Does Cost-Effectiveness Analysis Have a Role in US Managed Care Drug Formularies? An Empirical Study of Utilization, Costs, Outcomes, and Elasticity of Demand of a Value-Based Formulary

Thesis (Ph.D.)--University of Washington, 2015

The standard economic model for health insurance posits that in order to account for moral hazard in a population for which there is varying marginal benefit of treatment that is unknown to the insurer, cost-sharing should be proportional to the price elasticity of demand. Yet, many have observed that when faced with cost-sharing, consumers may reduce utilization to suboptimal levels due to underestimation of the marginal benefit of treatment. Hence, optimal health insurance design requires consideration of both insurer and consumer information asymmetries regarding the marginal benefits of treatment. This dissertation investigates whether cost-effectiveness analysis (CEA) may be useful for optimizing insurance in the face of insurer and consumer information asymmetries. In 2010, Premera Blue Cross, a large non-profit health plan in the Pacific Northwest implemented a value based formulary (VBF) benefit among their own employees and dependents that explicitly uses CEA to inform medication placement within copayment tiers. We exploit this natural experiment to empirically assess the impact of the VBF on medication utilization and other health services utilization and the impact of the VBF on medication costs and non-medication costs from the member, health plan, and overall perspectives. We also estimate price elasticities of demand for pharmaceuticals overall, by therapeutic class, by brand-generic status, and finally by the copayment tiers informed by CEA. In the first paper, we use individual-level data from July 2006 to June 2013 drawn from the employees of Premera and their dependents as well as data from employees and dependents of 5 employer sponsored plans administrated by Premera and chosen based on similarity to the intervention group in industry classification. After controlling for member demographics and plan characteristics and secular trends using an interrupted time series design with concurrent control, we find that the VBF shifted member medication utilization towards drugs placed in lower copayment tiers. The VBF also was associated with increased member medication costs and decreased health plan medication costs, leading to a net medication savings of $8 per member per month (PMPM) (95% confidence interval [CI], -$15, -$2). Over the 3 year period of the study, the medication cost savings totaled over $1.1 million USD. The findings regarding non-medication costs were comparatively small and not statistically significant. Total costs decreased by $9 PMPM (95% CI, -$49, $30) but was not statistically significant. We did not detect any changes in the probability or the number of emergency department visits, hospitalizations, or office visits. This evaluation suggests that the VBF may have reduced overall medication costs without negatively impacting utilization of other health services – a proxy for adverse outcomes. In the second paper, we use data from July 2009 to June 2011 drawn from the employees of Premera and their dependents to construct a medication level dataset of 284 unique medications. These medications accounted for 79.3% of the prescription medication volume over the period of observation. After controlling for member demographics and using a pre-post design, we find that our elasticity estimates of -0.14 for the probability of filling a medication were similar to the overall elasticity estimates of -0.17 from the RAND Health Insurance Experiment. We also find that the estimates by therapeutic class and brand-generic status also were generally similar to published studies. Finally, we estimate of price elasticity of demand by copayment tiers informed by CEA. We found that elasticity estimates for the probability of fill and days’ supply of medication respectively were -0.07 and -0.06 for the preventive tier, -0.09 and 0.08 for tier 1, -0.26 and -0.26 for tier 2, -0.27 and -0.32 for tier 3, and -0.45 and -0.55 for tier 4. Thus we observed a general trend of increasing elasticity with increasing copayment tiers. These results suggest that a cost sharing strategy based on elasticity estimates may be similar to a cost sharing strategy informed by CEA. Furthermore, since in the first paper we observed that the VBF was associated with decreased medication costs without negatively impacting health services utilization, we suggest that the use of CEA to inform medication copayment tiers may have a role in optimizing insurance benefit design. Contemporaneous with the implementation of the VBF on July 2010, the US health insurance market has experienced many changes. The passage of the Patient Protection and Affordable Care Act (PPACA) in March 2010 and the progressive implementation of many provisions in that law have fundamentally altered the insurance and provider markets. One provision in PPACA explicitly supports the development of value-based insurance design. This provision has led to the promulgation of federal regulations that have focused on eliminating cost-sharing for certain preventive services. The PPACA also has mandated coverage of preventive services recommended by Advisory Committee on Immunization Practices and the US Preventive Services Task Force. Due to the direct influence these national advisory bodies now have on the coverage of preventive services, others have advocated use of CEA by these bodies to inform the relative costs and benefits of their decisions. Our research suggests that the application of CEA to explicitly inform cost-sharing may allow the expansion of value-based principles beyond waivers of cost-sharing for specific “high value” services. Another important provision of PPACA is the formation of Accountable Care Organizations (ACOs) by health care providers that are accountable for the quality, cost, and overall care of Medicare beneficiaries. Under this provision, the ACOs enter into payment contracts that share both financial risk and savings for the care of beneficiaries. This shift in financial risk may cause ACOs to become more aware of the cost of health care. Further, since providers can greatly influence medications utilization behavior, ACOs may become an important lever for the introduction of value-based principles in integrated delivery system models. Finally, PPACA and the Health Information Technology for Economic and Clinical Health Act of 2009 have financially incentivized the use of electronic health records (EHR). It is possible that future iterations of VBFs could utilize patient-level information from EHRs to more accurately triangulate levels of cost-sharing with better estimates of treatment benefit. Further, the adoption of electronic health records may allow for future evaluations of VBFs that are able to assess true health outcomes. In sum, changes in the health insurance market present many opportunities and challenges for the adoption and evaluation of value-based principles in health insurance design. Policymakers and researchers should carefully assess such dynamics when considering the role of value-based insurance in health insurance markets in the future. Footnote: A joint regulation issued by the Department of Health and Human Services, the Internal Revenue Service and the Department of Labor, regarding the elimination copayments for certain preventive services states “The Departments recognize the important role that value-based insurance design can play in promoting the use of appropriate preventive services.”