Research Overview

My commitment to equity drives my research interest in Medicaid policy, access to care, and health equity. I conduct theoretically and conceptually grounded empirical research using advanced quantitative methods, as well as methodological work on data quality.

Previous and Current Work

My prior work examined the effects of Medicaid expansion on health insurance coverage and low-acuity ED utilization. For example, in one dissertation paper, I find that Medicaid expansion increased low-acuity ED visits in three states, particularly among men and younger beneficiaries, using interrupted times series. In another dissertation paper, I use the economic theory of consumer choice and non-linear least squares method to empirically test the common assumption that EDs are used as a substitute for unavailable primary care. Examining enrollees of Medicaid managed care in New York, I find that low-acuity ED visits and primary care (PC) are complements during nights and weekends in poorer and urban counties and when primary care is provided by advanced practice providers, suggesting referrals from primary care to the ED. Through this research, I honed my understanding of how Medicaid beneficiaries use the ED as a safety net.

The adage “garbage in, garbage out” captures the reality that our research is only as good as our data. An important part of my research focuses on data quality. I have explored the quality of mortality and hospital discharge data, leading to publications in Addiction and Health Services Insights. My current work as a postdoctoral fellow at Harvard Medical School combines my interests in Medicaid managed care and data quality. This work is mentored by Dr. Laura Hatfield, who leads the Data and Methods Core of an NIA-funded program project. The Transformed Medicaid Statistical Information System (T-MSIS) Analytic Files (TAF) have the potential to enable new research on Medicaid managed care, but their quality is untested. In the main project, I develop methods to address silent missingness in outpatient ED visits in TAF. Silent missingness refers to encounters that are missing without any indication of missingness. I develop a conceptual model of the data generating process in which managed care organizations (MCOs) affect both ED use (e.g., via utilization management) and silent missingness (e.g., via their own data validation processes). I then develop two approaches to estimating the true number of ED visits. In a simulation study, I show that both approaches outperform the naïve approaches of using TAF data as is or limiting analyses to states with high-quality data. I plan to implement the two methods in an application to TAF data. In another ongoing project, my colleagues and I examine the consequences of CMS’s truncation of diagnosis codes in TAF outpatient records to only two. These data quality projects complement my substantive interest in Medicaid policy. As managed care penetration grows in both Medicare and Medicaid, health services researchers can no longer afford to limit analyses to fee-for-service enrollees. My work tackles the quality of managed care encounter data head-on, rather than simply continue to exclude it.

Future Work

Medicaid is intertwined with the future of racial justice in the United States. Managed care organizations sit at the intersection of state Medicaid agencies, health care providers, and Medicaid beneficiaries. Although this positions them to improve health equity, their incentives are not necessarily aligned with those of state Medicaid agencies. Even when states commit to health equity, their policies must translate into incentives for managed care organizations (and some states are far from making such a commitment). My long-term goal is to understand how managed care advances or erodes health equity.

Please feel free to contact me to chat about my work and/or potential collaborations.

I have used a variety of advanced quantitative methods to conduct my research: for example, quasi-experimental methods, maximum likelihood, non-linear least squares, machine learning, and simulation analysis. I also have experience developing a conceptual model and designing a discrete choice experiment through collaborative work. Please see below for some examples of published work. Working papers are available upon request.