Does a Patchwork Approach to Health Insurance Expansion Exacerbate Public Insurance Drop-Out? Benjamin D. Sommers, Ph.D. Harvard University AcademyHealth Annual Meeting June 26, 2005 Background: Children Without Health Insurance * 8.4 million children (11.4%) in U.S. have no health insurance * Over 6 million uninsured children are eligible for Medicaid/CHIP * 1 in 8 children “drops out” of Medicaid/CHIP annually – becoming uninsured, despite still being eligible Sources: Census Bureau 2004, Selden et al. 2004, Sommers 2005 Conceptual Framework: Why Do Kids Disenroll? Parents must reenroll annually to prove eligibility. Why might they not do so? ☑ Acquiring Other Health Insurance ☑ Loss of Eligibility ☑ ‘Drop-Out’: Uninsured & Still Eligible - Only “drop-out” is a program failure. Motivation & Objectives * Incrementalism is the current approach to health insurance expansion: Does this exacerbate the drop-out problem? Question 1: Do separate CHIP's (vs. combined Mcaid/CHIP programs) fare differently in retention? Question 2: Does covering parents improve the retention of children? Is This A Problem? Mom & Dad need not apply... Is this Medicaid, CHIP, or both? Data & Methods • Child & household data from the CPS March Supplement (2000-2004), using two-year linked samples • Since 2001, CPS asks about CHIP separately from Medicaid • State-specific eligibility rules from the National Governors Association • State policies for Medicaid/CHIP from state welfare offices and previous studies Results: Key Descriptive Statistics • As of 2001, 17 states + D.C. ran combined CHIP/Medicaid programs, and 33 ran separate CHIP's. • 23% of CHIP kids had a parent in public insurance, vs. 61% of Medicaid kids • Roughly 2 million children switched back & forth between Medicaid & CHIP each year • Drop-Out: CHIP 15.6% , Medicaid 12.5% Effect of Combined vs. Separate Medicaid/CHIP Program on Drop-Out Unadjusted Odds Ratio for 'Combined' 95% CI: Adjusted for + Demographics* Policies** 0.69 0.68 0.64 (0.51-0.92) (0.49-0.93) (0.43-0.95) N = 6526 % Reduced Risk = 31% * Adjusted for gender, age, race, parental education, family income, parental & sibling insurance coverage, health status, urban residence, state & year. ** Adjusted for demographics + the following Medicaid policies: frequency of renewal, face-to-face interview at renewal, presumptive eligibility, state renewal outreach, 12-month continuous eligibility, monthly premium, physician reimbursement rate, physician participation rate, MCO penetration, spending per child, charity care availability, & state political ideology. Effect of Family Coverage in Medicaid on Drop-Out Among Children* Variable Parent in Mcaid/CHIP % Reduced Risk Sibling in Mcaid/CHIP % Reduced Risk Logistic Odds Ratio 0.65 2-Stage IV Logistic** Odds Ratio 0.19 (p<.01) (p<.01) 28% 72% 0.60 0.74 (p<.01) (p<.10) 32% 20% n = 9020 NOTES: * Adjusted for gender, age, race, parental education, family income, number of children, health status, urban residence, state & year. ** Instrumented for parental and sibling Medicaid/CHIP coverage using parental and sibling Medicaid/CHIP eligibility. Limitations • Correlational only, for the combined vs. separate programs • 12 month snapshots don’t let us see: – How long children stay enrolled – If drop-outs ever return to Medicaid • Technical issues for CPS data: – Medicaid /CHIP Undercount – Monthly vs. Annual Income Data – Attrition Bias Policy Implications Separate programs are administratively costly (2 million kids switching annually) and exacerbate drop-out → States should consider combining Medicaid & CHIP, or at least streamline transitions between programs. Covering parents and kids separately doesn’t make sense in terms of retention → Cover families, not individuals. Acknowledgments This research was conducted with the support of a fellowship from the National Science Foundation. Many thanks to Joe Newhouse, David Cutler, Kathy Swartz, and Melissa Wachterman for excellent advice throughout this project.