Recent evidence suggests that children in working poor families lack health resources, placing them at risk for inadequate access to care.
To examine financial and nonfinancial access and utilization of health services among children in working poor families, and to compare these data with those of children from both nonworking poor and moderate to affluent families.
A cross-sectional study of 13 785 children younger than 18 years.
Subjects from the 1997 National Health Interview Survey.
Main Outcome Measures
Prevalence and continuity of health insurance coverage, of delayed or missed care, and of unmet care needs; presence and type of usual source of care; and the amount of visits to physicians, emergency departments, and hospitals.
Compared with children of nonworking poor parents and moderate to affluent children, more working poor children were uninsured (22% vs 12% and 5%, respectively; P<.01) and experienced disruptions in insurance coverage (P<.01). After adjusting for other covariates, disparities in insurance coverage and continuity persisted, as did delays in care and unmet care needs; these delays were far higher for the working poor. Although these children had access to a regular source of care and had utilization rates comparable with those of other poor children, they differed markedly from moderate to affluent children on structural access and utilization (adjusted odds ratios, 1.5-3.4).
Children in working poor families experience far more barriers to care than other children. Health insurance expansions through the Children's Health Insurance Program and Medicaid, which reduce financial and nonfinancial barriers to care, may help correct these disparities.
ALTHOUGH access-to-care barriers have been amply documented for poor children in general, it is not clear to what extent children in low-income working families—the working poor—experience similar barriers to care. In 1997, approximately 25 million children in the United States lived in working families that earned less than 200% of the poverty level. In these families, approximately 2 of 3 parents were fully attached to the workforce. These children are usually grouped with other impoverished children, with the uninsured, or with children who experience unmet health care needs, rendering the working poor a virtually invisible population.1- 4 The sheer size of this population underscores the need to examine whether work, especially full-time year-round (FTYR) parental work, pays off in terms of improved access to care for their children. An examination of this issue is timely and compelling given recent welfare and immigration policies that place a strong emphasis on personal responsibility and work.5 New expansions in health insurance coverage through the Children's Health Insurance Program (CHIP) are making insurance more available to the working poor. To assess the effects of CHIP on access to care for children of the working poor requires baseline information.
In California, children of the working poor are less likely to have insurance coverage and more likely to experience disruptions in coverage than children of welfare families or nonworking poor families, even after controlling for other covariates.6 Having an FTYR working parent does not improve insurance coverage or use of health care. These children's access and utilization are much worse than those of children from moderate to affluent families. A recent national study indirectly points to the unmet needs of children of the working poor by demonstrating that uninsured children eligible for CHIP are far more likely to have 2 employed parents than children enrolled in Medicaid.2 However, scant information exists nationally on the joint effects of work and income on access to care for children.
Poor and near-poor children are at increased risk of having unmet health care needs, yet these cannot be addressed through health insurance alone.1 In addition to gaps in insurance coverage, which increase the chances that a child will receive health care from multiple sites,7 nonfinancial barriers to care exist for these children. These barriers are partly due to a high reliance on community clinics and county and city hospitals.8 Many of these facilities have not been funded to keep pace with the growth in the uninsured population and have longer waiting times for available appointments, excessive waiting room times, and provider shortages. Having a regular source of care can increase continuity; regular providers are more likely to know the child's medical history and are better able to monitor treatment through follow-up visits.9 The lack of a regular care source and of health insurance has been associated with an increase in emergency department use.10,11 Low levels of education among low-income working parents may lead to different health investments in their children compared with children from more affluent families with more education and disposable income.12,13
The purpose of this article is to examine the joint effect of parental work and income on children's health care access and utilization. We focus on low-income children of parents in the labor force who earn less than 200% of the poverty level (the working poor) and compare them with 2 other groups of children: those in low-income families who do not participate in the labor force (nonworking poor) and those from families in the labor force who earn more than 200% of the poverty level (moderate to affluent). We explore the extent to which work pays off by addressing 3 questions: Do children of working poor families differ from the 2 reference groups with respect to financial and structural dimensions of access to care? Are there differences in utilization of health services among the groups? Does insurance coverage improve access and utilization among children of the working poor? This analysis also provides a baseline for evaluating the effects of CHIP and the 1996 welfare reform act. Given the lack of consensus among policymakers regarding who constitutes the working poor, we offer 2 profiles of children of the working poor: (1) those who had at least 1 parent who was currently working or was employed for at least 6 months in the past year; and (2) those who had at least 1 parent working FTYR for at least 1 year at a main job 35 hours or more per week and earning 200% of the poverty level or less (ie, an annual income of $31 822 for a family of 4 using 1996 poverty thresholds).
Data were obtained from the National Health Interview Survey (NHIS), a continuing cross-sectional survey of the civilian, noninstitutionalized US population. Different households are sampled each year, and interviews are administered face-to-face by trained Census Bureau staff using computer-assisted technology.14 We used 1997 NHIS data to create the baseline against which to evaluate the future effects of CHIP, which was implemented in 1998.
In 1997, the NHIS randomly selected 1 child per family for a more in-depth interview on access and utilization. Of the 15 350 families with eligible children younger than 18 years, 14 290 had interview data on a selected child. Mothers or other adult family members provided information about their children; however, 17-year-olds could respond for themselves. Compared with children of respondents, children of nonrespondents were more likely to be African American and in good rather than excellent health. They did not differ with respect to disability status, nativity status, receipt of public assistance, or insurance coverage.
Sampled children's records were linked to their parents' records to determine parental work status. Children who did not live with a parent (n = 487) were excluded. After excluding records with missing work status, our sample size was 13 785. We imputed poverty level for 1640 children who had missing information on income and could not otherwise be classified into a study population. Imputation randomly assigned a poverty level category (in 25% increments) to individuals with similar geographic region, parental education, race/ethnicity, and broad income level if available.5 Recipients of Temporary Assistance to Needy Families (TANF) were assumed to have incomes less than 200% of the poverty level. The analytic sample consisted of 5342 children in working poor families (including a subsample of 3394 children [63.5%] living in FTYR working families), 1021 children in nonworking poor families, and 7422 moderate to affluent children living with parents earning 200% of the poverty level or more, 99% of whom were working.
Access to care was measured by financial and structural characteristics. Financial access included current coverage and disruptions in coverage during the past year. Several questions were used to classify children into the categories of private coverage, public coverage, or uninsured. Among the privately insured, children were classified into those insured through the parent's employer or "self-pay." Continuous insurance coverage was determined by asking respondents, "In the past 12 months, was there any time when [child] did not have any health insurance or coverage?"
Structural measures of access included the presence or absence of a regular source of care: a particular person or place to which a child usually goes for treatment when sick, for health advice, or for routine or preventive medical care. Parents of children with a usual source of care were asked about the kind of place they go. Parents were also queried about any experience in which their child's care was delayed or missed in the last 12 months because of financial constraints, and about any experiences in which their child needed medical or dental care, prescriptions, eyeglasses, or mental health care but was unable to get it.
Utilization variables included entry into care, measured by a visit to a physician or other health care professional in the last 12 months and stratified by perceived health status into children with excellent and less than excellent health. Among those with at least 1 visit in the past year, we assessed whether children had 1 visit, 2 to 3, or 4 or more; whether they had 2 or more emergency department visits; and the mean number of hospital admissions, all stratified by perceived health status.
Independent variables included demographic characteristics of the child such as age, sex, and race/ethnicity, as well as child health measures including perceived health status and the presence of disability, defined as a reduction in capacity to perform the average type or amount of daily living activities. Additional characteristics consisted of family structure and size, and receipt of public assistance. Socioeconomic status was measured by highest level of parental education, federal poverty level, and parental employment. Geographical characteristics included region of the United States and whether the child lived in a metropolitan statistical area (MSA).
We first compared the 3 study populations by demographic and health characteristics, family and geographic characteristics, access barriers, and utilization measures. We then did similar bivariate comparisons of the children of FTYR working parents and the 2 reference populations. We compared each of the working poor groups to each of the 2 reference groups, and we indicate statistically significant (P≤.05) differences in the tables. Further bivariate analyses allowed us to examine differences between the insured and uninsured among low-income children of working parents. We conducted multivariate logistic regression analyses to determine whether the differences in dichotomous outcomes between the 3 study populations (and between the FTYR working group and the 2 reference populations) persisted after adjusting for other variables. Linear regression was used to analyze the mean number of hospital admissions, adjusting for other factors. We adjusted for child's age, race/ethnicity, parental education, family structure, region, MSA, child's disability, and health status except where already stratified. These variables have been shown to affect health care access and utilization.3,4,8,11,15 We then added health insurance to the model.
Because the design of the NHIS is a complex, multistage sample, the SEs and significance tests are weighted to reflect population totals and corrected for stratification and sampling clustering. The analyses were conducted using SAS version 6.12 (SAS Institute Inc, Cary, NC) and SUDAAN version 7.5.3 (Research Triangle Institute, Research Triangle Park, NC) statistical software.
Compared with moderate to affluent families with predominantly 2 incomes and college-educated parents, children in working poor families were more likely to have an unemployed parent or only 1 parent employed (P<.01), with less than a completed high school education (P<.01) (Table 1). Consequently, their income levels were lower (P<.01). Forty-six percent were living below the federal poverty level, whereas 54% had incomes falling between 100% and 200% of the poverty level. However, children in working poor families were not as poor as those from nonworking low-income families, 81% of whom lived in poverty (P<.01). The differences are similar when comparing FTYR working poor families with the 2 reference groups.
Only 13% of working poor families received public assistance (TANF) compared with 54% of nonworking poor families (P<.01). The differential between working poor and nonworking poor families in public assistance use was even larger among FTYR earners (P<.01). Working poor families were almost 4 times more likely to have 2 parents in the home compared with nonworking poor families (P<.01), but less likely than moderate to affluent families (P<.01). A large family size was also more common among the working poor (including the FTYR group) than the reference groups (P<.01).
Compared with moderate to affluent children, those in working poor families were younger (P<.01), 3 times more likely to be Hispanic (P<.01), twice as likely to be African American (P<.01), and less likely to be Asian American (P<.06). They were less likely to be perceived to have excellent health status (P<.01) and more likely to be disabled (P<.03). Comparable differences were found between children in the FTYR working poor group and moderate to affluent children, except for a similar prevalence of disability. Compared with children in nonworking poor families, those in working poor families (including those in FTYR working families) were less likely to be African American (P<.01) and more likely to be white non-Hispanic (P<.01), and to enjoy better perceived health status (P<.01) and less disability (P<.01). Additionally, fewer working poor children were concentrated in MSAs compared with the other 2 groups (P<.01). Children of the working poor were more likely than moderate to affluent children to live in the West (P<.01) and South (P<.01), and were more likely than nonworking poor children to live in the South (P<.01).
Children in working poor families were far more likely to be uninsured (22%) than children in nonworking poor families (12%; P <.01) or moderate to affluent families (5%; P<.01) (Table 2). In addition, compared with children of nonworking poor families, the working poor were less likely to be covered by Medicaid (31% vs 78%; P<.01). Medicaid income eligibility levels vary by state, but eligibility is mandatory for children aged 1 to 5 years in families that earn 133% of the poverty level or less and for older children from families with incomes of 100% of the poverty level or less. Compared with moderate to affluent children, children of the working poor were less likely to be privately insured (47% vs 90%; P<.01). Only 44% had employment-based coverage in comparison with 86% among moderate to affluent children (P<.01); self-paid coverage was also lower (P<.03). Furthermore, disruptions in insurance coverage were almost twice as high among children of the working poor compared with the nonworking poor (P<.01), and more than 3 times higher than among moderate to affluent children (P<.01). Although parental FTYR work did not improve the prevalence of insurance coverage among working poor children, it improved the proportion of private coverage (P<.01).
Far more children in working poor than nonworking poor families (7% vs 4%; P<.04) or moderate to affluent families (7% vs 3%; P<.01) experienced delayed or missed care in the previous year because of financial constraints (Table 2). Although children of working poor families were as likely as children of nonworking poor families to get needed medical care, mental health care, or prescriptions, access to dental care was more of a problem for the working poor (P<.01). Access to all dimensions of care was much more difficult for children of the working poor (P<.01), including those of FTYR working parents (P<.01), than for moderate to affluent children.
The prevalence of a regular source of care was similar for children of working and nonworking low-income families (90% vs 91%), but this percentage was significantly lower than for moderate to affluent children (97%; P<.01). These differentials were also significant for children of FTYR working poor families compared with moderate to affluent children (P<.01). Among those having a regular source of care, children of working poor families were far more likely than children of nonworking poor families to seek care in a physician's office or health maintenance organization (P<.01) and less likely to seek care in a clinic or health center (P<.01) or hospital ambulatory setting (P<.03). Compared with working poor children who sought care in clinics or hospitals, those who sought care in the private sector were less likely to forgo needed medical care (P<.05) and prescription medications (P<.02) and more likely to have visited a physician in the past year (P<.02). Compared with moderate to affluent children, children of working poor families (including children in FTYR working families) were less likely to seek care in the private sector (P<.01) and more likely to seek care in health clinics (P<.01), hospitals (P<.01), or emergency departments (P<.01). Among children with no regular source of care, those in working and nonworking poor families sought care predominantly in community clinics or health centers, whereas moderate to affluent children sought care in physicians' offices (data not shown).
Children in working poor families who were in excellent health were as likely as children in nonworking poor families to have entered the health care system in the past year, but far less likely than moderate to affluent children (P<.01) (Table 2). However, among working poor children in less than excellent health (including FTYR working poor families), 16% had not visited a physician in the past year, a rate significantly worse than for the other 2 reference groups (12%; P<.05 and 9%; P<.01).
Among children who had visited a physician in the past year, fewer children in the FTYR working group had 4 or more visits compared with the moderate to affluent children, whether in excellent health (P<.01) or in less than excellent health (P<.04). Furthermore, children in working poor families (including those in FTYR working families) were far less likely than those of nonworking poor families to have 2 or more emergency department visits in the past year (P<.05) but were far more likely than moderate to affluent children (P<.01), irrespective of health status. A similar mean number of hospital admissions was found among the 3 study populations, irrespective of health status.
Structural access and utilization were further stratified by insurance coverage for children of the working poor, which is the population targeted for CHIP. As expected, insured children were far more likely than uninsured children to have a regular source of care (P<.01), to enter the health care system (P<.01), and to have a higher number of physician visits (P<.01) (data not shown). In contrast, uninsured children were far more likely than insured children to delay or miss care because of financial constraints (P<.01) and to forgo needed medical care (P<.01), dental care (P<.01), mental health treatment (P<.02), and prescription medications (P<.01) (data not shown). Similar differentials by insurance status were found among the FTYR subgroup.
Compared with children of nonworking poor parents, children of working poor parents were more likely to lack insurance coverage. These differences persisted after controlling for child's age, race/ethnicity, health and disability status, family structure, parental education, region, and MSA (odds ratio [OR] = 1.8; 95% confidence interval [CI], 1.4-2.4) (Table 3). Among insured children, disruptions in insurance coverage were more frequent for children of the working poor than the nonworking poor, even after adjusting for other covariates (OR = 2.0; 95% CI, 1.3-3.1). Although children of the working poor were as likely as children from nonworking poor families to lack a regular source of care, the working poor were somewhat more likely to seek care in a physician's office or health maintenance organization, after adjusting for other covariates (OR = 1.2; 95% CI, 0.9-1.4). The working poor were also more likely than the nonworking poor to delay or miss care because of financial constraints and to express unmet needs for care, after adjusting for other covariates (OR = 1.8; 95% CI, 1.2-2.8 vs OR = 1.5; 95% CI, 1.1-2.1, respectively). After adjusting for other demographic and health status covariates, poor children of working parents did not differ significantly from poor children of nonworking parents on measures of health care utilization, except for a lower likelihood of entering the health care system among the working poor (OR = 1.2; 95% CI, 0.9-1.6). These results did not differ according to health status and were not altered by adjusting further for insurance status. Similar patterns were observed when comparing low-income children in FTYR working and nonworking families.
The disparities in access and utilization were far greater for children of the working poor compared with moderate to affluent families, with the exception of a high number of physician visits (≥ 4, conditional on entry into the health care system) and the number of hospitalizations, which were both similar. The differentials narrowed, then persisted after adjusting for demographic factors and health status, and continued to narrow (particularly for structural access) after controlling further for insurance status (Table 4). Similar but narrower discrepancies were found when comparing children of FTYR working poor families with those from moderate to affluent families.
This article portrays the sociodemographic characteristics, patterns of health care access, and utilization rates among children of the working poor. It shows that compared with poor children from nonworking families and moderate to affluent children, children of the working poor experience far more difficulties in accessing health care. In addition, utilization of health services is markedly lower than for moderate to affluent children. Collected the same year CHIP was established, these data from 1997 can be a useful baseline from which to evaluate improvements in access to care for working poor children through health insurance expansions.
Specifically, the findings show that 22% of children in working poor families are uninsured compared with 12% of children in nonworking poor families and 5% in moderate to affluent families. Even after adjusting for demographic factors and health status, children of the working poor are less likely to have health insurance compared with children in the other 2 reference groups. For these children, parental employment does not provide adequate access to employment-based coverage, yet it reduces the likelihood of eligibility for Medicaid.
Full parental commitment to the labor force marginally increases employer-based coverage. Only 50% of the children of FTYR working poor parents have this type of private coverage, whereas 22% remain uninsured. Disruptions in coverage among those insured children are 2 to 3 times more prevalent among children of the working poor compared with children in nonworking poor and moderate to affluent families. Even children of FTYR working poor parents have odds of coverage disruptions that are 1.7 times higher than those of children from nonworking poor families and 2.1 times higher than those of moderate to affluent children, after adjusting for other factors. Welfare-to-work policies that seek to increase work incentives for poor parents must address insurance barriers by not only making insurance available to these families, but also making it more continuous. Recent evidence suggests that because many people now view public assistance as a temporary benefit, fewer recipients of TANF are applying for Medicaid or maintaining enrollment.16
Lack of insurance and continuity in coverage among children of the working poor contributed to a larger proportion of these children having to delay or miss care because of financial constraints compared with the reference child populations. These differentials are not reduced by adjusting for demographic and health factors but are somewhat offset by health insurance coverage. Recent expansions in health insurance coverage through CHIP may help to reduce the disparities in timely access to care.
Despite these disparities, children of the working poor were as likely as those from nonworking poor families to have a regular source of care, and they had similar utilization of health services after adjusting for confounders. As the results indicate, one reason for this favorable outcome may be related to a significantly higher use of physician offices or health maintenance organizations by children of the working poor compared with children of the nonworking poor, who tend to seek care in the public sector. Children who seek care in the private sector tend to have better structural access and utilization. As different states implement and expand CHIP insurance programs, they should attempt to facilitate enrollment with private providers or delivery systems.
Our findings also indicate that children in working poor families have much less structural access to care and use of health services than children in moderate to affluent families, even after adjusting for other factors. Being fully attached to the labor force reduces the discrepancy but does not equalize children's access to care or utilization of health services. If health insurance were available, disparities in structural access and entry into care could narrow. Nevertheless, the results of the regression models using moderate to affluent children as the reference group suggest that health insurance provision may not suffice to remove the barriers to adequate care for children of working poor families. An additional challenge for health insurance programs like CHIP will be to ensure that subscribing health plans offer providers and delivery systems incentives to provide adequate care.
The Children's Health Insurance Program was designed to provide health insurance to approximately 5 million eligible uninsured children in families with incomes falling below 200% of the poverty level. By providing funds to expand Medicaid or to establish new programs, Congress has given states flexibility in designing and implementing their programs. Although CHIP plans are progressing in most states, enrollment has fallen short of expectations because of multiple barriers. These include the complexity of administrative and eligibility structures, cost-sharing provisions that may discourage enrollment, poorly designed outreach programs, insufficient efforts to ensure that children maintain coverage once they are enrolled in CHIP or Medicaid, and poor coordination among CHIP, Medicaid, and TANF programs.17 These access barriers, in addition to cultural and linguistic barriers, need to be addressed if CHIP is to cover the target population. Because states with high uninsurance rates also have lower levels of average health status and more access problems, they may require more resources to achieve success in health care reform.18 Future studies should continue to monitor statewide differences in access to care for children of the working poor.
The baseline NHIS data have some shortcomings. The measure of health insurance is complex and subject to parental reporting error. In addition, the insurance measure reflects coverage at the time of the interview, whereas health care use is measured in the last 12 months. Health care utilization recall over a 12-month period may also be subject to error.19 The type of health plan, which may affect access and utilization, was not examined because of concerns about inaccurate reporting.20 In cross-sectional studies, the causal direction of the relationship between variables cannot be ascertained. Moreover, adjusting for demographics and other factors does not preclude the possibility that unmeasured factors could account for the observed differentials for the working poor. Although income was imputed for 12% of the sample, pairwise relationships between access-to-care variables and the study groups were similar for those with imputed and nonimputed income.
Despite these limitations, this study demonstrates that compared with other child populations that have been studied much more extensively, children in working poor families face serious barriers to care. Access and utilization of health services for the working poor need to be carefully monitored as welfare and health insurance reforms provide new challenges and opportunities for working poor families in the United States.
Accepted for publication February 4, 2001.
This study was funded by the Henry J. Kaiser Family Foundation, Washington, DC, to support the ongoing work of the Kaiser Commission on Medicaid and the Uninsured.
We thank Connie Gee for clerical support and Steven Samuels, PhD, for statistical consultation.
Corresponding author and reprints: Sylvia Guendelman, PhD, Division of Health Policy and Management, 404 Warren Hall, University of California, Berkeley, Berkeley, CA 94720-7360 (e-mail: firstname.lastname@example.org).
Guendelman S, Pearl M. Access to Care for Children of the Working Poor. Arch Pediatr Adolesc Med. 2001;155(6):651–658. doi:10.1001/archpedi.155.6.651