ABSTRACT
The U.S. is experiencing a sea change in healthcare delivery, with ever more physicians employed by
hospitals. Cardiology, in particular, has changed dramatically: a decade ago, 90% of cardiologists were in
private practice, while today, in many states, a majority are employed by hospitals. Yet almost nothing is
known about how this “vertical integration” affects patient care or outcomes. We will study the effects of
hospital-cardiologist integration for patients hospitalized with incident acute myocardial infarction (AMI) or heart
failure (HF), using a novel integration measure and a large, longitudinal Medicare dataset. We will use time
variation in the amounts that Medicare FFS pays providers for cardiac tests as a plausibly exogenous driver of
integration, which permits difference-in-differences analysis and “causal” estimates of the effects of integration.
Integration proponents argue that integration can improve healthcare “value” – increase quality, reduce
utilization and cost, or both, through better care coordination and innovation in care delivery. However, there is
limited evidence to support these claims. Meanwhile, critics note that integration often leads to higher prices
paid by commercial insurers. Cardiology is an excellent area in which to study the effect of integration due to
dramatic growth in vertical integration, the importance of cardiovascular disease, and the complexity of AMI
and HF treatment, which increases the potential for integration to affect care and outcomes.
Our study sample is all cardiologists (~40,000) who care for the 4 million Medicare patients who experience
an incident AMI (1.5 million) or HF hospitalization (2.5 million) from 2002-2021. This large, longitudinal sample
provides sufficient power to assess a broad range of care processes, patient outcomes, utilization and costs
that the potential benefits of integration (care coordination, monitoring, innovation) can plausibly effect. As a
basis for this study, we developed, and pilot tested in Colorado, a measure of vertical integration, which
substantially outperforms prior measures in avoiding both false negatives (not identifying integrated
cardiologists) and false positives (identifying cardiologists as integrated when they are not).
It is critically important to patients, providers, insurers, and policymakers to determine how vertical
integration affects healthcare quality, outcomes, and cost. CMS is assessing whether to enforce pay parity
(thus removing the financial spur for integration) and legislators in several states are debating whether to
restrict integration, with limited evidence on its effects. This project is innovative because of its 1) use of a new,
validated, accurate measure of vertical integration, 2) use of a DiD research design which permits credible
causal inference, 3) evaluation of patients with incident AMI and HF hospitalization across a broad range of
care processes, patient outcomes and costs that integration can potentially effect, and 4) use of a large,
longitudinal Medicare dataset that provides sufficient power to detect small differences in outcomes.
