standardized to 2020 US dollars using the consumer price index. P values in the figure were from tests for prespecified interactions between union membership and mean weekly earnings by sex and racial and ethnic groups (defined as Asian, Hispanic, or non-Hispanic Black vs non-Hispanic White). t Tests were used to compare mean weekly earnings between groups. Square data points represent the adjusted mean difference in weekly earnings between union and nonunion workers within the group; whiskers, 95% CIs of the mean difference.

Table 2. Association Between Labor Unionization and Pay, Work Hours, and Noncash Benefits, 2009-2021^a.

Outcomes	Overall		Occupational group
	Estimate (95% CI)	P value	Physicians and dentists		Advanced practitioners		Nurses		Therapists			Technicians and support staff
			Estimate (95% CI)	P value	Estimate (95% CI)	P value	Estimate (95% CI)	P value	Estimate (95% CI)	P value	Estimate 95% CI)		P value
	Mean difference		Mean difference		Mean difference		Mean difference		Mean difference		Mean difference
Mean weekly earnings, $b	123 (88 to 157)	<.001	48 (−157 to 253)	.65	31 (−115 to 178)	.68	165 (113 to 216)	<.001	228 (71 to 384)	<.001	67 (21 to 113)		.004
Employer contribution to health insurance, $/yb,c	1106 (843 to 1369)	<.001	983 (−214 to 2179)	.11	1617 (554 to 2679)	.003	1067 (647 to 1487)	<.001	1274 (112 to 2437)	.03	905 (517 to 1293)		<.001
Mean work hours per wk	1.11 (0.46 to 1.75)	<.001	0.63 (−3.60 to 4.85)	.77	3.07 (0.33 to 5.80)	.03	0.21 (−0.66 to 1.07)	.64	2.49 (0.10 to 4.88)	.04	1.81 (0.80 to 2.83)		<.001
	Risk ratio		Risk ratio		Risk ratio		Risk ratio		Risk ratio		Risk ratio
Pension or other retirement benefits	1.33 (1.26 to 1.41)	<.001	1.23 (1.00 to 1.46)	.06	1.57 (1.32 to 1.82)	<.001	1.28 (1.19 to 1.37)	<.001	1.13 (0.87 to 1.39)	.33	1.41 (1.25 to 1.57)		<.001
Health insurance pland	1.35 (1.17 to 1.53)	<.001	1.45 (0.93 to 1.97)	.005	1.59 (0.96 to 2.22)	.03	1.38 (1.07 to 1.69)	.005	0.97 (0.34 to 1.59)	.92	1.25 (0.94 to 1.56)		.09

^a All results listed used the nonunion group as the reference group. All models adjusted for age, sex, race and ethnicity, education, occupation, public or private sector employment, US state of residence, metropolitan or rural area of residence, and central city status.

^b US dollars were inflation adjusted to 2020 US dollars using the consumer price index. Twenty-eight individuals (25 nonunionized, 3 unionized) did not report their mean weekly earnings and were removed from this analysis.

^c US dollars were inflation adjusted to 2020 US dollars using the consumer price index. Employer contribution to health insurance premium was reported from 2009 through 2018.

^d Employer paid-for full premium–covered health insurance plan.

Comparisons of the association between unionization and pay in subgroups are shown in Figure 2. Although being unionized was associated with significantly higher reported mean weekly earnings for female workers ($1148 vs $1008; mean difference, $140 [95% CI, $103 to $177]; P < .001) but was not for male workers ($1238 vs $1187; mean difference, $51 [95% CI, −$33 to $135]; P = .24), the test for interaction did not meet statistical significance (P = .052). Male workers reported significantly more mean weekly earnings than female workers, regardless of unionization status. The mean difference between sexes among nonunion workers was $179 (95% CI, $144 to $215; P < .001) but the mean difference between sexes among union workers was $90 (95% CI, $6 to $174, P = .03). The association between unionization and pay differed significantly by racial and ethnic minority status (test for interaction, P = .02). Stratified analysis showed that being unionized was associated with significantly higher reported mean weekly earnings for non-Hispanic White ($1157 vs $1066; mean difference, $91 [95% CI, $49 to $132]; P < .001) and for racial and ethnic minority workers ($1170 vs $1001; mean difference, $169 [95% CI, $112 to $226]; P < .001). When compared across racial and ethnic groups, non-Hispanic White workers reported significantly more mean weekly earnings than members of racial and ethnic minorities among nonunionized workers (mean difference, $65 [95% CI, $40 to $91]; P < .001) but there was no significant difference between the 2 groups among unionized workers (mean difference, −$13 [95% CI, −$78 to $52]; P = .70).

The benefits associated with unionization differed by occupation (Table 2). Particularly for physicians and dentists, there was no significant difference between unionized and nonunionized workers in reported mean weekly earnings ($2116 vs $2068; mean difference, $48 [95% CI, −$157 to $253]; P = .65), employer’s contribution to health insurance ($7143 vs $6160; mean difference, $983 [95% CI, −$214 to $2179; P = .11), mean weekly work hours (46.36 vs 45.73; mean difference, 0.63 [95% CI, −3.60 to 4.85]; P = .77), and likelihood of having a pension or other retirement benefits at work (67.3% vs 54.8%; RR, 1.23 [95% CI, 1.00-1.46]; P = .06).

Similar findings were observed in sensitivity analyses excluding individuals who reported being covered by a labor union but not being a member (eTable 4 in Supplement 1). Unionized workers were statistically significantly more likely to have full or partially covered health insurance plans, with a smaller increase in magnitude when compared with the main analysis comparing full vs nonfull (partial or none) premium covered health plans (eTable 5 in Supplement 1).

Discussion

In this cross-sectional study of 14 298 US health care workers, 13.2% of workers across health care professions reported being labor unionized, with most being labor union members. Unionization was associated with higher wages and better benefits for health care workers without much change in working hours.

Previous studies have demonstrated an association between collective bargaining and wages and benefits for workers across different industries and skill levels.^16,17 This literature has also documented how rates of unionization are also associated with reductions in both sector-specific and economy-wide wage inequality.¹⁸ In contrast, empirical evidence on unionization in health care has been limited to date. The Bureau of Labor and Statistics only tracks union activities across broad categories (eg, occupation and industry) and lacks sufficient granularity to characterize individual groups of health care workers (eg, nurses, physicians, and dentists).¹⁹

This study addresses gaps in the literature by presenting evidence on labor unionization across the health care profession from 2009 through 2021, with the findings consistent with those documented by the Bureau of Labor and Statistics and in existing labor scholarship. First, the overall rate of unionization among health care workers in the study sample was consistent with those reported in the Bureau of Labor and Statistics data on workers categorized under “healthcare practitioners and technical occupations” (13.2% vs 11.7%).¹⁹ Second, the geographic differences in unionization prevalence in this study were aligned with those reported nationally in other sectors of the economy and may reflect the role of broader policy and political factors (eg, existence of right-to-work laws) in collective bargaining.²⁰ Third, the study sample reflected the demographics of the health care workforce documented in previous literature, namely, that the sample was predominantly female and individuals working as nurses and technicians and support staff.²¹ Fourth, the difference between the reported mean weekly wages of unionized and nonunionized workers in this study sample were directionally similar to the Bureau of Labor and Statistics data on workers in both health care occupations and workers in all industries.^22,23 Fifth, the higher wages and rates of access to benefits such as pensions and fully covered health insurance among unionized health care workers in the study sample were also aligned with the trends documented for unionized workers in other industries.^24,25 Sixth, the associated benefits of unionization differed by occupation, particularly for physicians and dentists, which is consistent with previous analyses showing less benefit to those in higher income brackets.²⁶ Additionally, physicians and dentists can be reimbursed under different payment models (eg, salary, fee-for-service, or value-based payment), which differs from other health care workers who typically earn an hourly wage from their employer.

To our knowledge, this is the first systematic examination of the relationship between unionization and working conditions across all segments of the health care workforce. This study has several strengths. First, the CPS data set captures rich data on demographic factors, occupation, pay, and noncash benefits of a nationally representative sample. The quality of this data allowed for characterization of each respondent with high resolution, such that many factors associated with the outcomes of interest in this study could be adjusted for. Second, the large sample size, and its national representation, allowed for inferences to be made about labor unions across the US health care system. Third, consistent annual survey data allowed for investigations of trends over time, to detect any changes in labor unionization until 2021.

Limitations

This study has several limitations. First, responses to the CPS may be susceptible to reporting bias. Second, CPS does not differentiate whether employed health care workers may also be enrolled in graduate training programs—a unique consideration for health care compared with other industries given the role of government subsidies and wage deflation for intern and resident salaries.²⁷ Third, CPS does not offer insight into workers’ experiences with regards to job satisfaction, stress, or mistreatment. Such factors are a key consideration given that unionization has been touted as a strategy for mitigating burnout, although evidence to date has been mixed.^7,28 Fourth, the study findings do not allow inferring that associations with union membership are caused by union membership. They do not distinguish whether associations are related to membership or to the characteristics of the organization in which the workers were employed. Fifth, CPS does not capture data on the risks of labor unionization. Employers have been reported to violate federal law in nearly half of all union election campaigns to thwart unionization efforts.²⁹ The harmful downstream actions by an employer on their health care employees are unclear.

Conclusions

From 2009 through 2021, labor unionization among US health care workers remained low. Reported union membership or coverage was significantly associated with higher weekly earnings and better noncash benefits but greater number of weekly work hours.

Supplement 1.

eTable 1. Health Care Occupation Codes for Occupation Group Categorization

eTable 2. Prevalence of Labor Unionization Among Health Care Workers by Year, United States, 2009-2021

eTable 3. Association between Labor Unionization and Mean Weekly Earnings among 2,000 Health Care Workers Living in Metropolitan Areas, 2009-2021

eTable 4. Association between Labor Union Membership and Pay, Work Hours, and Noncash Benefits, Overall and by Occupational Group, 2009-2021^a

eTable 5. Association between Labor Unionization and Having Employer-sponsored Health Insurance Plans

Supplement 2.

Data Sharing Statement