Research

WORKING PAPERS:

Association of Extended Postpartum Medicaid Coverage during COVID-19 with Postpartum Hospitalizations. Revise and resubmit. Coauthored with Maria Steenlad and Erica Eliason.

Do Urgent Care Centers Decrease Spending and Increase Access to Care? Link to thesis version.

Financial Strain Associated with Out-of-Network and Surprise Bills. Coauthored with Adam Biener, Chandler McClellan, and Samuel Zuvekas.

Measuring Hospital Utilization During the COVID-19 Pandemic using NHSN and HCUP. Coauthored with Alison Binder, Andrea Cool, Margaret Dudek, and Pamela Owens.

Healthcare Spending and Insurance Coverage by Income. Coauthored with Betsy Cliff and Ruochen Wang.

PUBLISHED PAPERS:

Catholic Hospital Affiliation and Postpartum Contraceptive Care, (2024) JAMA Internal Medicine. Coauthored with Jessica Monnet.

We examined whether postpartum outcomes among patients with a delivery were associated with the Catholic affiliation of their nearest hospital. We used coarsened exact matching to analyze rates of postpartum contraception and subsequent deliveries for  4, 101 ,443 deliveries in 11 states. Living in a zip code in which the closest hospital was Catholic was associated with a 0.95–percentage point decrease in the probability of surgical sterilization at delivery, a 0.21–percentage point decrease in the probability of surgical sterilization in the year after discharge, and a 0.47–percentage point increase in the probability of subsequent delivery within 3 years. Our finding suggest that living in a zip code in which the closest hospital was Catholic was associated with a modest decrease in the probability of postpartum surgical sterilizations and a modest increase in the probability of subsequent deliveries.

Heterogeneous Effects of the Affordable Care Act on Emergency Department Visits and Payer Composition among Older Adults by Race and Ethnicity, (2024) American Journal of Health Economics.  Coauthored with Asako Moriya and Yaa Akosa Antwi.  

We estimate the impact of the Affordable Care Act (ACA) on emergency department (ED) visits and the composition of insurance coverage for White, Black, and Hispanic older adults. Our estimation strategy uses changes in the discontinuity of insurance coverage at age 65 and variation in state decisions to expand Medicaid under the ACA. We find that uninsured ED visits decreased for older adults in all three racial and ethnic groups in Medicaid expansion and non-expansion states. The magnitude of the decrease varied from four visits per 1000 people among White older adults in non-expansion states to 23 visits per 1000 people among Black and Hispanic older adults in expansion states. The insurance coverage gains came primarily from Medicaid in expansion states and private insurance in non-expansion states, regardless of race or ethnicity. We find suggestive evidence of an increase in ED visits for Black and Hispanic populations that had low insurance coverage rates before 2014.

Patient Travel Distance and Concordance with Geographic Market Definitions, (2024) Annals of Internal Medicine. Coauthored with Sandra Decker and Rebecca Gourevitch.

We examined patient travel patterns for healthcare services using data from the Medical Expenditure Panel Survey (2018-2021). Our analysis estimated median travel times of 12.7 minutes for primary care, 17.1 minutes for specialty care, and 18.1 minutes for inpatient stays, with consistently longer times for patients outside metropolitan areas. While 73.8% of ambulatory visits occurred within patients' counties, only 50.9% of primary care visits fell within Primary Care Service Areas and 60.5% of inpatient stays within Hospital Service Areas. These findings indicate that common geographic market boundaries often inadequately capture actual care utilization patterns. When feasible, we recommend flexible, data-driven approaches that incorporate observed care-seeking patterns rather than relying on fixed geographical boundaries for measuring healthcare access and competition.

Children's Mental Health: Living with Parents who had Adverse Childhood Experiences, (2024) Health Affairs. Coauthored with Sandra Decker, Xue Wu, and Samuel Zuvekas.

In 2021, 42.8 percent of US children ages 5–17 lived with an adult who had at least four adverse childhood experiences (ACEs). We found differences in exposure by race and ethnicity, income, and public versus private insurance status. Childhood exposure to adult ACEs was negatively correlated with child mental health, particularly among children in low-income and publicly insured families.

COVID-19 Admission Rates and Changes in Care Quality in US Hospitals, (2024) JAMA Network Open. Coauthored with Pamela Owens, Sandra Decker, et al.

This study examined the association of COVID-19 admission rates with changes in hospital care quality for patients without COVID-19.  Using data from 3283 acute care hospitals in 36 states and more than 19 million patient discharges, we compared quality indicators in 2020 to 2019. We found that pressure ulcers and in-hospital mortality for nonsurgical care increased in 2020 during weeks with high COVID-19 admissions compared with weeks with low COVID-19 admissions. Increases were statistically significant and clinically meaningful; for example, pressure ulcer, heart failure mortality, and hip fracture mortality rates all increased by at least 20% during weeks with high compared with low COVID-19 admissions. These findings suggest that COVID-19 surges were associated with decreases in hospital quality, highlighting the importance of future strategies to maintain care quality during periods of high use.

Interruptions in Insurance Coverage and Prescription Drug Utilization: Evidence from Kentucky, (2024) Medical Care Research and Review.

This study examined how interruptions in insurance coverage affect purchases of prescription drugs for young adults. It used data spanning 2014 to 2018 from Kentucky’s prescription drug monitoring program, which tracked the universe of federally-regulated (Schedule II–V) prescription drugs dispensed in the state. The study employed a regression discontinuity design based on the age limit at 26 for dependent insurance coverage for children. At age 26, the probability of purchasing a prescription decreased by 5%, with all subcategories of prescriptions affected. The share of generic prescriptions increased for stimulants (the only category observed with substantial branded prescriptions). By age 27, prescription purchases returned to levels observed at 25, but the share purchased with public insurance and the generic share for stimulants remained higher. The findings suggest that interruptions in insurance coverage decrease prescription drug utilization by young adults and that public insurance programs such as Medicaid are important for resuming treatment.

COVID-19 Admission Rates and Changes in US Hospital Inpatient and Intensive Care Unit Occupancy, (2023) JAMA Health Forum. Coauthored with Sandra Decker, Pamela Owens, and Thomas Selden. 

The COVID-19 pandemic had unprecedented effects on hospital occupancy, with consequences for hospital operations and patient care. This paper used data from 3960 US hospitals in 45 US states to examine the association between COVID-19 admission rates and inpatient and intensive care unit (ICU) occupancy rates in 2020. We found that weekly rates of COVID-19 admissions were less than 4 per 100 beds for 64% of hospital-weeks and at least 10 per 100 beds in 16% of hospital-weeks. During weeks with low COVID-19 admissions (<1 per 100 beds), inpatient occupancy decreased by 13%, whereas during weeks with high COVID-19 admissions (≥15 per 100 beds), inpatient occupancy increased by 8% and ICU occupancy increased by 68%. We conclude that COVID-19 admission rates were associated with substantial changes in occupancy, with ICUs especially strained during surges.

Can the Boundaries of the Firm Make it Hard to Get In? The Consequences of Firm Scope and Scale on Patient Access to Care, (2023) Health Services Research, coauthored with Thomas Koch, Brett Wendling, and Samuel Zuvekas.

Over the past decade, healthcare providers have reorganized into larger firms with broader scope. In this paper, we used 2008-2019 data from Medicare fee-for-service claims and the Medical Expenditure Panel Survey (MEPS) to quantify changes in the market structure of primary care physicians and examine its relationship with access to care. We found that the percentage of people living in concentrated ZIP codes (HHI above 1500) increased from 37% in 2008 to 53% in 2019. During the same period, the median market share of multispecialty firms rose from 30% to 48%. MEPS respondents in highly concentrated ZIP codes (HHI over 2500) were 5.9 percentage points (95% CI: -1.4 to -10.4) less likely to report having access to immediate care than respondents in unconcentrated ZIP codes. The association was largest among Medicaid beneficiaries, a 17.3 percentage point reduction (95% CI: -5.1 to -29.4). We found no association between HHI and indicators for having a usual source of care and annual checkups. The multispecialty market share was negatively associated with checkups, but not other measures of access.

Mandatory Prescription Drug Monitoring Programs and Overlapping Prescriptions of Opioid and Benzodiazepines: Evidence from Kentucky, (2023) Drug and Alcohol Dependence, coauthored with Thuy Nguyen and Thomas Buchmueller.

In response to the opioid epidemic, many states implemented mandates requiring providers to check prescription drug monitoring programs (PDMPs) before prescribing opioids. We examine how overlapping benzodiazepine and opioid prescriptions changed after Kentucky implemented a PDMP mandate in July 2012. We conducted an interrupted time series analysis using monthly data from Kentucky’s PDMP from 2010 to 2016. Separate analyses were conducted for overlapping prescriptions from a single provider or multiple providers, and by sex and age group. We also conducted an individual-level longitudinal analysis that compared changes in utilization patterns after the mandate went into effect to changes in earlier periods during which the mandate was not in effect. Kentucky’s PDMP mandate was associated with an immediate 7.5 % decline in the rate of overlapping benzodiazepine and opioid prescriptions and a significant change in the trend from increasing to decreasing. Approximately half of the immediate effect in level terms was explained by decreases in overlapping prescriptions written by a single provider. Our longitudinal analysis suggests that over one year the mandate reduced initiation of overlapping prescriptions by 29.3 % and reduced continuation of overlapping prescriptions by 9.4 %. The effects of the policy were largest for women and men aged 36–50. Though not the main rationale for the policy, Kentucky’s PDMP mandate reduced overlapping prescriptions of benzodiazepines and opioids. Further efforts to reduce overlapping prescriptions should consider the effects on populations such as women over 50, who have high rates of overlapping prescriptions.

The Effects of the Medicaid Expansion on Hospital Utilization, Employment, and Capital, (2022) Medical Care Research and Review, coauthored with Brady Post.

In recent years, hospitals reacted to changes in demand caused by the Affordable Care Act Medicaid expansions. We conducted a difference-in-differences analysis that compared changes to hospital demand and supply in Medicaid expansion and nonexpansion states. We used 2010–2016 data from the American Hospital Association and the Healthcare Cost Report Information System to quantify changes to hospital utilization and characterize how hospitals adjusted labor and capital inputs. During the period studied, the Medicaid expansion was associated with increases in emergency department visits and other outpatient hospital visits. We find strong evidence that hospitals met increases in demand by hiring nursing staff and weaker evidence that they increased hiring of technicians and investments in equipment. We found no evidence that hospitals adjusted hiring of physicians, support staff, or investments in other capital inputs.

Provider Compliance With Kentucky’s Prescription Drug Monitoring Program’s Mandate To Query Patient Opioid History, (2021) Health Affairs, coauthored with Colleen Carey and Thomas Buchmueller.

Forty states mandate that providers query their patients’ prescription histories in the state’s prescription drug monitoring program (PDMP) before prescribing controlled substances. However, little is known about providers’ use of PDMPs, either with or without a mandate. We measured the share of opioid prescriptions with PDMP queries in Kentucky from 2010 to 2018, before and after the implementation of the first comprehensive PDMP mandate in the US. Providers queried the PDMP for 12 percent of opioid prescriptions before the mandate; after the mandate, they queried for 56 percent of prescriptions. The share of prescriptions queried was lowest for patients without recent opioid use (3 percent before the mandate, 25 percent after) and highest for pain management specialists (31 percent before, 72 percent after). Over time, high-compliance providers reduced prescribing to the riskiest patients, whereas low-compliance providers continued to prescribe to them. Although the share of prescriptions queried greatly increased after the mandate, compliance remained incomplete, including for patients with high-risk patterns of opioid use.

How Well Do Doctors Know Their Patients? Evidence from a Mandatory Access Prescription Drug Monitoring Program, (2020) Health Economics, coauthored with Thomas Buchmueller and Colleen Carey.  

Many opioid control policies target the prescribing behavior of health care providers. In this paper, we study the first comprehensive state‐level policy requiring providers to access patients' opioid history before making prescribing decisions. We compare prescribers in Kentucky, which implemented this policy in 2012, to those in a control state, Indiana. Our main difference‐in‐differences analysis uses the universe of prescriptions filled in the two states to assess how the information provided affected prescribing behavior. We find that a significant share of low‐volume providers stopped prescribing opioids altogether after the policy was implemented, though this change accounted for a small share of the reduction in total volume. The most important margin of response was to prescribe opioids to fewer patients. Although providers disproportionately discontinued treating patients whose opioid histories showed the use of multiple providers, there were also economically meaningful reductions for patients without multiple providers and single‐use acute patients.