Project Summary
Background: High-deductible health insurance plans, which are characterized by low premiums coupled with
substantial costs for accessing medical services, are one of the most common tools for containing growing
health spending in the United States. Currently, nearly one-third of U.S. adults with employment-based health
insurance coverage—some 50 million people—have a deductible higher than $2000. In line with theoretical
predictions and intended policy goals, robust evidence shows high-deductible health plans reduce spending by
reducing medical service use. Yet, questions remain about whether these reductions come at the expense of
care valuable for patient health. These concerns are especially acute for low-income enrollees who are at
greater risk for chronic conditions. As high-deductible health plans continue to proliferate in the United States,
actionable information is needed about their impact on use of clinically valuable health care services and
whether high-deductible plans exacerbate existing income-related disparities healthcare utilization.
Objective: This project will assess the effect of high-deductible health plan enrollment on medical service
utilization and spending and how the effect varies across household income.
Methods: I use a quasi-experimental research design known as a difference-in-differences analysis to
measure outcomes among individuals before and after high-deductible plan enrollment relative to a
comparison group whose health insurance does not change. To mitigate bias caused by individual selection
into plans, I analyze the subset of individuals who enroll due to an employer-mandated switch. I perform the
analysis on a large individual-level dataset that contains detailed medical and pharmaceutical claims
information for approximately 3.1 million people with employment-based insurance who lived in a U.S. state
from 2015 to 2019. Unlike most claims datasets, the data link individuals to their census block group of
residence, a precise geographic identifier that can be merged with publicly available income data at the same
level of detail. I measure overall spending and utilization and then focus on specific bundles of services shown
to be clinically valuable for preventing or treating disease or, conversely, to have little to no value in
maintaining health, so-called low-value care. I compare results on these outcomes by income using quintiles
of the observed income distribution.
Impact: The analysis will generate policy relevant and readily translatable estimates of the effects of high-
deductible plans on health service use, including use of valuable versus low-value care, and allow for
comparison of differences across income. This project will help policymakers and public health experts better
target interventions to mitigate potential negative effects of high-deductible plans on valuable service utilization
and inform the current debate about the role of health insurance in health disparities.
