Essays in health economics
- This dissertation explores various topics in the economics of healthcare, specifically the behavior of physicians (supply-side) and healthcare consumers (demand-side). The first chapter is coauthored with Sarah Eichmeyer and investigates how physician opioid-prescribing behavior impacts patient outcomes and behavior. In the past two decades, death rates from opioids have seen a fivefold increase and opioid prescribing has emerged as a leading public health problem in the United States. Clinical guidelines leave many opioid prescribing decisions to physician judgement; we study the extent to which a single opioid prescription in an emergency department, for these marginal cases, can induce long-term dependence and impact health and economic outcomes of a patient. We tackle these questions by leveraging quasi-random assignment of patients to physicians, who vary in their propensity to prescribe opioids. We analyze the universe of electronic health record data for a particularly vulnerable population---veterans---and find that a single opioid prescription can have strong adverse effects on a veteran's long-term outcomes. A single opioid prescription induces a 1.2 percentage point (pp) increase in the probability of long-term prescription opioid use, a 0.34pp increase in development of an opioid use disorder, and a 0.075pp increase in opioid overdose mortality. We find suggestive evidence of both use of and death by heroin and synthetic opioids. Moreover, in settings where the supply of legal prescription opioids is restricted, veterans are more likely to resort to illicit opioids, highlighting the complex interdependencies between legal and illicit sources of opioid supply. The second chapter, also coauthored with Sarah Eichmeyer, builds on the first chapter by expanding into the primary care setting and the broader effects of having a high opioid-prescribing primary care provider (PCP). Primary care is the most frequently utilized health service and is the source for nearly half of all opioids prescribed in the United States. This chapter studies the impact of exposure to high prescribing PCP on opioid abuse, and physical and mental health among veterans. Using over two decades of electronic health records, we exploit variation in opioid prescribing tendency across providers in the same facility, in conjunction with quasi-random assignment of providers to new patients. We find that assignment to a PCP who prescribes opioids at a 3pp higher rate (equivalent to the difference between a 90th and 10th percentile prescriber within a facility) is associated with an increase in the probability of long-term opioid use by 0.72pp, development of an opioid use disorder by 0.12pp, and five-year opioid overdose mortality by 0.008pp. Veterans' mental health deteriorates; the three-year likelihood of attempted suicide or self-harm increases by 0.023pp and depression diagnosis increases by 0.18pp. Investigating into the mechanisms, we find evidence consistent with high opioid prescribers being less likely to refer patients to alternative pain management, adhere to clinical recommendations on naloxone distribution, or refer patients to substance use disorder treatment. The final chapter is coauthored with Audrey Guo and studies how healthcare consumers dynamically optimize their medical utilization in response to nonlinear dynamic healthcare prices---a fundamental question in health insurance markets. This chapter tests the neoclassical prediction that a fully forward-looking agent only responds to their expected end-of-year price. Our unique identification strategy studies families during the year of childbirth who will likely satisfy their annual deductible, thereby knowing their expected end-of-year price. We find that during the year of a childbirth, fathers increase medical spending by 11% per month after their deductible is satisfied, rejecting the null of fully forward-looking consumers. This behavior cannot be explained by fathers increasing utilization in response to the childbirth itself. Furthermore, this myopia translates to a 21-24% decrease in total annual medical spending, relative to the counterfactual of fully forward-looking behavior, and is concentrated in elective procedures; we find no response in low value or urgent care. Our findings suggest the need for modeling non-linear incentives while accounting for myopic behavior when studying the medical utilization responses to health insurance
|Type of resource
|electronic resource; remote; computer; online resource
|1 online resource
|Zhang, Jonathan Xia
|Chan, David C, Jr
|Degree committee member
|Chan, David C, Jr
|Stanford University, Department of Economics.
|Statement of responsibility
|Submitted to the Department of Economics
|Thesis Ph.D. Stanford University 2020
- © 2020 by Jonathan Xia Zhang
- This work is licensed under a Creative Commons Attribution Non Commercial 3.0 Unported license (CC BY-NC).
Also listed in
Loading usage metrics...