The economics of safety-net programs: evidence from the 340B Drug Pricing Program

Abstract

This dissertation consists of four chapters. In the first chapter, I present research from a research agenda that I have pursued alongside collaborators in the School of Public Health (Sayeh Nikpay, Claire McGlave, and Elizabeth Watts) studying the growth of the 340B Drug Pricing Program, a safety-net program that is responsible for nearly 10% of annual pharmaceutical spending in the United States. The two articles presented in this chapter are adapted from versions that have been published in JAMA Health Forum and Health Affairs Scholar. These two articles describe the rapid growth in the 340B Drug Pricing program, particularly among chain pharmacies. My coauthors and I document that nearly three quarters of big chain pharmacies participate in the 340B program.In the second chapter, I study whether vertical integration can increase participation in social safety net programs. In particular, I study several examples of integration between pharmacies and patient or inventory management softwares, aimed at assisting hospitals manage their participation in the 340B program. In recent years, several large pharmacy chains (Walgreens, CVS, Optum, and Accredo) have integrated with software that processes claims and manages inventory - and several are integrated with insurers as well. These pharmacy chains bundle contract pharmacy and 340B administrative services, which I hypothesize lowers the transaction costs of participating in this social safety net program. Simultaneously, integration with an insurer may increase revenues from the 340B program. I study the effect of such integration on the growth of this social safety-net program, and find that vertical integration had a significant impact on participation in this program, leading to increased contract activity after the integration, without a reduction in contracts with competitors. This provides evidence that the integration broadens the scope of the program, widening the net with which a hospital is able to capture discounts. The third chapter is coauthored with Colleen Carey and Sayeh Nikpay. We study the impact of recent litigation on the ability of hospitals to claim discounts under the 340B program. The discounts are generated when participants, called “covered entities,” dispense significantly discounted outpatient drugs to insured patients. Until recently, a prescription was eligible for discounts when written by a prescriber affiliated with the covered entity and filled at a pharmacy contracting with the covered entity. A recent lawsuit - Genesis Healthcare Inc., v. Xavier Becerra - expanded the pathways for 340B discount eligibility, thereby making an unknown number of prescriptions eligible for 340B discounts. We simulate the impact of the new 340B eligibility definition using 20% Medicare Part D data linked to 340B program data and inpatient and outpatient fee-for-service Medicare claims. Comparing pre and post-“Genesis” definitions, we find that the share of Medicare Part D prescriptions that are eligible for 340B discounts would increase from 12% to 16% and result in an additional $8.2B in Medicare Part D spending becoming eligible for 340B discounts. We also demonstrate that the new definition could double the number of 340B eligible prescriptions for which multiple covered entities can claim discounts. In the fourth and final chapter, I study administrative and hassle costs (“leaks”) in the operation of public welfare program, and how private companies can both present efficiencies due to their scale while simultaneously distorting outcomes away from the optimal scenario. To study this question, I develop a model of hospital pharmacy contracting in the 340B program in which, if a hospital contracts with a given retail pharmacy, the hospital receives 340B revenue from patients that fill a prescription at that pharmacy. In each period, hospitals choose a set of pharmacies with which to contract, collect revenues, and pay fixed and sunk costs that depend in part on whether a pharmacy is a chain or independent pharmacy. This sunk cost parameter represents any reduction in administrative or “hassle” costs that would lead a covered entity to contract with a chain pharmacy, rather than an independent pharmacy that may generate more revenue. In order to address the massive computational costs and combinatorics inherent in such a model, I apply a moment inequality approach to estimation. My results demonstrate that while a reduction in administrative costs may help grow the scale of a welfare program, it simultaneously presents distortions relative to the optimal configuration. I find that a hospital is willing to contract with a pharmacy that has fewer patients, if that pharmacy has network effects. However, while this encourages an increased number of contracts, I find that the resulting configuration of pharmacies is not optimal. A counterfactual exercise in which I remove these effects demonstrates that the observed configuration results in 40% fewer captured patients, and thus revenue (a 7 percentage point decrease).