In this country, >12.2 million children under 22 years old do not have health insurance (Table 1). It is well documented that children who lack health insurance are less likely to have access to needed health care services such as preventive care, are less likely to use health care services, and are more vulnerable to health problems.¹

Table 1. Children and Youth in the United States Without Health Insurance, by Age and Family Income, as a Percentage of the Federal Poverty Level*, 1987, 1990, and 1993  [View Table]

The Medicaid program, a joint federal/state program that finances health care for low-income vulnerable Americans, insures more children in this country than any other public or private entity.⁵ Beginning in 1989, states were required to increase their Medicaid programs by expanding eligibility to pregnant women and children. Today, all states must cover pregnant women and children to age 6 up to 133% of the federal poverty level (FPL) and children through age 13 up to 100% of the FPL. In addition, 38 states in 1995 had adopted Medicaid eligibility expansion options that went beyond the federal mandates. Also, 36 states have medically needy programs that provide at least some Medicaid services to persons who also meet the nonfinancial standards for inclusion in one of the groups covered under Medicaid, but who do not meet the income or resource requirements for categorically needy eligibility.⁶

Today, in an effort to balance the federal budget, Congress is debating proposals that will restructure the Medicaid program as an entitlement program to children. Some estimate that proposed budget cuts could eliminate Medicaid coverage for as many as 4.4 million children nationwide and cut federal Medicaid funding by $182 billion over 7 years, reducing funding to the states by 30% in 2002.⁷ But before any action is taken, policymakers and others should recognize the enormous impact this program has had on the pregnant women and children in each state. For example, in fiscal year 1994, more than half (54%) of Medicaid beneficiaries were children, although they only accounted for 23% of total Medicaid expenditures.⁶ Also in fiscal year 1994, Medicaid financed the health care for one-fourth of all children in the United States under 21; one-third of all US births; one-third of all children under 6; and one-half of all infants.^6,8

Several years ago, we developed a methodology for pooling multiple years of Current Population Survey (CPS) data to estimate the number of uninsured children in each state.⁹ Those data have been used extensively by state policymakers in assessing the effectiveness of Medicaid expansions. This article updates our previous work by 1) examining national estimates of uninsured children, by age and family income, at three points in time: 1987, 1990, and 1993; 2) analyzing changes in the number and proportion of uninsured children in each state between 1987- 1990 and 1991-1993; and 3) looking at the number and proportion of children in each state with employer-based/private health insurance and Medicaid coverage.

We report three major findings. First, our data indicate that overall the number of uninsured children grew by nearly 1 million between 1987 and 1993. The proportion of infants and children <7 years old without health insurance declined; the proportion of uninsured children between the ages of 7 and 22 increased. Second, we observed noteworthy state variation in the number and proportion of uninsured children; between 1987 and 1990 and 1991 and 1993, six states experienced a significant increase in the number of uninsured children, only two states experienced a significant decrease in the number of uninsured children, although no progress was found in reducing the number of uninsured children in 43 states. Finally, we found a significant decrease in the proportion of children covered by employer-based private insurance in three-fourths of the states and a significant increase in Medicaid coverage in four-fifths of the states during this period.

METHODS

Pooling CPS Data

Although the March CPS has been a popular data source for state and national estimates of demographic and economic characteristics,¹¹ small sample sizes pertaining to less populated states yielded estimates that vary in precision from state to state. Pooling together multiple years of CPS data is one strategy to increase state-level sample sizes. This method, however, complicates the computation of sampling errors due to two characteristics of the March CPS sample design; namely, year-to-year overlap and panel-to-panel dependence.

To facilitate the reader's interpretation of state level estimates and to conform to National Center for Health Statistics survey reporting guidelines, (Sirken MG, Shimizu I, French D, Brock D. Manual on Standards and Procedures for Reviewing Statistical Reports. Washington, DC: National Center for Health Statistics, Department of Health, Education, and Welfare, [unpublished information]) a method was developed to compute sampling errors for level and percent distribution estimates based on pooled CPS data. This method, described in detail in a technical appendix (this technical appendix on computing standard errors of the state estimates is available from the authors at the American Academy of Pediatrics, 141 Northwest Point Boulevard, PO Box 927, Elk Grove Village, IL 60009-0927), adjusts for year-to-year overlap and panel-to-panel dependence in the March CPS sample to provide unbiased measures of sampling errors for state level analyses. In this report, sampling errors (measured in standard errors) are reported for all state level estimates; standard errors >20% of the estimates are noted.

Determining Insurance Status

Consistent with the Census Bureau, the estimates presented for the proportion of persons who are covered by Medicaid include the small percentage of nonelderly persons who are covered by Medicare (for children, generally <1%). Family income is derived from the total family income variable computed by the Census Bureau that includes all sources of income. (Total family income as computed by the Census Bureau includes all earnings, investment returns, unemployment and worker's compensation, child support, alimony, public assistance and welfare, social security, supplemental security income, disability, and retirement income. We did not adjust total family income by family size).

Interpreting Insurance Rates

RESULTS

Table 2 and Table 3 describe the number and proportion of uninsured, employer-based/privately insured and Medicaid covered children in each state from 1987 to 1990 and 1991 to 1993. We have chosen to report both numbers and proportions for the following reasons: 1) by reporting the number of uninsured children, the reader is given a sense of the actual magnitude of the child population at risk, and 2) by reporting proportions, the reader is able to compare state-by-state variations in child health insurance coverage.

Table 2. Number* of Uninsured, Employer-based/Privately Insured, and Medicaid Covered Children and Youth in Each State, 1987-1990 and 1991-1993 [View Table]

Table 3. Percentage of Uninsured, Employer-based/Privately Insured, and Medicaid Covered Children and Youth in Each State, 1987-1990 and 1991-1993 [View Table]

Six states saw a significant increase in the number of uninsured children, two states experienced a significant decrease in the number of uninsured children, and 43 states showed no significant change in the number of uninsured children during this period (Table 2). The proportion of uninsured children changed significantly in nine states during the period under study, rising in five states and declining in four (Table 3). However, the dynamics driving the rise and fall of the uninsured child population lie in the changing employer-based/private and Medicaid insured populations.

In 37 states, the proportion of children with employer-based/private health insurance declined significantly during the period under study (Table 3). In 41 states, the proportion of children with Medicaid coverage increased significantly during this same period. No states experienced a significant increase in employer-based insurance and no states experienced a significant decrease in Medicaid coverage during this time.

Seventeen of 19 states suffering from a significant decrease in the number of employer-based/privately insured children, experienced simultaneously a significant increase in their Medicaid coverage (Table 2). Subsequently, 14 of those 17 states reported no significant increase in the number of uninsured children despite significant drops in employer-based/private health insurance for children. Although we are not implying that the Medicaid program picked up the health care coverage of those individuals who lost their dependent health insurance from an employer, the Medicaid expansions were effective in stabilizing the number of uninsured children in the states with two exceptions; 1) where the loss of employer-based/private insurance outpaced the Medicaid expansions or 2) where child population growth outpaced the increase in Medicaid coverage. Of the six states experiencing a significant increase in their uninsured child population, three states (New Jersey, Illinois, and Michigan) experienced significant drops in employer-based/private health insurance coverage for children and suffered a concurrent increase in their uninsured child population, despite significantly higher Medicaid coverage. Massachusetts was the only state to experience a significant drop in employer-based/private health insurance without a simultaneous increase in Medicaid coverage. Maine and Nevada, on the other hand, experienced simultaneously a significant increase in Medicaid coverage and substantial growth in their uninsured child population.

CONCLUSIONS

Second, we observed noteworthy state variation in the number and proportion of uninsured children; between 1987-1990 and 1991-1993, six states experienced a significant increase in the number of uninsured children, only two states experienced a significant decrease in the number of uninsured children, although no progress was found in reducing the number of uninsured children in 43 states.

Finally, the proportion of children with employer-based health insurance declined significantly in three-fourths of the states, although the proportion of children with Medicaid coverage increased significantly in more than four-fifths of the states during the study period. In light of current proposals to reduce and restructure the Medicaid program, it is no longer possible to expect the Medicaid program to decelerate the growth in the uninsured population.

Newacheck et al pointed to an alarming trend of children losing private health insurance coverage at an accelerating rate, due to reductions in employer-based dependent health insurance coverage.¹⁶ Employer-based health insurance coverage of children aged 0 to 18 declined 11.5% in the 15 years between 1977 and 1992, from 67.7% to 56.2%. In the same article, they found that between 1988 and 1992, Medicaid expansions worked to offset the contraction in employer-based coverage during the same period, resulting in relative stability of the proportion of the uninsured child population. Based on the trends mentioned above, the authors observed that families losing private coverage because of the changing economy or employer cutbacks in benefits for dependents will increasingly rely on Medicaid as a coverage of last resort. They went on to suggest that policymakers should consider these trends as they debate the future of the Medicaid program.

By pooling together March 1988 to 1994 CPS Surveys, we were able to examine the pervasiveness of these trends, namely eroding employer-based/private insurance coverage for children being offset by Medicaid expansions, in the individual states. This analysis also enables us to examine the consequences proposals being debated by Congress may have on the nation's different regions and states. With employer-based/private health insurance declining significantly in three-fourths of the states and Medicaid coverage significantly increasing in four-fifths of the states, our findings clearly point to 1) Medicaid's role as a vital safety net and 2) the fact that the Medicaid expansions are working. Since 1989, >4.3 million children have enrolled in the Medicaid program as a result of the expansions, bringing to 17.6 million the total number of children enrolled in the program.

But even with increased Medicaid enrollment, the number of uninsured children still grew by nearly 1 million in the past 7 years. As many as 30 million children in this country are at risk of either losing their current coverage or remaining uninsured. Three elements are at work here: 1) absent a reversal in the decline of employer-based private insurance coverage for dependents, millions of children will continue to lose coverage; 2) if Congress does indeed decide to cut Medicaid funding and shrink the entire program, millions of children will continue to lose coverage; and 3) if the Medicaid program is restructured through block grants to the states, the scope of benefits offered and services provided will be reduced and millions of children will continue to lose coverage. As policymakers debate the future of the Medicaid program, they must consider seriously its role as the country's largest insurer of children's health and its efficacy as a vital safety net for the nation's children.

As our earlier study stated, our data can be used to provide states with comparable baseline information from which states' progress in reducing uninsured rates among children can be measured. As this study shows, children in many states are at greater risk today than they were between 1987 and 1990, despite the laudable efforts made by states to expand their Medicaid programs. If this country's largest insurer of children's health care is jeopardized, then our country's future is also at risk.