Declining Health Insurance Coverage for Health Care Personnel and ...

 RESEARCH AND PRACTICE 

No Care for the Caregivers: Declining Health Insurance Coverage for Health Care Personnel and Their Children, 1988–1998 | Brady G. S. Case, AB, David U. Himmelstein, MD, and Steffie Woolhandler, MD, MPH

Two years ago, more Americans were uninsured than at any other time in the previous 2 decades, yet the unemployment rate in 1999 was the nation’s lowest since 1969.1,2 Because most Americans secure health insurance through an employer, accounting for this paradox has been a central task of health policy research. Explanations have focused on rising costs of coverage,3–5 as well as on growing service-sector and part-time employment,6,7 declining unionization,8 segregation of high- and low-wage workers into different firms,9 shrinking job tenure,10 and error in the measurement of coverage.11,12 The role of institutions of care—hospitals, medical offices, nursing homes, and home care agencies—has received less attention despite the health care industry’s dramatic growth and long-standing importance as an employer in inner-city, rural, and minority communities.13–15 We examined trends in the health insurance coverage of personnel working in health establishments and their children over the past decade.

METHODS We analyzed data from the annual March supplements of the Current Population Survey (CPS) for 1989 through 1999. The CPS is an annual Census Bureau survey of a nationally representative sample of approximately 150 000 noninstitutionalized US residents in about 50 000 households. Respondents are asked about their insurance status during the previous calendar year, as well as occupational and demographic characteristics. One quarter of the respondents are asked about union membership. We used Standard Industrial Classification (SIC) codes to identify principal place of employment in the week before the survey. Persons employed in the offices of physicians or

Objectives. This study examined trends in health insurance coverage for health care workers and their children between 1988 and 1998. Methods. We analyzed data from the annual March supplements of the Current Population Survey (CPS), a Census Bureau survey that collects information about health insurance from a nationally representative sample of noninstitutionalized US residents. Results. Of the health care personnel younger than 65 years, 1.36 million (90% confidence interval [CI] = 1.28 million, 1.45 million) were uninsured in 1998, up 83.4% from 1988; the proportion uninsured rose from 8.4% (90% CI = 7.8%, 9.1%) to 12.2% (90% CI = 11.5%, 12.9%). Declining coverage rates in the growing private-sector health care workforce—and declining health employment in the public sector, which provided health insurance benefits to more of its workers—accounted for the increases. Households with a health care worker included 1.12 million (90% CI = 1.05 million, 1.20 million) uninsured children, accounting for 10.1% (90% CI=9.5%, 10.8%) of all uninsured children in the United States. Conclusions. Health care personnel are losing health insurance coverage more rapidly than are other workers. Increasingly, the health care sector is consigning its own workers and their children to the ranks of the uninsured. (Am J Public Health. 2002;92:404–408) other health practitioners (SIC codes 812, 820–830), in hospitals (SIC code 831), in nursing and personal care facilities (SIC code 832), or in other health services (SIC code 840) were defined as health personnel. Non– health personnel were persons employed in other settings. People were defined as publicsector employees if they reported employment by federal, state, or local government; they were defined as private-sector employees if they reported work for a private employer. Persons older than 65 years were excluded from the analysis because virtually all are covered by Medicare. Children in households of health care personnel were defined as persons younger than 18 years who shared living quarters (rooms, an apartment, or a house) with at least 1 health care worker. We classified occupations according to the Census Bureau’s Occupational Classification Codes for Detailed Occupational Categories. Physicians, registered nurses, and licensed practical nurses were each identified by single codes (codes 84, 95, and 207, respectively). Other occupations were grouped as

404 | Research and Practice | Peer Reviewed | Case et al.

follows: managers and management-related occupations (codes 3–37); aides (codes 445–447); food service, cleaning, building service, and laundry workers (codes 433– 444, 448–455, 747, and 748); clerical and administrative support workers (codes 205, 276, 303, and 390); and other workers (all other codes). Neither SIC nor Occupational Classification Codes identify public health workers. People were considered insured if they reported any health insurance, public or private. Population estimates were derived by using weights supplied by the Census Bureau. These weights adjust for the complex sample design and the failure to obtain interviews from some households. Weights for survey data for the March 1994 annual CPS and thereafter were obtained from an updated sample framework based on the 1990 decennial census. Weights for data prior to the 1994 CPS were based on the 1980 census. The change had little effect on estimates of the percentage uninsured but did alter estimates of the number uninsured.16 For exam-

American Journal of Public Health | March 2002, Vol 92, No. 3


ple, use of the 1990-based weights for 1992 data resulted in an estimate of the total number of uninsured Americans that was 1.2 million higher than the 38.6 million estimated with the 1980-based weight.17 Percentage estimates (e.g., 15.0% of the Americans uninsured in 1992) were closely similar regardless of weight used. In addition, in 1993 the Census Bureau began computer-assisted interviewing, and in 1994 health insurance questions were modified to minimize underreporting of coverage, particularly private health insurance coverage.16,18,19 The latter change raised CPS estimates of health insurance coverage and decreased estimates of the number and proportion uninsured for years after 1994.19 In keeping with Census Bureau conventions for demographic data, we calculated 90% confidence intervals (CIs) with parameters supplied by the Census Bureau to correct for expected correlations among individuals within a sampled household according to the following formulas: (1) (2) (3) (4)

Sx = (ax 2 + bx)0.5 Sy,p = [bp(100 − p)/y]0.5 Sl–m = (Sl2 + Sm2)0.5 Sl/m = l/m[(Sl /l)2 + (Sm /m)2 − 2r(Sl Sm /lm)]0.5

where Sx, Sy,p, Sl-m, and Sl/m are the standard errors of the estimated number x, percentage y, difference l–m, and ratio l/m, respectively; p is the total number of persons (the base of the percentage); l and m are estimated numbers or percentages; r is the correlation between l and m; and a and b are the Census Bureau’s adjustment parameters for the particular characteristics of the population under study (e.g., age, employment, educational status). The correlation r is assumed to be zero in most estimates of the standard error of a ratio; however, the assumed value of r is 0.7 if the denominator m is a number estimate of families or households and the numerator l is a number estimate of people in those families or households with specified characteristics. Standard errors for statewide estimates are obtained with a and b parameters multiplied by state-specific f 2 factors; for regional division standard errors, a population-weighted average of state f 2 factors is used.16

RESULTS Between 1988 and 1998, the number of uninsured health care personnel increased 83.4%, from 743 000 (90% CI = 681 000, 806 000) to 1.36 million (90% CI = 1.28 million, 1.45 million), and the proportion uninsured rose from 8.4% (90% CI = 7.8%, 9.1%) to 12.2% (90% CI = 11.5%, 12.9%) (Table 1). The number of uninsured workers in other industries grew more modestly over the period, up 37.8%, whereas the proportion who were uninsured increased from 14.7% (90% CI = 14.5%, 15.0%) to 17.9% (90% CI = 17.7%, 18.2%). Coverage of health care workers varied by place of employment. Twenty percent (90% CI = 17.9%, 22.1%) of nursing home personnel were uninsured in 1998, compared with 8.2% (90% CI = 7.3%, 9.0%) of hospital workers, 8.7% (90% CI = 7.0%, 10.4%) of those employed in medical offices, and 15.9% (90% CI = 14.3%, 17.4%) of workers at other health care establishments. The private sector accounted for all of the growth in the number of uninsured health personnel; the number of uninsured personnel in private institutions rose by 614000 (90% CI=517000, 712000)—a 101.4% increase between 1988 and 1998. The relative growth of private health care employment—38.4% compared with a 19.6% decline in public employment—contributed to this large increase. However, a rise in the proportion of private-sector health workers uninsured—from 8.6% (90% CI=7.8%, 9.3%) to 12.5% (90% CI=11.7%, 13.3%)—accounted for most of the increase in the number of uninsured private-sector health personnel. The proportion of all public-sector health personnel who were uninsured remained stable, at 7.6% (90% CI=5.9%, 9.2%) in 1988 and 7.8% (90% CI=5.9%, 9.6%) in 1998. Workers in public hospitals and nursing homes were much less likely to be uninsured than were workers in comparable private institutions in 1998. Uninsurance remained steady at 6.5% (90% CI = 4.6%, 8.3%) in 1988 and 6.3% (90% CI = 4.1%, 8.4%) in 1998 among public hospital workers, but it rose from 5.4% (90% CI = 4.5%, 6.3%) to 8.5% (90% CI = 7.5%, 9.4%) in private hospitals. Of the public nursing home workers, 7.5% (90% CI = 2.3%, 12.6%) were uninsured in 1998, down from

March 2002, Vol 92, No. 3 | American Journal of Public Health

11.9% (90% CI = 5.7%, 18.1%) 10 years earlier, whereas in private nursing homes, the proportion uninsured grew to 20.9% (90% CI = 18.7%, 23.2%) from 18.8% (90% CI = 16.3%, 21.3%). Total employment at public hospitals also declined substantially—by 30.4%—but increased at private hospitals. Among occupational groups, aides had the highest uninsurance rate in 1998—23.8% (90% CI=21.7%, 25.9%)—and accounted for 37.0% of the uninsured health personnel. Food service, cleaning, building service, and laundry workers also had relatively low rates of health coverage—19.7% (90% CI=11.7%, 27.7%) were uninsured in 1998. The proportion of uninsured licensed practical nurses rose steeply between 1988 and 1998, from 7.3% (90% CI=4.2%, 10.3%) to 14.5% (90% CI=10.3%, 18.7%), and uninsurance among physicians increased from 3.3% (90% CI=1.5%, 5.2%) to 5.4% (90% CI=3.3%, 7.6%). Personnel in management-related occupations saw the sharpest fall in coverage, from 2.8% (90% CI=1.4%, 4.2%) uninsured in 1988 to 7.3% (90% CI=5.5%, 9.1%) in 1998, and experienced a 282% (90% CI=104%, 460%) increase in the number uninsured. Unionized workers had a higher rate of insurance than did nonunionized health workers in both years and experienced a milder decline in coverage. In 1988, 6.3% (90% CI = 4.5%, 8.1%) of the union members employed in health care settings were uninsured, compared with 8.8% (90% CI = 8.1%, 9.6%) of the nonunion workers. By 1998, the proportion of union members uninsured had risen to 7.8% (90% CI = 6.0%, 9.7%), whereas among nonunion health workers, 13.3% (90% CI = 12.6%, 14.1%) were uninsured. Unionization rates among health personnel decreased from 10.9% (90% CI = 9.8%, 11.3%) in 1988 to 9.7% (90% CI = 9.0%, 10.3%) in 1998. In 1998, 20.5% (90% CI = 17.8%, 23.1%) of Black health care workers were uninsured, compared with 10.4% (90% CI = 9.7%, 11.2%) of White health personnel (Table 2). Black women, who constituted 12.9% of the health care workforce in 1998, accounted for 22.5% of the uninsured personnel. Hispanic workers—who may be of any race—were twice as likely as non-Hispanics to be uninsured: 24.7% (90% CI = 20.6%, 28.7%) vs 11.2% (90% CI = 10.5%, 11.9%).

Case et al. | Peer Reviewed | Research and Practice | 405


TABLE 1—Percentage and Number of Health Personnel and Workers in Other Industries Without Health Insurance, by Occupational Characteristics, 1988 and 1998 % Uninsured (90% CI)

Health personnel Private sector Public sector Full-time Part-time Union member Not union member Place of employment Practitioner’s office Hospital Private hospital Public hospital Nursing home Private nursing home Public nursing home Other health care establishment Occupation Doctor of medicine Registered nurse Licensed practical nurse Management Aides Food, cleaning, building, and laundry services Clerical and administrative Other occupation Workers in other industries Private sector Public sector

No. Uninsured, 1000s of Workers (90% CI)

1988

1998

1988

8.4 (7.8, 9.1) 8.6 (7.8, 9.3) 7.6 (5.9, 9.2) 8.0 (7.3, 8.8) 10.1 (8.5, 11.7) 6.3 (4.5, 8.1) 8.8 (8.1, 9.6)

12.2 (11.5, 12.9) 12.5 (11.7, 13.3) 7.8 (5.9, 9.6) 11.5 (10.7, 12.2) 15.2 (13.4, 16.9) 7.8 (6.0, 9.7) 13.3 (12.6, 14.1)

743 (681, 806) 606 (549, 662) 103 (80, 127) 559 (505, 613) 184 (153, 216) 59 (41, 76) 669 (609, 728)

1363 (1279, 1447) 1220 (1140, 1299) 85 (64, 106) 1028 (955, 1101) 335 (293, 377) 85 (64, 106) 1341 (1258, 1424)

6.3 (4.5, 8.0) 5.6 (4.8, 6.4) 5.4 (4.5, 6.3) 6.5 (4.6, 8.3) 18.1 (15.7, 20.4) 18.8 (16.3, 21.3) 11.9 (5.7, 18.1) 9.1 (7.6,10.6)

8.7 (7.0, 10.4) 8.2 (7.3, 9.0) 8.5 (7.5, 9.4) 6.3 (4.1, 8.4) 20.0 (17.9, 22.1) 20.9 (18.7, 23.2) 7.5 (2.3, 12.6) 15.9 (14.3,17.4)

63 (44, 81) 251 (215, 288) 190 (158, 222) 61 (43, 79) 256 (220, 293) 239 (203, 274) 17 (8, 27) 173 (143, 203)

129 (103, 155) 400 (354, 455) 358 (315, 401) 41 (27, 56) 365 (321, 408) 351 (309, 394) 10 (3, 17) 470 (420, 519)

3.3 (1.5, 5.2) 3.2 (2.1, 4.3) 7.3 (4.2, 10.3) 2.8 (1.4, 4.2) 18.6 (16.4, 20.9) 15.1 (6.9, 23.3) 7.7 (6.1, 9.2) 6.1 (4.8, 7.3) 14.7 (14.5, 15.0) 15.4 (15.1, 15.7) 6.1 (5.6, 6.5)

5.4 (3.3, 7.6) 5.0 (3.8, 6.2) 14.5 (10.3, 18.7) 7.3 (5.5, 9.1) 23.8 (21.7, 25.9) 19.7 (11.7, 27.7) 12.9 (11.2, 14.7) 8.6 (7.4, 9.9) 17.9 (17.7, 18.2) 18.9 (18.6, 19.2) 7.3 (6.8, 7.7)

16 (7, 26) 46 (31, 62) 27 (15, 40) 21 (11, 32) 295 (255, 335) 101 (78, 125) 117 (92, 142) 119 (94, 144) 15 240 (14 968, 15 512) 12 200 (11 954, 12 446) 952 (881, 1023)

1998

31 (19, 44) 89 (67, 110) 52 (36, 69) 81 (61, 102) 504 (453, 556) 129 (103, 155) 242 (207, 278) 234 (199, 269) 21 000 (20 682, 21 317) 17 220 (16 930, 17 509) 1245 (1165, 1325)

Note. CI = confidence interval.

Workers earning less than $25 000 annually, who constituted 52.7% of the health care personnel, were more than 3 times as likely as higher-income workers to lack coverage— 19.1% (90% CI = 17.9%, 20.3%) vs 6.0% (90% CI = 5.3%, 6.7%)—and accounted for 74.1% of all uninsured health personnel. Geographic variations in coverage of health personnel generally mirrored variations in coverage of the general population (Table 3). For instance, in Texas, both health personnel and the general population had the nation’s highest rates of uninsurance—21.5% (90% CI = 17.6%, 25.4%) and 24.5%, respectively—in 1998. Children residing with a health care worker accounted for an increasing percentage of all

uninsured children (Table 4). In 1988, 7.52 million children (90% CI = 7.33 million, 7.72 million) lived in a household with an adult health care worker. Of these children, 9.4% (90% CI = 8.6%, 10.1%)—or 705 000 (90% CI = 644 000, 766 000) children—were uninsured, accounting for 8.4% (90% CI = 7.7%, 9.1%) of all uninsured children. Ten years later, 9.77 million (90% CI = 9.51 million, 9.95 million) children lived in a household with a health care worker. Of these children, 11.5% (90% CI = 10.8%, 12.3%)—or 1.12 million (90% CI = 1.05 million, 1.20 million) children—were uninsured, accounting for 10.1% (90% CI = 9.5%, 10.8%) of all uninsured children in 1998.

406 | Research and Practice | Peer Reviewed | Case et al.

DISCUSSION Health care personnel and their children are losing health coverage more rapidly than are other Americans; 1.36 million health care workers provide care that they and their children cannot expect to receive. The surge in the number of uninsured health workers was driven by the private health care sector, which expanded substantially while insuring a shrinking proportion of its workers. Meanwhile, public-sector health employment decreased. This dynamic of job privatization was absent in other industries, in which public employment grew more briskly than private employment. We, and others,20

American Journal of Public Health | March 2002, Vol 92, No. 3


TABLE 2—Percentage and Number of Health Personnel Without Health Insurance, by Demographic Characteristics, 1988 and 1998 % Uninsured (90% CI)

White Black Hispanic Non-Hispanic Female Male Married Unmarried Annual family income, % of poverty level