This dissertation studies health care policies aimed at better measuring and improving quality of life. The policies were implemented at the Veterans Health Administration, the largest integrated health care system in the U.S. My aims are to understand whether the policies reduce despair at the individual level, and to explore implications for disparities at the population level. Hence the title of this dissertation, Despair and Disparities.
In Chapter 1, I estimate the health costs of untreated depression by studying the impacts of a large screening initiative. In 2008, over one thousand clinics across the VA began automatically prompting physicians to screen patients with a standardized two-question survey, the PHQ-2. This survey includes questions about the frequency of two emotions highly associated with clinical depression over the prior two weeks; the resulting score is a widely adopted measure of depression severity. I use this novel source of data and the sudden nature of the rollout to establish two facts. First, screening positive on the PHQ-2 is strongly associated with one-year mortality. The association persists after controlling for a rich set of measures of physical health, including chronic conditions, vital signs on the day of the visit, and laboratory test results in the year prior to the visit, suggesting that the PHQ-2 contains information about mortality beyond what is captured by standard measures of an individual’s health status. Second, I leverage the sudden rollout of the PHQ-2, and the fact that it induced larger increases in detection depression for some groups than others, to estimate the causal impact of screening on mortality. I estimate that a 1 p.p. increase in detection of depression causes a 0.08 p.p. reduction in one-year mortality. I then use this estimate to simulate the impacts of a national screening program and conclude that the gains in population health would be substantial. Of note, because depression disproportionately affects low-income individuals, such a program would likely reduce cross-income group disparities in mental and physical health.
In Chapter 2, I study the impacts of a program that embedded mental health professionals in primary care clinics, thus exposing millions of patients to more proximate mental health care. To estimate causal effects of the program, Primary Care Mental Health Integration (PC-MHI), I leverage staggered adoption across primary care clinics and estimate that the program caused a 2.4\% reduction in mortality. Because PC-MHI increased access to psychotherapy and psychiatric medications without increasing measures of non-psychiatric care, these health benefits likely flow through improved access to mental health care. Compared to the average patient in the sample, patients induced into treatment by PC-MHI were lower-income, had lower levels of social capital, and suffered from drug use disorders and liver disease at higher rates. Given the close resemblance of these patients to the population suffering deaths of despair, I conclude that increased access to psychiatric treatment could help remedy the devastating consequences of the deaths of despair epidemic.
In Chapter 3, I study the impacts of palliative care, a rapidly expanding medical subspecialty aimed at monitoring and improving quality of life. When patients are near the end of life, doctors often prescribe treatments, such as chemotherapy or mechanical ventilation, in hopes of extending life. These treatments are often costly to both the individual and society, could reduce quality of life, and may have harmful side effects that could, paradoxically, reduce survival. Leveraging staggered adoption across hospitals, I find that palliative care causes a 12\% reduction in cost without affecting survival. This suggests that marginal spending near the end of life is flat of the curve. It also suggests that if a physician attuned to quality of life is involved in the patient’s care, outcomes can help society without harming the patient.
Collectively, these chapters contribute to the body of knowledge in economics and medicine in three primary ways. First, they address a measurement problem: existing research on quality of life is sparse in part because of a paucity of reliable data sources. This dissertation includes several pieces of data that are not traditionally used in economics research, including PHQ-2 scores, pain scores, and vital signs. My hope is that this opens doors for more research on quality of life, including methodological progress that brings cardinal utility back into the welfare discussion.
These chapters also contribute to existing work by using observational data to document the causal impacts of large policy initiatives in the medical system. Reliably estimating causal effects is critically important for policymaking. In the medical field, unlike economics, a large share of evidence comes from randomized controlled trials. Though such trials are internally valid, they are typically small, raising concerns about external validity. By using quasiexperimental research designs to document causal impacts of policies affecting millions of patients, these chapters reap some benefits of both internal and external validity, and can thereby inform policy efforts aimed at improving population health in a rigorous manner.
Finally, these chapters study a unique set of policies: those drawing attention to quality of life. Today's health care system is increasingly centered around technology. The value of care provided by hospitals and providers is increasingly monitored, and physicians are spending less time at the bedside. In this economy governed by efficiency, what often gets pushed aside is subjective well-being: that which takes time to understand, defies simple classification, is unable to be neatly tabulated by the hospital’s billing department.
But what about the economy of feelings, the value of being heard? By studying policies aimed at understanding and improving quality of life, and showing that they produce benefits both to individuals and society, these chapters call for a health care system that listens more closely and cares more compassionately.
