The Effects of State Medicaid Expansions for Working-Age Adults on - PowerPoint PPT Presentation

The Effects of State Medicaid Expansions for Working-Age Adults on Senior Medicare Beneficiaries Melissa McInerney, Jennifer Mellor, and Lindsay Sabik

Research Questions • Do Medicaid expansions to working-age adults affect healthcare spending and utilization among older Medicare beneficiaries? – A one percentage point increase in the share of adults eligible for Medicaid reduces annual healthcare spending by $87 per Medicare beneficiary – Larger reductions of $477 per person among dual eligibles • If there are spillovers, are there adverse health effects? – Self-reported health does not decline – Mortality does not increase

Contributions • Develop testable predictions about spillover effects • Microdata from the Medicare Current Beneficiary Survey (MCBS) Cost and Use files – Identify duals and non-duals – Test for heterogeneous effects by state Medicaid generosity – Examine impact on health status • Examine all Medicaid expansions to working-age adults in the 2000s

Why might there be spillovers? • Medicaid expansions to parents increase Medicaid participation and overall insurance coverage – Kronick and Gilmer, 2002; Aizer and Grogger, 2004; Busch and Duchovny, 2005; Hamersma and Kim, 2013; McMorrow et al., forthcoming • Adults newly covered by Medicaid use more healthcare – DeLeire et al., 2013; Finkelstein et al., 2012; Taubman et al., 2014; Long, 2008; Long and Masi, 2009; Long and Stockley, 2011; Miller, 2012; Long et al., 2012; Long et al., 2012/2013; Sommers et al., 2012 • Estimates of crowd-out associated with adult Medicaid expansions more modest than CHIP expansions to children; most expansions to adults extend eligibility to persons at lower levels of eligibility than expansions to children – Buchmueller et al., 2015; Atherly et al., 2016

Prior Literature on Spillovers to Medicare Enrollees • Evidence of no effect or negative spillovers following Massachusetts health care reform – Joynt et al., 2015; Joynt et al., 2013; Bond and White • Seniors in NY, ME, and AZ experienced increased access to care as well as declines in cost-related care delays and mortality following Medicaid expansions – Sommers et al., 2012 • Prior study to examine expansions in whole US finds negative spillovers following substantial CHIP expansions – Glied, 2014

Conceptual Framework • Sloan, Mitchell, and Cromwell (1978) mixed-economy model – Physicians treat patients covered by the government or private payer. – Assumption: Medicaid reimburses at lower rate than Medicare. • Crowd-out: Medicaid expansions increase supply of services to Medicare enrollees • No crowd-out: Medicaid expansions have no effect on Medicare enrollees • Glied and Zivin (2002) models of physician behavior with FFS and managed care – FFS insurance generous (Medicare), managed care restrictive (Medicaid) • Only take restrictive coverage patients if excess capacity: no effect • If induce demand among patients with more generous insurance: increase care • Choose fixed costs of practice based on patient insurance profiles: reduce care – Effects larger if doctor treats Medicaid patients (i.e., duals) – Effects larger in states where Medicaid provider payments less generous

Dual Eligibles • Low-income seniors – Eligible for Medicare because of their age – Eligible for Medicaid because of their low income • Physicians who treat duals are reimbursed by Medicare • Medicaid covers cost sharing for the patient • Physicians who treat dual eligibles are roughly twice as likely to treat working-age Medicaid patients

Data • Medicare Current Beneficiary Survey Cost and Use files – 2001 through 2009 – Age 65 and older – Enrolled in Medicare Parts A and B for entire year – Live in community for entire year – 71,709 individuals – Separately identifies duals and non-duals – Mean medical spending: $11,400 per year • Simulated eligibility measure of Medicaid expansions to working-age adults – Percent of state population aged 20-64 eligible for Medicaid in year t-1 • Currie and Gruber, 1996; Gruber and Simon, 2008 – Medicaid income eligibility rules for both parents and childless adults • Measure of Medicaid generosity – 2003 Medicaid-to-Medicare fee ratio for primary care (Zuckerman et al., 2003)

Two-Part Models of Healthcare Spending P ( y ist > 0| SimElig s , t - 1 , X ist ) = F ( b 1 SimElig s , t - 1 +G X ist ) ln( E ( y ist | SimElig s , t - 1 , X ist , y ist > 0)) = a 1 SimElig s , t - 1 +P X ist E ( y ist | SimElig s , t - 1 , X ist ) = P ( y ist > 0| SimElig s , t - 1 , X ist )* E ( y ist | SimElig s , t - 1 , X ist , y ist > 0) X: age, age 2 , sex, race/ethnicity, household income, household income 2 , household size, educational attainment, marital status, veteran status, reside in urban area number of chronic conditions, smoking, BMI state unemployment rate, state percent Medicaid enrollees in comprehensive managed care plans, hospital wage index (total spending, inpatient and outpatient), physician practice cost indices (total spending, medical provider events) state fixed effects, year fixed effects, state-specific linear time trends

Effects of Eligibility Expansions on Healthcare Spending by Type of Insurance, Marginal Effect from TPM of SimElig(t-1) Full Sample Dual Eligibles Non-Duals Total Payments -86.9** -476.6*** -38.9 (34.2) (82.8) (40.6) Inpatient 7.3 -229.7*** 32.6*** Payments (14.2) (57.9) (12.4) Medical Provider -53.0** -94.2*** -47.3 Payments (25.4) (35.6) (29.5) Outpatient -21.6*** -52.1*** -18.1** Payments (8.0) (12.9) (8.5) Prescribed Med. -9.0*** -1.7 -9.0*** Payments (2.8) (9.4) (3.3)

Are Duals Receiving Less Care, or Less Costly Care? Coefficients and Marginal Effects on SimElig(t-1) , Duals Probit GLM (Marginal Marginal Effect Coefficient Effect) of SimElig(t-1) Total Payments 0.03 -496.4*** -476.6*** (0.01) (84.2) (82.8) Inpatient -0.01*** -620.5*** -229.7*** Payments (0.006) (197.8) (57.9) Medical Provider 0.03** -102.8*** -94.2*** Payments (0.01) (36.8) (35.6) Outpatient -0.001 -68.9*** -52.1*** Payments (0.007) (16.5) (12.9) Prescribed Med. 0.01 -4.3 -1.7 Payments (0.01) (10.1) (9.4)

Are negative spillovers larger in states with less generous Medicaid? Marginal Effects from TPM, Duals Total Inpatient Medical Outpatient Prescribed Payments $ Provider $ Hospital $ Medicine $ SimElig(t-1) x -684.2*** -190.6** -186.0*** -117.3*** 21.9 state (154.9) (92.2) (38.2) (38.6) (30.5) generosity below 10 th ptile SimElig(t-1) x -404.0*** -291.8*** -29.4 -31.6* -9.8 state (135.8) (74.0) (39.2) (17.3) (9.6) generosity at or above 10 th ptile p-value, test of 0.20 0.39 0.005 0.06 0.32 equality

Additional analyses • No strong evidence that capacity constraints is the primary mechanism • Effects on spending fall over time • No evidence of sample selection/healthier dual population – No increase in mortality in year t – No evidence of “welcome mat” effect for healthier duals • No evidence that spending declines due to rising Medicare Advantage enrollment • No evidence of pre-existing trend driving results

Effects of State Medicaid Expansions to Working-Age Adults on Senior Health Outcomes, Results from LPM, Duals =1 if =1 if Health =1 if Health =1 if Die in General Limits is Worse Year t+1 Health is Social Compared Fair or Poor Activities to One Year Ago SimElig(t-1) -0.0001 -0.0007* -0.0001 -0.0002 (0.0005) (0.0004) (0.0006) (0.0025) Mean of 0.209 0.123 0.214 0.072 dep. Var.

Effects on health? Additional evidence • No evidence Medicaid expansions alter dual eligibles’ utilization of preventive care • No evidence that Medicaid expansions increase likelihood a dual in FFS has an avoidable hospitalization – Some evidence that increases in eligibility reduce likelihood of hospitalization for diabetes-related complications among diabetics and hospitalization for angina among at-risk patients • No long-term effects on health

Discussion • Medicaid expansions to working-age adults have negative spillover effects on spending and utilization among the already-insured • Leading explanation for results: physician responds to heterogeneous payment environment by changing fixed practice components • No evidence of reduced welfare corresponding with these spending reductions – May be reduction in so- called “overtreatment” • Large expansions from 2001-2009 similar magnitude to ACA expansions – Average increase in states with large expansion in our study period: 12.6 ppts – Average increase among states expanding Medicaid in 2014 (but did not have a prior large expansion to non-elderly adults) 5 to 15 percentage points

Thank you! melissa.mcinerney@tufts.edu