Effects of the State Children’s Health Insurance Program (SCHIP) on Access to Medical Care and Use of Medical Services
Annual Meeting of the AcademyHealth Orlando, FL, June 2007
Hua Wang
Cornell University
Edward C. Norton, R. Gary Rozier UNC-Chapel Hill
Background
Health insurance reduces relative health care cost and increases demand for care
About 18% of children whose family incomes were between 150-199% FPL were uninsured in 1994-1996
Congress created SCHIP in 1997 to further expand public health insurance to uninsured low-income children All states implemented SCHIP by 2000
– 6.1 million children enrolled by the end of 2005 FY
SCHIP Differs from Medicaid
Income eligibility: on average
–Upper limit: 215% FPL –Medicaid: 155% FPL for children<6; 100% FPL for those>5
States have more flexibility in program design & implementation (Medicaid expansion or separate) More aggressive outreach and simplified enrollment procedures
Mandated coordination in outreach and enrollment
–Increases Medicaid enrollment (spillover effect)
Research Questions
Did SCHIP implementation improve lowincome children’s access to medical care and use of medical services? Addressed from two perspectives:
– Low-income children’s perspective: What is the effect of SCHIP implementation on the outcomes of interest for all low-income children? – Enrollees’ perspective: Did enrollment in public insurance programs as a result of SCHIP implementation improve outcomes of interest for enrolled children?
Previous Studies
Low-income children’s perspective: – Davidoff et al. (2005): Focused on children with chronic health conditions – Joyce & Racine (2005): Up-to-date immunization Enrollees’ perspective: – State specific AL, CO, FL, IA, KS, NH, NC, PA, TX, & NY – Separate SCHIP programs only
Methods Overview
Two sets of analyses:
– Effect of SCHIP implementation on lowincome children? (Low-income children’s perspective) Time and county fixed effects model
– Effect of public insurance program enrollment on enrolled children? (Enrollees’ perspective) Instrumental variables regressions
National Health Interview Survey
Sample Child Files 1997-2002
National data collected continuously by the NCHS since 1957 for the civilian noninstitutionalized population during in person household interviews
Comprehensive information on health insurance, access to care, health care use, and health status
SCHIP coverage collected beginning in 1999 may be misreported as Medicaid enrollment. The SCHIP sample maybe under-representative. Analysis sample: N>24,500 – Children ages 0-17 years – Family income below state SCHIP eligibility thresholds
Analysis 1:
Effects for low-income children
Staggered SCHIP implementation across states results in variation in SCHIP availability
– explanatory variable of primary interest
Multivariate regressions controlling for time and county fixed effects
Allow the effect of SCHIP to vary with time since implementation
– SCHIP available for 1-11m or ≥12m
Analysis 1:
Low-income children (cont.)
Dependent variables – Access to Care: Usual place for care, delayed care, unmet need for medical care, unmet need for drug, unmet need for mental care – Use of medical services: General doctor visit, specialty doctor visit
Control variables – Child age, gender, race, US born, general health – Family size, education, income – County & time dummies, county unemployment Estimated by linear probability regressions with adjustment of survey designs
Analysis 2:
Effects of coverage for the enrolled
Instrumental variables regressions:
– Commonly used econometric method to deal with selection bias, e.g. associated with coverage – Instruments for public coverage: SCHIP available ≥12 months or not Medicaid & SCHIP eligibility thresholds (% FPL) Waiting periods (months uninsured required) Any family member receives Food Stamp or not Insurance offer from work place or not
Estimated by two-stage-least-squares with same control variables Comparison group is uninsured counterparts
Regression Results
SCHIP≥12mo. Usual place for care Delayed care Unmet medical need Unmet drug need Unmet mental need
General doc. visit Special doc. visit
Public Coverage .174** (.006) .106** (.006) .075** (.005) .047** (.005) .014** (.003)
.158** (.008) .059**(.005)
.024 (.013) .002 (.011) .003 (.009) .009 (.008) .014 (.008)
.074** (.019) .002 (.013)
Robust standard errors in parentheses; * significant at 5%, ** significant at 1%
Implications: Sample children vs.
those with higher family income
8% 6% 4% 2% 0% -2% -4% -6% -8%
Diff._Sample
Diff._Fully insured
Place for care
Delayed care
Unmet General med. care doc. visit
Special doc. visit
Conclusions
For low-income children: – SCHIP implementation significantly increased visits to general doctors – Conservative methods imply greater effect among enrolled or eligible children
For enrolled children: – Public program enrollment due to SCHIP implementation significantly improved access to care and use of medical services compared with uninsured counterparts
Conclusions (cont.)
Different types of SCHIP programs (Medicaid expansion, separate program, combined) did not show consistent differential effects
Limitations
Due to limited information in the data, this study did not calculate eligibility for SCHIP or focus on children enrolled in SCHIP (however, in both cases the spillover effect on Medicaid enrollment was taken into account) Determined short-term effect of SCHIP, which may be different from its long-term effect
Future Research
Estimate the effect of SCHIP implementation for SCHIP eligible and SCHIP enrollees separately from its spillover effects on Medicaid eligible or Medicaid enrollees
Track SCHIP’s long-term impact
Acknowledgment
National Center for Health Statistics (NCHS)/AcademyHealth Health Policy Fellowship Program.