Presentation on theme: "Reduction of Medicaid Expenditures from State Prescription Programs in Illinois and Wisconsin Donald S Shepard, PhD* Desiree Koh, * Cindy Thomas, PhD*"— Presentation transcript:
Reduction of Medicaid Expenditures from State Prescription Programs in Illinois and Wisconsin Donald S Shepard, PhD* Desiree Koh, * Cindy Thomas, PhD* Grant Ritter, PhD* Daniel Gilden,+ William Stason, MD,MS* Christine Bishop, PhD* *Brandeis University; +JEN Associates Supported by the Centers for Medicare & Medicaid Services under Contract No. CMS /T.O. #2 to Brandeis University AcademyHealth Annual Research Meeting, June 8-10, 2008
22 Framework Prescription coverage Better use of drugs and medical services Maintain health Lower nursing home use Less Medicaid entry
33 Past research l Rector (2004), Safran (2005), Leung (2005) – About 30% of low income people skip some prescribed medications l Soumerai et al. (1991) –Limiting drugs to vulnerable population increased nursing home admissions l Gilman (2004) – Members of Prescription Assistance Programs (PAP) skip fewer doses than comparable controls l Shepard (2006) – SeniorCare halved risk of skimping l Leung (2005) – Risk is related to individual characteristics
44 Program background l In mid-2002, Illinois and Wisconsin initiated “SeniorCare” (SC) pharmacy assistance programs (PAPs) that provide low-income persons aged 65+ with publicly funded prescription drug assistance. l Maximum co-payments per prescription are generally $4 in IL and $15 in WI. l Enrollees with incomes up to 200% of the federal poverty limit (FPL) are funded under a Medicaid waiver designed to help seniors improve prescription drug use, maintain health and reduce financial vulnerability due to prescription costs.
55 Three strata studied l 68,292 Wisconsin members, who were all new enrollees (1,189 interviewed), l 121,000 Illinois members previously in Circuit Breaker, a limited PAP that excluded mental health and gastro-intestinal drugs and automatically rolled over into SC (termed ‘IL rollovers, 374 interviewed); l 47,782 Illinois members not previously in this PAP (termed ‘IL new,’ 664 interviewed).
66 Skimping by Income Group
77 Study Design for Medicaid Analysis l Ohio served as the comparison state. l Using Medicare claims and zip codes, matched Illinois and Wisconsin enrollees exactly on demographic and disease categories to similar Ohio Medicare beneficiaries. l Used propensity scores to match closely on disease severity and socio-economic characteristics based on census information and Social Security benefits.
88 Population Studied l Needed precise matching on income for examining Medicaid entry l Limited this analysis to buy-in Medicare beneficiaries in the three states Received subsidies for Medicare premiums and deductibles Qualified Medicare Beneficiaries, QMB Special Low Income Medicare Beneficiaries, SLMB l We matched 7,699 Illinois and 1,798 Wisconsin buy-in beneficiaries to comparable buy-in Ohio controls.
99 Research Objective: Evaluate First Year Impacts on l Nursing home entry l Medicaid entry l Medicaid expenditures
10 Approach l Descriptive analysis l Multivariate analysis
11 Nursing Home Entry among Wisconsin Buy-in Beneficiaries and Matched Ohio Controls
12 Medicaid Entry among Wisconsin Buy-in Beneficiaries and Matched Ohio Controls
13 Hazard Function for Nursing Home Entry, Wisconsin, part 1 VariableParameter Estimate Standard Error Statistical Signifi- cance Hazard Ratio Inpatient 0-3 Months of Index Home Health 0-3 Months of Index SNF 0-3 Months of Index JAI Morbidity Score Indicator for a Arthritis diagnosis Indicator for a Chronic heart disease diagnosis Indicator for a Congestive heart failure diagnosis Indicator for a COPD diagnosis Indicator for a Cerebrovascular disease diagnosis Indicator for a Diabetes diagnosis
14 Hazard Function for Nursing Home Entry, Wisconsin, part 2 VariableParameter Estimate Standard Error Statistical Signifi- cance Hazard Ratio SSA Dept Count=1; SSA Pym (in 1,000s) SSA Dept Count> SSA Dep Count>1 * SSA Pymt (in 1,000s) % Census Block: Income $0-$10, % Census Block: Income $10,000-$20, % Census Block: Income $20,000-$30, % Census Block: Income $30,000-$40, % Census Block: Income >$40, % Census Block: HMO Participant State Rx Enrollee
15 Adjusted nursing home entry (SC/OH)
16 Crude risk ratio for Medicaid entry (SC/OH)
17 Relative spending per entrant (SC/OH)
18 Relative spending per enrollee (SC/OH)
19 Illinois summary l Due to preexisting PAP, SeniorCare did not reduce Medicaid entry, but did reduce nursing home entry and spending. l Cumulative rate of nursing home entry of Illinois SeniorCare buy-in beneficiaries (2.4%) was half the rate of the matched Ohio controls (4.4%). l Medicaid spending over the first year when averaged over all Illinois buy-in SeniorCare members (with standard errors of the mean) was $631 ($26) vs. $1,605 ($83) for matched buy-in Ohio controls l Per enrollee savings $974 ($87) or 61%. l Savings in Illinois did not quite equal the state’s share of first-year program costs per enrollee year ($1,394).
20 Wisconsin summary l SeniorCare buy-in enrollees had half the rate Medicaid entry in the first year (11%) than matched Ohio controls (22%) l Wisconsin SC had half the rate of nursing home entry (2.2%) compared to Ohio controls (4.5%) l Had $1,190 ($163) or 81% lower Medicaid spending per buy-in enrollee. l Wisconsin savings on buy-ins were greater than the state’s share of first-year program costs per enrollee year ($1,032).
21 Extrapolation possible? l Question: Do the data allow examining impacts on nursing home and Medicaid for all SeniorCare enrollees? l Answer: No l Why not? Ascertainment of income
22 Medicaid entry by family income
23 Approximate probability of Medicaid entry
24 Household incomes by neighborhood
25 Neighborhood: weak predictor of family income
26 Unsuccessful extrapolation beyond buy-ins
27 Conclusions l 50% reductions in skimping applied to all SeniorCare enrollees l Comparable declines in nursing home entry among buy-ins. l First year savings in buy-in population not quite enough to pay for the program costs in Illinois l These savings were more than sufficient in Wisconsin. l Prescription drug coverage for vulnerable populations pays off with less nursing home entry and lower costs.
28 Limitations l Differences in nursing home and Medicaid policies among states could confound interpretation
29 Strength: Consistent improvements in l Self reported behavior (skimping) l Costly services (nursing home entry) l Medicaid expenditures
30 Research implications l Observations and natural experiments very powerful. l Must understand and control for selection effects. l Stay within the data.
31 Policy implications l Enrollment of needy elders in both states benefited from outreach, straightforward design, and federal subsidies that extended to 200% of the FPL. l These findings show the value of completing “coverage” with access to prescription drugs.