Health Affairs, 23, no. 6 (2004): 170-180 doi: 10.1377/hlthaff.23.6.170 © 2004 by Project HOPE	New Online   Pay Cuts For Medicare Docs   Access To Care Woes   Public Coverage More Efficient   Empowering Consumers

This Article Abstract Reprint (PDF) Supplemental Exhibit Submit a response to this article Alert me when this article is cited Alert me when eLetters are posted Alert me if a correction is posted Services E-mail this article to a friend Similar articles in this journal Similar articles in ISI Web of Science Similar articles in PubMed Alert me to new issues of the journal Add to My Personal Archive Download to Citation Manager Reprints & Permissions Citing Articles Citing Articles via HighWire Citing Articles via ISI Web of Science (2) Citing Articles via Google Scholar Google Scholar Articles by Blewett, L. A. Articles by Rodin, H. Search for Related Content PubMed PubMed Citation Articles by Blewett, L. A. Articles by Rodin, H. Related Collections Access To Care Business Of Health Consumer Issues Health Reform Insurance Coverage - Children Maternal And Child Health State/Local Issues Health Spending

Health Tracking

TRENDS

Covering Kids: Variation In Health Insurance Coverage Trends By State, 1996–2002

Abstract

 
We estimated state-specific changes in health insurance coverage rates for children between 1996–1998 and 2001–2002. We found considerable variation in the changing distribution of health insurance coverage for children across states, with significant increases in public program coverage in twenty-nine states and significant decreases in uninsured children in twenty-seven. Children in families with incomes below 200 percent of the federal poverty level were the most likely to enroll in public programs. We provide an overview of state outreach and administrative simplification efforts and raise concerns about the persistent variation in children’s health insurance coverage across states.

Factors affecting public program enrollment. SCHIP, passed as part of the Balanced Budget Act (BBA) of 1997, represents the largest expansion of public health insurance programs since the 1965 passage of Medicare and Medicaid. As of June 2004, SCHIP insured approximately 3.7 million children.¹ Combined with the larger Medicaid program, SCHIP and Medicaid represent an important source of health insurance coverage for children, covering almost one in every four children in 2002 (23.9 percent).²

Section 1931 of the Social Security Act, enacted under the Personal Responsibility and Work Opportunity Reconciliation Act (PRWORA) of 1996, also presented new opportunities for states to increase public program enrollment of children. These new provisions increased administrative flexibility and opportunities to streamline eligibility and enrollment. States could raise income eligibility levels and earned-income disregards, eliminate certain asset tests, and extend coverage to two-parent working families.³ States could also provide presumptive eligibility for children up to a year old.⁴ Karl Kronebusch and Brian Elbel found that some of these administrative changes, including presumptive eligibility and self-declaration of income, had a sizable impact on increasing children’s enrollment in public programs.⁵

State outreach efforts also played a role in increasing coverage for children.⁶ State spending on outreach is eligible for the federal SCHIP match, and almost all states employed outreach programs, including television, radio, and print advertising, as well as activities in targeted enrollment sites such as schools, community clinics, and other provider-based settings.⁷

National outreach efforts were largely funded by the Robert Wood Johnson Foundation (RWJF) with its four-year, $55 million Covering Kids campaign to increase enrollment in Medicaid and SCHIP. As part of this effort, the Covering Kids and Families program coordinates an annual Back-to-School campaign. The 2002 campaign included 1,200 outreach and enrollment activities, resulting in 39,000 news stories, with corporate partners providing additional outreach.⁸ Genevieve Kenney and colleagues found increased familiarity with Medicaid and SCHIP programs and eligibility requirements among parents of low-income children.⁹

Evidence of increasing coverage for children. The distribution of health insurance coverage for children is changing, as documented by public program enrollment and national survey data. It should be noted that Medicaid/SCHIP administrative counts of the number of people enrolled do not often agree with the survey estimates of Medicaid coverage.¹⁰ Also, the national surveys cited below use different measures of health insurance coverage, which makes comparisons across surveys difficult. Despite these differences, all of the data point to similar trends: an increase in public program enrollment and a decrease or stable rate of uninsurance for children.

A study using data from the National Health Interview Survey (NHIS) shows a decline in the proportion of uninsured children between 1997 and 2002, from 13.9 percent to 10.1 percent. At the same time, public coverage for children increased from 62.1 percent to 69.3 percent for children in poor families (under 100 percent of poverty), and from 24.3 percent to 42.2 percent for children in near-poor families (100–199 percent of poverty).¹¹

The most current data from the CPS-DS show that from 2000 to 2003, uninsurance among children remained stable, while that among adults increased. There was also a more than 4.7 percent increase in public health insurance coverage for all children.¹² The Urban Institute’s analysis of data from the National Survey of America’s Families (NSAF) show a decrease in the uninsurance rate of nearly 6 percent for children in low-income families from 1999 to 2002, as well as an overall decrease in the number of uninsured children, from 9.6 to 7.8 million.¹³

Finally, the nationally representative Community Tracking Study (CTS) Household Survey shows that between 1998–1999 and 2000–2001, uninsurance among children declined by nearly 20 percent, and public coverage for low-income children (under 200 percent of poverty) increased from 28.9 percent to 36.7 percent.¹⁴ A recent update found that these trends continued from 2001 to 2003, with the uninsurance rate of low-income children falling an additional 4.3 percent to 11.4 percent and public coverage for low-income children increasing to 49.3 percent.¹⁵

Our analysis examines trends in private health insurance coverage, public coverage, and uninsurance at the state level, controlling for economic and other factors, to illustrate how the distribution of coverage for children has changed over time. We extend previous analyses by examining the dynamics in coverage and presenting state-level comparative data on the impact for two time periods, 1996–1998 and 2001–2002. Our analysis focuses on aggregate rates of change in health insurance coverage at the state level. We do not decompose the variation across states by program-specific characteristics; we acknowledge the need for additional research in this area, to better understand the local dynamics that affect health coverage.

Study Data And Methods

Top Study Data And Methods Study Results Discussion Editor's Notes NOTES:

 
Data. We use data on children age eighteen and younger from the CPS-DS for two time periods: 1996–1998 and 2001–2002.¹⁶ In 2000 the CPS-DS greatly altered how health insurance coverage data were obtained by adding a verification question. This question directly asks respondents who did not report having a specific type of coverage whether they are uninsured. Although it would be preferable to have verified data for 1996–1998, the only way to ensure consistent measurement between the two time periods is to "unverify" the data for 2001–2002.¹⁷

Variables. The dependent variable is the child’s type of health insurance coverage, classified as public, private, or uninsured. Public coverage includes all children reported as being covered by Medicaid, SCHIP, state programs, and Medicare.¹⁸ Private coverage includes children covered by employer-based, privately purchased, and military health insurance. We classify children as being covered by public insurance if they have both private and public coverage in both time periods, because surveys tend to undercount the Medicaid population and people on Medicaid tend to report that they have some type of public insurance.¹⁹ The uninsured category includes all children for whom a specific type of coverage is not reported.

Analysis. The explanatory variables in our model include characteristics of children and the families with whom they live. Children’s characteristics include age, health status, sex, poverty level, urban or rural residence, race, Hispanic ethnicity, and native-born status. Family characteristics include work status, highest education achieved by any family member, parents’ marital status, and age of family reference person. Also included are state dummy variables and state time-period interaction variables. Although additional variables could be added to further refine the model, their addition would not greatly alter the robust results we obtained.

We estimated a multinomial logistic regression predicting whether a child was uninsured, covered by public insurance, or covered by private insurance for each time period. We included family economic and demographic characteristics along with the child’s demographic characteristics as covariates. We used Stata version 7’s multinomial logistic regression procedure to adjust the parameter estimates for the complex sample design of the CPS, with the lowest level of identifiable geography as the stratum and the household as the cluster.²⁰ Using the coefficients derived from the multinomial logistic regression model, we formed two counterfactuals using a generalization of the "recycled probabilities" methodology.²¹ This procedure allows one to compare mean rates of uninsurance between the two time periods, controlling for significant factors likely to influence rates of uninsurance. The method of recycled probabilities produces adjusted health insurance coverage rates that take into account factors such as age, income, education, employment, and access to employer coverage. The procedure enables apples-to-apples comparisons within states from one time period to another, controlling for other relevant variables that may influence coverage rates. For a multinomial logistic regression model with a categorical dependent variable (being uninsured, privately insured, or publicly insured), the method of recycled probabilities is the appropriate statistical tool to construct adjusted averages.

We examined adjusted and unadjusted differences in the overall rates of public coverage, private coverage, and uninsurance between the two time periods. To compute significance tests for differences between the rates of uninsurance, we used an extension of the delta method to obtain estimates of the standard errors of the differences in recycled means (specific to each respondent-characteristic).²²

Study Results

Top Study Data And Methods Study Results Discussion Editor's Notes NOTES:

 
Exhibit 1 provides the unadjusted and adjusted rates of change in health insurance coverage for children, by state. The adjusted rates are based on the recycled rate of change in private coverage, public coverage, and uninsurance, controlling for family and child characteristics likely to influence coverage. Using the adjusted results, we found significant decreases in the percentage of uninsured children between the two time periods for twenty-seven states and significant increases in the number of children enrolled in public insurance in twenty-nine states. We found fewer states with significant changes in the percentage of children with private health insurance coverage: Twelve states had a significant decrease in private coverage for children, while five states had a significant increase.

Top Study Data And Methods Study Results Discussion Editor's Notes NOTES:

State rollbacks. Although state revenues improved somewhat in 2004, twenty-two states still report cost overruns in Medicaid.²³ And even though children are typically not a large part of the Medicaid outlays (representing 49 percent of Medicaid enrollees but only 16 percent of Medicaid spending), children’s programs are not protected from budget cuts, as evidenced by recent state action.²⁴

As part of its fiscal year 2004 budget, Massachusetts capped enrollment in its Children’s Medical Security Plan (CMSP) and raised premiums, resulting in an estimated increase of 20 percent in the number of uninsured children.²⁵ As of November 2003, six states had enacted SCHIP enrollment freezes, which may increase inequities in coverage among children. Higher-income children have retained coverage, while lower-income children and eligible newborns have lost coverage.²⁶ In another recent study, seven of thirteen state officials interviewed indicated that their states had drastically cut funding for outreach programs.²⁷

These cuts are starting to show up in recent enrollment data. New data from the Centers for Medicare and Medicaid Services (CMS) show a decline in SCHIP enrollment of 200,000 children, from 3.9 to 3.7 million children between June 2003 and March 2004.²⁸ During the first six months (June–December 2003), Texas alone accounted for 52 percent of the decline.²⁹ This followed major legislative action in Texas that restricted eligibility, placed limits on enrollment, and increased premiums and cost sharing for SCHIP.³⁰

We can anticipate more cuts in 2004. The state fiscal relief provided by Congress—$10 billion in general fiscal relief for unrestricted purposes and $10 billion in a temporary increase in the federal matching payments to states—expired 30 June 2004.³¹ As part of the relief package, states were required to maintain Medicaid eligibility levels that were in effect as of 2 September 2003. However, states are no longer restricted from making changes to Medicaid program eligibility to achieve budget savings. To balance state budgets, policymakers will face the choice of raising taxes to increase program revenue, eliminating existing coverage programs, or reducing program eligibility or benefits, or both.³²

Related issues. Our study also highlights the potential benefit of a more thorough examination of state program policies, the role of employer-sponsored insurance (ESI), and implications of public program expansions’ crowding out private insurance. It will be difficult to tease out the specific effects of SCHIP on reducing the number of uninsured children, given the complexities of isolating the impact of SCHIP and all of the other factors affecting the change in coverage occurring at the same time, including the extent of state and national outreach, Medicaid program changes, and the type of SCHIP program implemented. For example, of the twenty-nine state programs with significant increases in public program participation in our analysis, the type of SCHIP program is fairly evenly distributed: Eleven states implemented a Medicaid expansion program; ten, a separate SCHIP program; and eight, a combined program.³³

Recent research has highlighted the variation of ESI across states largely because of individual employment and labor-market characteristics as well as other community contextual effects.³⁴ State variation in ESI also helps explain the persistent state variation in coverage rates. And in terms of policy solutions, states have had little impact on employers’ decisions to offer health insurance coverage—leaving the public programs such as SCHIP and Medicaid to fill the gaps.³⁵

There is also ongoing concern among policymakers about the degree to which public coverage crowds out private employer-based or individual coverage when new or expanded public programs are made available.³⁶ Our analysis indicates that there may be some limited substitution; ten states had a significant decrease in private coverage and a corresponding significant increase in public coverage. Yet, interestingly, five states showed a significant increase in private coverage, with three of these showing a comparable decrease in levels of uninsurance.

Policy recommendations. Our research has highlighted the impact that new programs and policies have had on increasing health insurance coverage for children. But it has also demonstrated great variation in these gains by state. Providing a more equitable distribution of coverage of children across states will require not only additional revenue but a concerted bipartisan commitment from policymakers. Perhaps such an effort could initially target coverage for all low-income children (below 200 percent of poverty). Such an effort would require new state and federal financial support, an increase in federally mandated minimum eligibility levels, and innovations to get and keep children enrolled.

We recommend, initially, that national policymakers consider extending federal fiscal relief to states to support and maintain states’ progress toward reducing the number of uninsured children and to explicitly move toward more equitable coverage of children across states, starting with children in families with the lowest incomes.

Editor's Notes

Top Study Data And Methods Study Results Discussion Editor's Notes NOTES:

 
Lynn Blewett (blewe001{at}umn.edu) is an associate professor in the Division of Health Services Research and Policy, School of Public Health, University of Minnesota, in Minneapolis. Michael Davern is an assistant professor there, and Holly Rodin, a graduate research assistant and doctoral candidate.

The authors thank Linda Bilheimer of the Robert Wood Johnson Foundation (RWJF); Mary Kennedy, Medicaid director, Minnesota Department of Human Services; Gestur Davidson and Kristin Dybdal at the State Health Access Data Assistance Center (SHADAC); and three anonymous reviewers for their comments on an earlier draft of this paper. This analysis was funded by a grant from the RWJF to SHADAC at the University of Minnesota, Division of Health Services Research and Policy, School of Public Health. The views expressed are those of the authors and do not necessarily reflect the views of the RWJF.

NOTES:

Top Study Data And Methods Study Results Discussion Editor's Notes NOTES:

 

1. Centers for Medicare and Medicaid Services, "FY 2004 First Quarter—Program Enrollment Last Day of Quarter by State—Total SCHIP," 10 June 2004, cms.hhs.gov/schip/enrollment/2004pit1qt.pdf (9 August 2004).
2. R.J. Mills and S. Bhandari, "Health Insurance Coverage in the United States: 2002," Current Population Reports, September 2003, www.census.gov/prod/2003pubs/p60-223.pdf (28 June 2004).
3. M. Ellwood, "The Medicaid Eligibility Maze: Coverage Expands, but Enrollment Problems Persist: Findings from a Five State Study," Pub. no. 2164, September 1999, www.kff.org/medicaid/2164-index.cfm (17 August 2004).
4. CMS, Supporting Families in Transition: A Guide to Expanding Health Coverage in the Post-Welfare Reform World, 13 December 2002, cms.hhs.gov/medicaid/welfareref/welfare.asp (17 August 2004).
5. K. Kronebusch and B. Elbel, "Simplifying Children’s Medicaid and SCHIP: What Helps? What Hurts? What’s Next for the States?" Health Affairs 23, no. 3 (2004): 233–246.[Abstract/Free Full Text]
6. T.M. Selden, J.S. Banthin and J. Cohen, "Waiting in the Wings: Eligibility and Enrollment in the State Children’s Health Insurance Program," Health Affairs 18, no. 2 (1999): 126–133; and [CrossRef][Medline]T.M. Selden, J.S. Banthin and J. Cohen, "Medicaid’s Problem Children: Eligible but Not Enrolled," Health Affairs 17, no. 3 (1998): 192–200.[CrossRef][Medline]
7. M. Perry et al., Marketing Medicaid and CHIP: A Study of State Advertising Campaigns, October 2000, www.kff.org/medicaid/2213-index.cfm (17 August 2004).
8. Covering Kids and Families, "Communication Tools: About Back-to-School," 2004, coveringkidsandfamilies.org/communications/bts/about (26 March 2004).
9. G. Kenney, J. Haley, and A. Tebay, "Familiarity with Medicaid and SCHIP Programs Grows and Interest in Enrolling Children Is High," Snapshots of America’s Families 3, no. 2, 31 July 2003, www.urban.org/url.cfm?ID=310817 (17 August 2004).
10. K.T. Call et al., "Uncovering the Missing Medicaid Cases and Assessing Their Bias for Estimates of the Uninsured," Inquiry 38, no. 4 (2002): 396–408; and State Health Access Data Assistance Center, "Do National Surveys Overestimate the Number of Uninsured? Findings from the Medicaid Undercount Experiment in Minnesota," Issue Brief no. 9 (Minneapolis: University of Minnesota, School of Public Health, January 2004).
11. R.A. Cohen, H. Ni, and C. Hao, "Trends in Health Insurance Coverage by Poverty Status among Persons under Sixty-five Years of Age: United States, 1997–2002," 26 February 2004, www.cdc.gov/nchs/products/pubs/pubd/hestats/insurance.htm (28 June 2004).
12. U.S. Census Bureau, "Health Insurance Historical Tables," Table HI5, 2004, www.census.gov/hhes/hlthins/historic/index.html (17 September 2004).
13. J. Holahan and M. Wang, "Changes in Health Insurance Coverage during the Economic Downturn: 2000–2002," Health Affairs, 28 January 2004, content.healthaffairs.org/cgi/content/abstract/hlthaff.w4.31 (15 March 2004).
14. P.J. Cunningham, "SCHIP Making Progress: Increased Take-up Contributes to Coverage Gains," Health Affairs 22, no. 4 (2003): 163–172.[Abstract/Free Full Text]
15. B. Strunk and J. Reschovsky, "Trends in U.S. Health Insurance Coverage, 2001–2003," Tracking Report no. 9, August 2004, www.hschange.com/CONTENT/694/?topic=topic01 (3 August 2004).
16. The CPS-DS health insurance items pertain to the prior calendar year. The data for the 1995–1997 time period are drawn from the 1996–1998 CPS-DS surveys, and the data for the 2001–2002 time period are drawn from the 2002–2003 CPS-DS surveys.
17. C.T. Nelson and R.J. Mills, "The March CPS Health Insurance Verification Question and its Effect on Estimates of the Uninsured" (Presentation at the Annual Meeting of the American Statistical Association, Atlanta, Georgia, August 2001).
18. Children can be eligible for Medicare if they qualify under Social Security disability definitions. Medicare includes a very small number of children.
19. Call et al., "Uncovering the Missing Medicaid Cases."
20. M. Davern et al., "Evaluating Various Methods of Standard Error Estimation for Use with the Current Population Survey’s Public Use Data" (Presentation at the Section of Survey Research Methods, Joint Statistical Meetings, San Francisco, California, 2 August 2003).
21. Stata Corp, STATA Statistical Software: Release 7.0 (College Station, Texas: Stata Corp, 2001). The recycled estimates are a form of "model based" estimate that are similar to regression analysis predicted values or "small area estimates." Multivariate statistical models of these type allow analysts to "borrow strength" from other similar observations in the data set. This strength greatly reduces the variance of the recycled estimate (or coefficient in a regression model) from a straight direct survey estimate (such as an unadjusted rate of coverage in time 1 versus time 2). Statistical models allow for a great reduction in the variance of an estimate relative to the direct survey estimates. See National Research Council, Committee on National Statistics, Small Area Income and Poverty Estimates: Priorities for 2000 and Beyond (Washington: National Academies Press, 2000). They also have the advantage of holding constant a large amount of observed child and family characteristics that the direct survey estimates do not.
22. W.H. Greene, Econometric Analysis (Upper Saddle River, N.J.: Prentice Hall, 2003).
23. National Conference of State Legislatures, "Fiscal Storm Shows Signs of Subsiding," Press Release, 21 November 2003, www.ncsl.org/programs/press/2003/pr031121.htm (25 March 2004).
24. Henry J. Kaiser Family Foundation, "Kaiser State Health Facts Online: Medicaid and SCHIP," www.statehealthfacts.org/cgi-bin/healthfacts.cgi?action=compare&category=Medicaid+%26+SCHIP&welcome=1 (23 June 2004).
25. Mass Health is the name used for both Medicaid and SCHIP programs in Massachusetts. Children’s Health Access Coalition, "Cutting Children’s Care: How Capped Enrollment and Premiums Have Put Children’s Health Care out of Reach" (Boston: Health Care for All, 17 December 2003).
26. D.C. Ross and L. Cox, "Out in the Cold: Enrollment Freezes in Six State Children’s Health Insurance Programs Withhold Coverage from Eligible Children," 11 December 2003, www.kff.org/medicaid/4159.cfm (28 June 2004).
27. I. Hill, H. Stockdale, and B. Courtot, "Squeezing SCHIP: States Use Flexibility to Respond to the Ongoing Budget Crisis," New Federalism: Issues and Options for the States, Series A, no. A-65, 2 June 2004, www.urban.org/url.cfm?ID=311015 (17 August 2004).
28. CMS, "FY 2004 First Quarter."
29. V.K. Smith, D.M. Rousseau, and M. O’Malley, "SCHIP Program Enrollment: December 2003 Update," July 2004, www.kff.org/medicaid/7134.cfm (17 August 2004).
30. A. Dunkelberg and M. O’Malley, "Children’s Medicaid and SCHIP in Texas: Tracking the Impact of Budget Cuts," July 2004, www.kff.org/medicaid/7132.cfm (17 August 2004).
31. These provisions were included in the Jobs and Growth Tax Relief Reconciliation Act of 2003. Federal matching rates are calculated using measures of state and national average incomes. In FY 2002, five states had the highest Medicaid matching rates of 70 percent or more with an enhanced matching rate of 80 percent (MI, WV, NM, MT, and AR); eleven states had the lowest Medicaid matching rates of 50 percent and an enhanced matching rate of 65 percent; (NY, NJ, NH, NV, MN, MA, MD, IL, DE, CT, and CO).
32. V.K. Smith et al., "States Respond to Fiscal Pressure: A Fifty-State Update of State Medicaid Spending Growth and Cost Containment Actions," January 2004, www.kff.org/medicaid/7001.cfm (17 August 2004).
33. V.K. Smith and D.M. Rousseau, "SCHIP Program Enrollment: June 2003 Update," December 2003, www.kff.org/medicaid/4148.cfm (17 August 2004). Additional comparative state policy information on SCHIP is available in online Supplemental Exhibit 1, content.healthaffairs.org/cgi/content/full/23/6/170/DC1.
34. R. Kronick, T. Gilmer, and T. Rice, "The Kindness of Strangers: Community Effects on the Rate of Employer Coverage," Health Affairs, 2 June 2004, content.healthaffairs.org/cgi/content/abstract/hlthaff.w4.328 (17 August 2004); and Y. Shen and S. Zuckerman, "Why Is There State Variation in Employer-Sponsored Insurance?" Health Affairs 22, no. 1 (2003): 241–251.[Abstract/Free Full Text]
35. Shen and Zuckerman, "Why Is There State Variation?"
36. G. Davidson, L.A. Blewett, and K.T. Call, "Public Program Crowd-Out of Private Coverage: What Are the Issues?" Report no. 5, June 2004, www.rwjf.org/publications/synthesis/reports_and_briefs/issue5.html (17 August 2004).

                      What's this?

This article has been cited by other articles:

				D. Hargunani Healthcare Coverage for Children: Are We Making Progress? AAP Grand Rounds, April 1, 2005; 13(4): 46 - 46. [Full Text] [PDF]