The Role of Medicaid in Ensuring Children's Access to Care

BETWEEN 1986 and 1995, the number of Americans receiving Medicaid benefits increased from 23 million to 36 million.¹ Children have accounted for a disproportionate share of enrollment growth, primarily as a result of eligibility expansions mandated by the Congress in the mid 1980s and early 1990s.^{2 - 3}

The major Congressional actions taken to expand Medicaid eligibility for children are summarized in Figure 1. The first major legislative changes came in the mid 1980s, when Congress severed the historic link between welfare and Medicaid for children, allowing states to enroll certain groups of children into Medicaid who were not eligible for Aid to Families with Dependent Children (AFDC). Initially, states were given the option of expanding their Medicaid income eligibility thresholds for children from low-income families with federal Medicaid matching payments as the incentive. Although many states responded to these options, others did not.

Beginning in 1988, Congress enacted a series of laws mandating eligibility expansions at the state level to achieve uniform eligibility levels nationwide. Under these mandates, states were required to adopt the expanded eligibility provisions for children or risk losing their entire federal Medicaid matching grants (including funds for adult Medicaid beneficiaries). As a result of these laws, states must phase in Medicaid eligibility to all resident children younger than 19 years living in families with incomes below the official federal poverty level (133% of poverty for children younger than 6 years) by the year 2002. Currently, children younger than 15 years living in families with incomes below the poverty level are entitled to Medicaid. Next year the law will apply to children younger than 16 years and so on until all children younger than 19 years living in poor families are eligible for Medicaid in 2002.

Figure 2 illustrates the impact of these changes on children's enrollment in the Medicaid program. Over the decade spanning 1986 to 1995, the number of low-income children served by the program swelled from fewer than 10 million to more than 17 million.¹ The rapid upswing in enrollment growth began in the late 1980s and continued into the 1990s, paralleling the federally mandated expansions. Enrollment growth does not appear to be attributable to changes in the prevalence of poverty among American children. From 1986 to 1995, the proportion of children living in families with incomes below the poverty level increased marginally from 20.5% to 20.8%.⁴ Indeed, the slowdown in Medicaid enrollment growth from 1994 to 1995 may be attributable to continued strength in the US economy.

Poor children, the primary target of the Medicaid expansions, have historically lagged behind nonpoor children in access to care and use of health services. For example, prior to the Medicaid expansions, poor children had fewer physician contacts,^{5 - 7} were less likely to receive preventive care on a timely basis,^{7 - 8} and obtained routine care in qualitatively different settings than nonpoor children.^{9 - 10} By reducing or eliminating financial barriers, we would expect the rapid growth in Medicaid enrollment of children from poor families to lead to improvements in their access and use of care. In fact, studies using data collected prior to the large-scale expansions of Medicaid have demonstrated that the program was effective in removing at least some barriers to care for program recipients.^{11 - 14}

Now that the expansions enacted in the mid 1980s and early 1990s are largely implemented, it is important to understand how effective Medicaid has been in improving access to and use of health services for children in poor families. To address this question, we conducted a cross-sectional analysis of the relationship between Medicaid coverage and several new measures of access and use of physician services using data from the 1995 National Health Interview Survey (NHIS),¹⁵ the most current year available.

The NHIS is a continuing nationwide household survey that is conducted by the US Bureau of the Census for the National Center for Health Statistics.¹⁵ The purpose of the survey is to collect information on the health status and use of health services by the US civilian noninstitutionalized population. Each year the survey instrument consists of a core questionnaire on health status, utilization, and demographic characteristics, and a set of supplemental questionnaires on topics of current interest. During 1995, 2 special supplemental questionnaires on access to care and health insurance coverage were included in the survey.

Data on 29,711 children younger than 18 years from the core and supplemental surveys were used in this analysis to examine the effects of Medicaid on a variety of access and utilization measures (usual source of care, site of care, access to a regular clinician, ability to obtain needed care, physician contacts, and hospitalizations). This was done by comparing poor children with Medicaid coverage to poor children without insurance and nonpoor children with private health insurance. The latter group was selected to provide a benchmark for assessing the effects of Medicaid, since nonpoor privately insured children presumably experience the best possible access to care and therefore represent a high standard of comparison. In the same vein, poor children without insurance serve as a baseline for comparison. Without insurance and with limited family means to pay for services out-of-pocket, these children presumably experience the greatest financial barriers to care and thus provide a useful comparison group at the opposite end of the access spectrum. To simplify our presentation, we excluded poor children with private insurance, nonpoor children with Medicaid, and nonpoor children without insurance from the analysis. Results for these groups are available on request from the authors.

In the NHIS, an adult member of the household knowledgeable about the health of child family members serves as the respondent. In most cases, this is the mother. The response rate for the core survey was 93.8% and 91.5% for the insurance and access supplements. The compound response rate for all survey components was 85.8%. Item nonresponse, including missing and unknown responses, for each of the dependent variables was less than 10%. Cases with missing item responses were excluded from relevant estimates for the access and utilization variables.

Poverty status was assessed using the US Bureau of Census definition that takes into account family income from all sources and family size. In 1995, the poverty threshold for a family of 4 was $15,569.⁴ It should be noted that because the NHIS collects income data in bands (eg, $1000-$1999, $2000-$2999, and so on), the poverty status indicator is approximate. Medicaid and private health insurance status was assessed based on responses to questions included in the health insurance supplement about Medicaid and private insurance coverage status during the month prior to the interview date. A total of 17.8% of children had either missing income or health insurance responses and were excluded from the subgroup comparisons but are included in the table totals.

Estimates presented in the tables and text have been statistically weighted to reflect national population totals. The weights, provided by the data collection agency, are equal to the inverse of the sampling probability for each case, adjusted for nonresponse. Using the SUDAAN (Survey Data Analysis) software program,¹⁶ which takes into account the complex sample design and weighting structure of the NHIS, we conducted both bivariate (Table 1) and multivariate (Table 2) analyses of the access and utilization indicators described above.

The effect of Medicaid and poverty status on access and utilization is likely to be confounded by other demographic and health status variables. We adjusted for confounding using logistic regression analysis for categorical outcomes and ordinary least squares regression for continuous outcomes. Potential confounding variables included in the regression equations were selected based on a review of the theoretical^{17 - 23} and empirical literature^{22 - 26} on access to health care. These covariates included age, sex, race, ethnicity, family composition, region, population density, health status, disability, and restricted activity days. A second order term for restricted activity days was also included to improve model fit. Only the summary findings from the regression analyses are presented here; detailed results are available from the authors. Unless otherwise noted, only differences significant at the .05 level or higher are discussed in the text.

There are some important methodological limitations to this study. Because our analysis is based on cross-sectional data, it is impossible to draw definitive conclusions concerning causal relationships between insurance status and access to and utilization of care. Moreover, although our analysis of access and utilization differentials includes an adjustment for known confounding variables available in the data set, some confounding may remain unaccounted. For example, unmeasured factors that lead some poor children to enroll in Medicaid and other poor children to remain uninsured, even though they are eligible for Medicaid, may contribute to differences in outcome variables.

A usual source of care is defined as a place where a child can obtain routine preventive health care or treatment for illness and is considered a prerequisite for ensuring continuity of care. Overall, 94% of US children had a usual source of care in 1995 (Table 1). Poor children with Medicaid were far more likely than poor children without insurance to have a usual source of care (95.6% vs 73.8%). Although poor children with Medicaid lagged behind nonpoor children with private insurance on this measure of access, the difference was comparatively small (95.6% vs 97.4%).

Among children with a usual source of care, there was little difference between poor children with Medicaid and poor children without insurance in the distribution of sites. However, poor children with Medicaid received their care in different sites than those used by nonpoor children with private insurance. For example, poor children with Medicaid were only about two-thirds as likely as nonpoor privately insured children to receive their care in physician offices or health maintenance organizations (58.7% vs 85.9%). Instead, poor children with Medicaid were much more likely to receive their care in health centers or clinics (30% vs 11.8%), hospital outpatient clinics (6.5% vs 0.9%), and emergency departments (3.9% vs 1%).

Another aspect of continuity of care concerns whether children have access to the same clinician on repeat visits to their usual source of care. Some 13% of US children did not have a regular clinician at their usual source of care in 1995. Poor children with Medicaid were only about half as likely as poor children without insurance to be without a regular clinician (15.9% vs 27.2%). However, poor children with Medicaid were still more likely than nonpoor children with private insurance to lack a regular clinician (15.9% vs 9.1%).

Several items were included in the 1995 NHIS access questionnaire on ability to obtain needed health care. The results indicate that only small percentages of American children had unmet health care needs, at least as perceived by respondents to the survey (Table 1). There were, however, substantial differences in unmet needs according to income and insurance status. Poor children with Medicaid were much less likely than poor children without insurance to have unmet needs for medical care (2.1% vs 5.9%), dental care (4.4% vs 15.5%), and medications, eyeglasses, or mental health care (2.7% vs 7.1%). Nonetheless, poor children with Medicaid were more likely than nonpoor children with private insurance to have unmet needs for medical care (2.1% vs 0.6%), dental care (4.4% vs 2.8%), and medications, eyeglasses, or mental health services (2.7% vs 0.8%).

As shown in Table 1, poor children with Medicaid were much more likely to have at least 1 physician contact in the past year than poor children without insurance (83.9% vs 60.7%). In contrast, the percentage of poor children with Medicaid who used at least 1 physician service was essentially identical to nonpoor children with private insurance (84%). We also assessed volume of physician services for children with at least 1 contact in the past year. On this measure, poor children with Medicaid had more contacts than poor children without insurance (6.5 vs 4.0) and nonpoor children with private insurance (6.5 vs 5.2).

Finally, we compare use of inpatient hospital care (excluding deliveries) in Table 1. Poor children with Medicaid were much more likely to have at least 1 hospital episode in the past year than poor children without health insurance (4.7% vs 1.2%) and nonpoor children with private insurance (4.7% vs 2.6%).

The 3 analysis groups— poor children with Medicaid, poor children without insurance, and nonpoor children with private insurance—differ in important sociodemographic and health characteristics that are known to influence access and utilization of health care. Consequently, differences in these underlying characteristics could account for some of the access and utilization differentials described above. To adjust for potential confounding, we conducted a series of multivariate analyses using logistic and ordinary regression techniques. The results of these analyses are presented in Table 2. In this table, poor children without insurance are compared to poor children with Medicaid in the first column, while poor children with Medicaid are compared to nonpoor children with private insurance in the second column.

The first column of Table 2 shows that Medicaid is associated with a substantial improvement in all measures of access and utilization for children living below the poverty level. It is also apparent from the second column that poor children with Medicaid lag somewhat behind nonpoor children with private insurance on 4 of the access measures and 1 of the utilization indicators. However, the odds ratios for the comparisons in the second column are consistently smaller in magnitude than those in the first column. This implies that the gaps in access and utilization between poor children with Medicaid and nonpoor children with private insurance are smaller than those between poor children with Medicaid and poor children without insurance.

The rapid enrollment growth of children in the Medicaid program served as the impetus for this study. Our analysis focused on assessing Medicaid's role in improving access to care for poor children using a large, nationally representative sample of children. Because access is a multidimensional construct, we chose to use several measures of access and utilization to determine Medicaid's impact. Recognizing that Medicaid coverage is only one of many financial and nonfinancial determinants of access and utilization, we used multivariate techniques to account for the effects of a number of other potential determinants suggested by the theoretical and empirical literature.^{17 - 26}

Our analysis of the 1995 NHIS shows that when compared to poor children without insurance, poor children with Medicaid experience considerably better access and higher utilization rates. However, our results also indicate that Medicaid does not bring poor children's access entirely up to the level of nonpoor, privately insured children. Modest- to moderate-sized differentials remain in about half of the access and utilization indicators examined.

Understanding the factors that contribute to discrepancies in access and utilization between Medicaid and privately insured children is a necessary first step toward eliminating these remaining gaps. Medicaid programs across the nation have historically suffered from a number of structural and programmatic problems that have hindered their ability to secure high-quality care for children. Chief among these problems has been low levels of reimbursement and long delays in payment to clinicians.²⁷ In as much as reimbursement levels are related to the site of usual source of care,²⁸ it follows that low reimbursement rates would contribute to more children seeking care in locations other than private physician offices, such as health centers, hospital outpatient departments, and other clinics. Yet, the ability of these latter sites to offer children the full range of services that they need has long been impeded by funding shortages coupled with high volumes of patients with limited or no ability to pay. Consequently, some of the remaining disparities in access could be attributable to the much greater reliance of children enrolled in Medicaid on health centers, hospital outpatient departments, and other clinics as usual sites of care.

Dramatic shifts in the financing of children's health care are now leading to changes in where children enrolled in Medicaid obtain care and how that care is reimbursed. Specifically, with the large-scale transition of Medicaid from fee-for-service to managed care, children enrolled in Medicaid are increasingly receiving care in different settings than in the past. The proliferation of privately sponsored managed care arrangements has also lowered the amount of reimbursement that clinicians are receiving for patients with private insurance, diminishing the gap between privately and publicly sponsored patients.

The impact of these changes on poor children's access to care is unknown.²⁹ While managed care holds the potential for improving access,³⁰ the extent to which existing managed care arrangements can meet the needs of the Medicaid population has yet to be demonstrated. There is particular concern regarding the suitability of managed care for vulnerable Medicaid subpopulations, including children with disabilities, severe emotional disturbances, and language barriers.^{31 - 32} As states aggressively move to enroll their Medicaid populations in managed care, particularly in fully capitated arrangements, it will be important to monitor whether changes occur in access, utilization, and quality of care.

Our analysis suggests that Medicaid has played an important, if not always consistent, role in improving access to health care for children living in families with incomes below the poverty level. The salutary effects of the program may be heightened if states choose to implement several new Medicaid options legislated under the Balanced Budget Act of 1997.³³ Most significantly, states now have the option of immediately expanding Medicaid income eligibility thresholds up to 200% or more of the federal poverty level with substantially enhanced federal matching payments. States also have the option of guaranteeing Medicaid eligibility for 12 months, greatly reducing administrative costs and paperwork while encouraging continuity of care. Under prior law, states were required to redetermine eligibility whenever a beneficiary's financial circumstances changed. Because income fluctuations are common among low-income families, this requirement contributed to frequent interruptions in Medicaid coverage for children.³⁴ The new law also permits states to offer presumptive Medicaid eligibility for children. Under this provision, health care practitioners can initiate needed treatments after gathering limited information on family income and resources. Children remain eligible until the state Medicaid agency makes a final eligibility determination. Finally, the Balanced Budget Act provides a new source of funds for enrollment and outreach purposes. Given the large number of children eligible but not enrolled in Medicaid,³⁵ these funds could contribute to significantly enhanced participation of poor and near-poor children in the program.