Health Insurance Coverage by Occupation Among Adults Aged 18–64 Years — 17 States, 2013–2014

Lack of health insurance has been associated with poorer health status and with difficulties accessing preventive health services and obtaining medical care, especially for chronic diseases (1–3). Among workers, the prevalence of chronic conditions, risk behaviors, and having health insurance has been shown to vary by occupation (4,5). CDC used data from the 2013 and 2014 Behavioral Risk Factor Surveillance System (BRFSS) to estimate the prevalence of having no health care coverage (e.g., health insurance, prepaid plans such as health maintenance organizations, government plans such as Medicare, or Indian Health Service) by occupation. Among all workers aged 18–64 years, the prevalence of being uninsured declined significantly (21%) from 16.0% in 2013 to 12.7% in 2014. In both years there were large differences in the prevalence of being uninsured among occupational groups, ranging from 3.6% among the architecture and engineering occupations to 37.9% among the farming, fishing, and forestry occupations in 2013 and 2.7% among community and social services; and education, training, and library occupations to 37.0% among building and grounds cleaning and maintenance occupations in 2014 (p<0.001). In 2014, more than 25% of workers in four occupational groups reported having no health insurance (construction and extraction [29.1%]; farming, fishing, and forestry [34.6%]; food preparation and serving related [35.5%]; and building and grounds cleaning and maintenance [37.0%]). Identifying factors affecting differences in coverage by occupation might help to address health disparities among occupational groups.

BRFSS is an annual, state-based, random-digit–dialed landline and cell phone survey of noninstitutionalized adults aged ≥18 years residing in the United States.* Industry and occupation was first available as an optional module in BRFSS in 2013. In both 2013 and 2014, 17 states^† asked all survey participants about their health care coverage^§ and asked participants who were currently or recently employed at the time of their interview about their industry and occupation.^¶ Participants’ responses were coded to the 2002 version of U.S. Census Bureau occupation numeric codes. Census occupation codes were then grouped for analysis into major groups using the 2000 Standard Occupational Classification System. During 2014, 12 of the 17 states elected to expand Medicaid eligibility to persons with an income ≤138% of the federal poverty level,** and five did not (6).

The subpopulation of interest included respondents aged 18–64 years and currently employed for wages or self-employed in the 17 states. It excluded those on active military duty or whose occupation was missing or could not be coded. Respondents aged ≥65 years were excluded because they were presumed to be eligible for Medicare (7).

Data were weighted and analyzed to account for the complex BRFSS sampling design. The prevalence of being uninsured was estimated by occupational group and sociodemographic characteristics, stratified by year. Unadjusted prevalences by occupational group were calculated to present the magnitude of noncoverage for each occupational group. To control for effects of the potential confounders age, sex, race/ethnicity, language in which the survey was conducted, education, annual household income, marital status, employment status (currently employed for wages or self-employed), county urbanization, and state Medicaid expansion in 2014, and to provide estimates specifically reflecting the association between occupation and being uninsured, adjusted prevalences were estimated using logistic regression. The initial model included occupational group, year, the interaction between occupational group and year, confounders, and the two-way interaction term between each confounder and year. County urbanization and each confounder interaction term except for age by year and income by year were dropped from the final model because they were not statistically significant.

Among the 17 states, the survey response rates ranged from 31.1% to 59.2% in 2013^†† and 33.0% to 57.6% in 2014.^§§ In 2013 and 2014, the subpopulation of interest comprised 138,407 workers. Among these, 18,140 (13%) were excluded because occupation was missing or could not be coded, leaving 59,718 respondents in 2013 and 60,549 in 2014.

The overall prevalence of being uninsured among workers in 2013 (16.0%) declined 21% (p<0.001) (3.3 percentage points [p<0.001]) to 12.7% in 2014 (Table 1). The prevalence of being uninsured declined in all demographic groups in 2014; both the percentage point difference and the percentage change were statistically significant for all groups except persons aged 25–34 years, persons who took the survey in Spanish, those with household incomes ≥$50,000, and those who resided in urban and rural counties. The decline was statistically significant among workers who lived in the most populous counties, metropolitan (Table 1). The prevalence of being uninsured exceeded 20% in 2014 (well above the average of 12.7%) among workers who took the survey in Spanish, who had less than a high school education, who had an annual household income <$35,000, were of Hispanic ethnicity, self-employed, and never married.

In both 2013 and 2014, a lower percentage of workers were uninsured in the 12 states that expanded Medicaid eligibility than were in the five states that did not, and the prevalence of being uninsured declined more (23%) in states that expanded Medicaid than in those that did not (9%; p = 0.013), although the percentage point difference between the two groups of states was not statistically significant (Table 1).

In both 2013 and 2014, there were statistically significant differences among occupation groups in the unadjusted prevalence of being uninsured (p<0.001). In 2013, the unadjusted prevalence of being uninsured ranged from 3.6% among the architecture and engineering occupations to 37.9% among the farming, fishing, and forestry occupations. In 2014, the unadjusted prevalence of being uninsured ranged from a high of 37.0% (building and grounds cleaning and maintenance) to 2.7% (community and social services; and education, training, and library) (Table 2). More than 25% of the workers in four occupations (construction and extraction [29.1%]; farming, fishing, and forestry [34.6%]; food preparation and serving related [35.5%]; and building and grounds cleaning and maintenance [37.0%]) reported not having health insurance in 2014.

There were also statistically significant differences in adjusted prevalence of being uninsured by occupation (p<0.001) in both 2013 and 2014. The 2014 adjusted prevalence of being uninsured ranged from 19.4% in the farming, fishing, and forestry occupations to 5.4% in the education, training, and library occupations. Half the occupational groups experienced significant decreases from 2013 to 2014 in the adjusted prevalence of being uninsured (Table 3). The same four occupational groups with the highest unadjusted prevalences of being uninsured in 2014 also had the highest prevalences in 2014 after adjustment for potential confounders (Table 3).

Acknowledgments

Jan Birdsey, Susan Burton, P. Timothy Bushnell, Jeff Purdin, Pam Schumacher, Marie Haring Sweeney, CDC; 17 state BRFSS coordinators.

Corresponding author: Winifred L. Boal, wboal@cdc.gov, 513-841-4266.

¹Division of Surveillance, Hazard Evaluations and Field Studies, National Institute for Occupational Safety and Health, CDC, ²Spokane Mining Research Division, National Institute for Occupational Safety and Health, CDC.

References

1. Ayanian JZ, Weissman JS, Schneider EC, Ginsburg JA, Zaslavsky AM. Unmet health needs of uninsured adults in the United States. JAMA 2000;284:2061–9. CrossRefexternal icon PubMedexternal icon
2. Okoro CA, Zhao G, Dhingra SS, Xu F. Lack of health insurance among adults aged 18 to 64 years: findings from the 2013 Behavioral Risk Factor Surveillance System. Prev Chronic Dis 2015;12:E231. PubMedexternal icon
3. Okoro CA, Zhao G, Fox JB, Eke PI, Greenlund KJ, Town M. Surveillance for health care access and health services use, adults aged 18–64 years—Behavioral Risk Factor Surveillance System, United States, 2014. MMWR Surveill Summ 2017;66(No. SS–7). CrossRefexternal icon PubMedexternal icon
4. Towle M, Tolliver R, Bui AG, Warner A, Van Dyke M. Adding industry and occupation questions to the Behavioral Risk Factor Surveillance System: new opportunities in public health surveillance. Public Health Rep 2015;130:153–60. CrossRefexternal icon PubMedexternal icon
5. Massachusetts Department of Public Health. Putting data to work, 23 health indicators by occupation and industry: findings from the Massachusetts Behavioral Risk Factor Surveillance System, 2012–2013. Boston, MA: Massachusetts Department of Public Health; 2016. https://www.mass.gov/eohhs/docs/dph/occupational-health/full-report-health-indicators-15.pdfpdf iconexternal icon
6. Kaiser Family Foundation. Status of state action on the Medicaid expansion decision. Menlo Park, CA: Kaiser Family Foundation; 2018. https://www.kff.org/health-reform/state-indicator/state-activity-around-expanding-medicaid-under-the-affordable-care-act/?currentTimeframe=0&sortModel=%7B%22colId%22:%22Location%22,%22sort%22:%22asc%22%7Dexternal icon
7. Centers for Medicare & Medicaid Services. Medicare program—general information. Woodlawn, MD: Centers for Medicare & Medicaid Services; 2014. https://www.cms.gov/Medicare/Medicare-General-Information/MedicareGenInfo/index.htmlexternal icon
8. National Center for Health Statistics. Health insurance coverage: early release of estimates from the National Health Interview Survey, 2017. Hyattsville, MD: US Department of Health and Human Services, CDC, National Center for Health Statistics; 2018. https://www.cdc.gov/nchs/data/nhis/earlyrelease/insur201805.pdfpdf icon
9. National Institute for Occupational Safety and Health. Worker health chartbook, 2004. Washington, DC: US Department of Health and Human Services, CDC, National Institute for Occupational Safety and Health; 2014. https://www.cdc.gov/niosh/docs/2004-146/pdfs/2004-146.pdfpdf icon
10. Nelson DI, Concha-Barrientos M, Driscoll T, et al. The global burden of selected occupational diseases and injury risks: methodology and summary. Am J Ind Med 2005;48:400–18. CrossRefexternal icon PubMedexternal icon

Abbreviation: CI = confidence interval.
* Unadjusted, weighted estimates.
^† Percent change = [(prevalence in 2014 – prevalence in 2013)/prevalence in 2013] x 100.
^§ p<0.05.
^¶ Estimates have a relative standard error >50% and are not shown as they do not meet standards of reliability/precision.
** County of residence was classified as metropolitan (codes 1–3), urban (4–7), or rural (8–9), based on the U.S. Department of Agriculture’s 2013 Rural-Urban Continuum Codes. https://www.ers.usda.gov/data-products/rural-urban-continuum-codes.aspxexternal icon.
^†† Except as noted, expansion began January 1, 2014: Illinois, Maryland, Massachusetts, Michigan (April 1, 2014), Minnesota, New Hampshire (August 15, 2014), New Jersey, New Mexico, New York, North Dakota, Oregon, and Washington (n = 83,755; 69.6% of respondents).
^§§ Louisiana, Mississippi, Montana, Nebraska, and Utah (n = 36,512; 30.4% of respondents).

Abbreviation: CI = confidence interval.
* Unadjusted, weighted estimates.
^† From the 2000 Standard Occupational Classification System. https://www.bls.gov/soc/external icon.
^§ Percent change = [(prevalence in 2014 – prevalence in 2013) / prevalence in 2013] x 100.
^¶ p<0.05.
** Estimates have a relative standard error >30% and ≤50% and should be used with caution as they do not meet standards of reliability/precision.

Abbreviation: CI = confidence interval.
* Weighted estimates adjusted by age group, sex, race/ethnicity, language in which the survey was administered (English, Spanish, other), education, annual household income, marital status, employment status (currently employed for wages or self-employed), and state Medicaid expansion.
^† From the 2000 Standard Occupational Classification System. https://www.bls.gov/soc/external icon.
^§ Percent change = [(prevalence in 2014 – prevalence in 2013) / prevalence in 2013] x 100.
^¶ p<0.05.
** Estimates have a relative standard error >30% and ≤50% and should be used with caution as they do not meet standards of reliability/precision.