By Anna S. Sommers, Lisa Dubay, Linda J. Blumberg, Fredric E. Blavin, and John L. Czajka

By Anna S. Sommers, Lisa Dubay, Linda J. Blumberg, Fredric E. Blavin, and John L. Czajka

H E A L T H T R A C K I N G :
7 August 2007

Dynamics In Medicaid And SCHIP Eligibility
Among Children In SCHIP's Early Years:
Implications For Reauthorization
Fixing the problem of coverage discontinuity will
require continuing the strong state and federal partnership
that has characterized SCHIP's success in its first decade.

by Anna S. Sommers, Lisa Dubay, Linda J. Blumberg,
Fredric E. Blavin, and John L. Czajka


Two-thirds of children in the United States were income-eligible for Medicaid or the State Children's Health Insurance Program (SCHIP) at some point from 1996 to 2000. One in five children were income-eligible for both programs, and 73 percent of children ever eligible for SCHIP were eligible at other times for Medicaid. As SCHIP is reauthorized, Congress will need to give states the tools and financial commitment to assure that uninsured children are enrolled in and retain the coverage for which they are eligible. [Health Affairs 26, no. 5 (2007): w598-w607 (published 7 August 2007; 10.1377/hlthaff.26.5.w598)]

In 2004 approximately six million children were eligible for public health insurance through Medicaid or the State Children's Health Insurance Program (SCHIP) but remained uninsured at least part of the year.1 These numbers reflect one of the biggest challenges facing the public health insurance system for children today: how to reduce the number of uninsured eligible children and increase continuous coverage for children enrolled. These issues are a major policy concern because the uninsured are more severely ill when diagnosed and receive less therapeutic care than the insured, resulting in higher mortality.2 Parents of uninsured children report worrying all the time that their children will get hurt and need medical care.3 Children with discontinuous coverage are more than thirteen times as likely to delay care as are children who are continuously insured.4

Addressing these problems is complicated by the existence of the dual and sometimes competing programmatic structures of Medicaid and SCHIP. Only nine states and the District of Columbia administer SCHIP solely as a Medicaid expansion, relying on the same benefits and provider networks but applying different eligibility rules.5 In all other states, certain populations are enrolled in a separate SCHIP program, with more limited benefits and a different provider network, and some families are required to pay copayments and premiums. These separate programs enrolled 72 percent of the SCHIP caseload by the end of fiscal year 2006.6 Switching programs in these forty-one states may require a child to change providers, necessitate paying a premium, or lead to loss of specific benefits or disenrollment. Even when the burden of switching programs is minimized for families through administrative simplifications, it still remains a cost to states.

Churning in and out of public coverage is a well-documented burden in Medicaid.7 In addition, about two million children move between Medicaid and SCHIP each year.8 Less well understood is the underlying eligibility dynamics of children, which could be of a much larger magnitude than is apparent from studying enrolled populations. In 2002, thirty-six million children were eligible for either program during the year.9 Because eligibility depends on a child's age, family income, and assets, and, in SCHIP, access to private coverage, a change in any of these factors can trigger changes in eligibility.

The number of children potentially affected by these dynamics through coverage disruptions is determined by the overall size of the programs and policies that minimize these disruptions. If a child is enrolled, twelve-month continuous enrollment protects him or her from loss of coverage for a time. Awareness education can help dispense misconceptions. For instance, some parents view eligibility as "static" and after one denial do not reapply when income changes; others believe that working families are not eligible.10 Therefore, eligibility dynamics have the potential to affect access to care and continuity of coverage to a far greater degree if current policies that minimize their consequences are underfunded. As the reauthorization of SCHIP spurs renewed debate about how to reduce uninsurance among children, it is important to understand the number of children potentially affected.

In this study we analyze eligibility dynamics in Medicaid and SCHIP for children over the period 1996-2000, using data from the 1996 panel of the Survey of Income and Program Participation (SIPP). We present estimates of the share of children eligible for Medicaid, SCHIP, or both programs over the four-year panel. We examine the types of transitions in eligibility that children made and what happened to children before and after they first became eligible for SCHIP. We also explore insurance coverage among eligible children. Although this analysis was conducted using data from early in SCHIP's infancy, we discuss the implications of our findings for Congress as it seeks to reauthorize the now mature SCHIP program.

Study Data And Methods

Data. The 1996 SIPP panel collected information over a forty-eight-month period for a nationally representative sample of the U.S. civilian noninstitutionalized population. The panel was divided into four rotated samples, of which one rotation was interviewed in any given month beginning in April 1996 and ending in March 2000. Respondents were asked questions about the preceding four months, resulting in a set of monthly observations covering twelve waves of four months each for the period December 1995 through February 2000. The core questionnaire is built around labor-force participation, public program participation, income, and health insurance coverage of each person in the sample household.

We used data from the most recent reference month in the wave from each interview to construct all variables in our analysis.11 Eligibility "in a wave" refers to data from the most recent reference month in a wave.12

Eligibility algorithm. We used the Transfer Income Model version 3 (TRIM3) to impute eligibility in Medicaid and SCHIP for all children age eighteen and younger.13 Although the TRIM3 model was designed for use with the Current Population Survey (CPS), we mapped data from SIPP to the CPS data elements required by TRIM3 to produce estimates of eligibility for Medicaid and SCHIP using SIPP data.

TRIM3 incorporates state-specific rules for Medicaid eligibility under each of the following eligibility routes: Section 1931, Section 1115, medically needy, Supplemental Security Income (SSI) and Social Security Disability Income (SSDI), poverty-related expansions, and SCHIP.14 The simulation incorporates rules related to family structure, work and child care disregards, gross and net income thresholds, and categorical designations.

We did not model all spend-down criteria for medically needy eligibility because of limited data; nor did we model eligibility based on immigration status, because data on the legal status of noncitizens were not available in the 1996 SIPP panel. Eligibility for SCHIP included Medicaid-expansion programs under SCHIP and separate SCHIP programs based on rules in place during the wave. We did not model eligibility for SCHIP based on access to employer-sponsored insurance. Otherwise eligible children with employer coverage could enroll in SCHIP only if they became uninsured. Therefore, our estimates represent children who are income-eligible based on data on household structure, income, and assets.

We classified eligibility into two broad categories--eligible for Medicaid and eligible for SCHIP--and we divided the sample into three groups based on whether children were eligible for all waves in which they were present in the panel (always eligible), eligible for some but not all waves in which they were present (sometimes eligible), or not eligible for any waves in the panel (never eligible). A child could not be eligible for both programs during the same wave but could be eligible for both during the same spell.

Analysis sample. Our goal was to produce estimates of eligibility over a period of up to four years for a representative sample of children who were in the SIPP universe at the beginning of the panel or who were born to people in the SIPP universe at the beginning of the panel. We limited our sample to children who were age eighteen or younger at the start of the panel (N = 21,888). Some children turned nineteen over the course of the panel. We did not exclude these children from the analysis, but we excluded all waves after the child reached age nineteen (2,651 cases). Infants who were born to original panel members after the start of the panel were included (n = 3,226). Families were defined as those family members who can obtain private health insurance together under one policy (that is, spouse, dependent child under age nineteen in the home, full-time student under age twenty-three).

Limitations. Longitudinal data are vulnerable to bias from attrition. The 1996 SIPP panel data provided a longitudinal panel weight that accounted for asymmetric attrition across demographic groups and by a number of other characteristics; however, it did not adjust for some attributes relevant to our study, and the weight was constructed only for cases with no missing waves.15 We made several adjustments to the SIPP weights to address these issues, including the assignment of weights to children born after the start of the panel, and an adjustment for the underrepresentation of young mothers in the sample. These adjustments greatly improved the representation of the sample to minimize the effects of attrition in later waves.16 Second, our longitudinal sample is representative of the civilian population in 1996. Although we accounted for some growth in the population due to newborns through weighting, we could not account for population changes due to immigration.

Study Results

Overview of children's eligibility. Over the full course of the four-year panel, 66 percent of all children (fifty-nine million) were eligible for either Medicaid or SCHIP at some point during the panel (Exhibit 1). By comparison, point-in-time estimates are much lower and show that 33 percent of all children in the first wave of the panel and 46 percent in the last wave were eligible for either program.

Exhibit 1.
View larger version
[in this window]
[in a new window]

Medicaid and SCHIP serve many of the same children at different times (Exhibit 2). More than one-third of all children who were ever eligible for Medicaid during the panel were also eligible for SCHIP in other waves, while nearly three-fourths of all children who were ever eligible for SCHIP during the panel were also eligible for Medicaid in other waves.

Exhibit 2.
View larger version
[in this window]
[in a new window]

Overall, 32 percent of all eligible children were eligible for both Medicaid and SCHIP during the course of the panel, representing about one-fifth of all children, or almost nineteen million children. Another 57 percent were eligible only for Medicaid, and 12 percent were eligible only for SCHIP. About one in five children (18 percent) were always eligible for either Medicaid or SCHIP in every wave of the panel, while 48 percent of all children were eligible for only some waves, and the remaining third were never eligible.

The sometimes eligible children represent 73 percent of children who were eligible at any point, while the always eligible constitute 27 percent. About half of the sometimes eligible became SCHIP-eligible, compared with 29 percent of the always eligible. Eligibility for public coverage is often characterized by interruptions in eligibility. All of the sometimes eligible were ineligible for both programs in at least one wave, and 56 percent had two or more distinct spells of eligibility. Overall, 41 percent of all eligible children had multiple spells of eligibility. This does not account for children who were eligible for Medicaid and experienced changes in income or family structure that would lead to a change in how they became eligible for Medicaid (that is, welfare- versus poverty-related rules). These changes could lead families to believe that they are no longer eligible for Medicaid or could require new documentation.

Characteristics of eligible children. Exhibit 3 presents individual and family characteristics of the always, sometimes, and never eligible children in their first and last waves of the panel.17 Along a wide array of dimensions, children who were always eligible were more disadvantaged than those who were sometimes or never eligible. The always eligible were disproportionately infants (35 percent), compared to the sometimes (16 percent) and never (14 percent) eligible, and they were also disproportionately black or Hispanic. Parents of the always eligible were more likely younger, with 25 percent under age twenty-five, versus 7 percent and 1 percent of parents of sometimes and never eligible children, respectively. The always eligible were more likely to live in a family with no working adult by more than forty percentage points compared with the sometimes and never eligible children, and about 80 percent lived in families with income below the federal poverty level, compared with only 25 percent of sometimes eligible and 3 percent of never eligible children.

Exhibit 3.
View larger version
[in this window]
[in a new window]

The most notable change between the first and last waves occurred among always eligible children, for which the share of children with no working adult in the family dropped from 55 percent to 44 percent. Most of these families added one full-time worker. A smaller shift occurred among the sometimes and never eligible, of whom slightly larger shares reported two full-time workers by the last wave. Consistent with these changes, the share of always eligible children with incomes below poverty dropped from 80 percent to 69 percent, while larger shares of the sometimes and never eligible children reported incomes of 300 percent of poverty or higher in the last wave than in the first wave.

Children eligible for SCHIP. Of the children eligible for SCHIP in at least one wave of the panel, 68 percent were eligible for Medicaid in some wave prior to becoming eligible for SCHIP, while 33 percent had been covered by Medicaid (Exhibit 4). In the wave immediately preceding the first SCHIP-eligible wave, 39 percent were eligible for Medicaid, while the remainder was ineligible for both programs. Also in this wave, 25 percent of children had family income below poverty, and another 36 percent had income of 100-199 percent of poverty. Thus, it appears that most children newly eligible for SCHIP were already poor or low income prior to becoming eligible, while only 16 percent came from families with incomes of 300 percent of poverty or higher. During an economic contraction, a greater share of newly eligible children would likely come from higher-income families.

Exhibit 4.
View larger version
[in this window]
[in a new window]

Of those we could observe after the first wave they became eligible for SCHIP, two-thirds remained eligible for SCHIP for some period of time thereafter, including 23 percent who also became eligible for Medicaid as well. About 39 percent cycled to a spell of Medicaid eligibility. Seventeen percent were eligible for SCHIP in just one wave and were never eligible for either program in the panel again. These findings suggest that many of these families experience instability in family income that affects eligibility for public programs. Importantly, many changes were downward even though overall economic conditions were improving during the study period.

Insurance coverage among eligible children. Finally, we explored insurance coverage rates for children during eligible waves (Exhibit 5). Many eligible children--42 percent--faced at least some period of uninsurance while eligible, although only 7 percent remained uninsured for the entire period they were eligible. Forty-four percent of eligible children were covered by Medicaid or SCHIP during at least one eligible wave, while 66 percent were covered by private insurance.

Exhibit 5.
View larger version
[in this window]
[in a new window]

The sometimes eligible, which represents a higher income group that was more often eligible for SCHIP than the always eligible, were more likely than the always eligible to have been covered by private insurance at some point (78 versus 36 percent) and less likely to have been covered by public insurance (30 versus 81 percent). The always eligible were more likely than the sometimes eligible to have been uninsured at some point, even if ineligible periods are taken into account for the sometimes eligible children (data not shown), yet more than one-third of the sometimes eligible experienced a spell of uninsurance. Families of these two eligible populations differed with respect to family resources available to weather transitions without disruption in coverage. The sometimes eligible children were more protected by the employer-sponsored market, and the lower take-up into Medicaid/SCHIP reflects this.