COST-EFFECTIVENESS OF MEDICARE PART D PLANS IN SCHIZOPHRENIA

Purpose: The Medicare Part D prescription drug benefit has a coverage gap, imposing risks for discontinuing medications, particularly in mental health disorders where drug costs are high. However, some plans offer partial gap coverage and some beneficiaries get full coverage through subsidies. It is unclear how different benefits affect health outcomes in schizophrenia.  

Methods: We constructed a Markov model using all 2007 Medicare beneficiaries with schizophrenia enrolled in stand-alone Part D plans. We estimated the cost-effectiveness of three types of plans: 1) no coverage in coverage gaps, 2) generic drug coverage only, or 3) full drug coverage in low-income beneficiaries, termed low-income subsidy (LIS) plans. To control for differences in each plan’s beneficiaries, outcomes (hospitalization, mortality, and costs) were adjusted for age, sex, race, and comorbidity. We obtained quality of life utilities from the literature and took a societal perspective. The model, during its 1-year time horizon, assumed that all beneficiaries were on drug therapy and no plan switching occurred. In the base case, we assumed no differences in outpatient treatment utility (base case: 0.81) between plans, which we examined further in sensitivity analyses.

Results: In aged and disabled patients with schizophrenia, no significant differences in mortality between plans were seen. As shown below, yearly hospitalization rates were lower and average yearly drug costs were higher in LIS after adjustment; total medical costs were similar between plans.  

Status	Coverage	Hospitalization Rate	Drug costs	Total costs
Aged	No Coverage	52.9%	$5,344	$31,354
	Generic-only	50.2%	$5,514	$30,038
	LIS	44.7%	$8,308	$31,239
Disabled	No Coverage	39.9%	$6,452	$23,653
	Generic-only	37.3%	$6,622	$22,336
	LIS	32.3%	$9,416	$23,537

In the cost-effectiveness analysis, generic-only coverage cost less and was more effective than no coverage. Compared to generic-only, LIS cost >$850,000/QALY. However, effectiveness differences between plans were small; if utility in the LIS plan improved (from decreased anxiety about coverage and/or better health) due to more generous coverage, LIS cost <$100,000/QALY if its associated utility was >0.01 higher than the utility of other plans.

Conclusion: In our model, generic drug coverage was cost-saving compared to no coverage in schizophrenia while improving health outcomes. Drug costs for LIS plans were substantially more than other plans, but LIS’s more generous benefits may be economically reasonable if they result in improved health utility.