Mortality Effects of Healthcare Supply Shocks: Evidence Using Linked Deaths and Electronic Health Records
The contraction in health care consumption at the start of the pandemic provides insight into central economic questions of waste and productivity in the U.S. health care system. Using linked mortality and Electronic Medical Records, we compare people who had outpatient appointments scheduled for dates in 30 day periods immediately before and after the Covid-19 emergency declaration. Appointment cancellation rates were 77% higher for people with appointments in the shutdown period. Intent to treat estimates imply that having a scheduled appointment date right after the emergency declaration increased one-year mortality rates by 4 deaths per 10,000. Instrumental variable estimates suggest that a cancelled appointment increased one-year mortality by 29.7 deaths per 10,000 among compliers, implying that a 10% increase in health care appointments reduces mortality rates by 2.9%. The mortality effects are rooted in two mechanisms: a complier sub-population with high marginal benefits from care, and a cascade of delayed or missed follow-up care that lasted for about 3 months. Healthcare spending accounted for 19.7% of U.S. GDP in 2021, and controlling health spending is a major policy objective. Our results quantify health tradeoffs from cutting every-day non-emergency visits, illustrating the importance of cost-control efforts that differentiate between medical care with the largest and smallest benefits for patient health.
We would like to thank Diane Alexander, Sumedha Gupta, Alex Hollingsworth, Amanda Kowalski, Timothy Moore, Jim Poterba, Julian Reif, Adrienne Sabety, Daniel Sacks, Jon Skinner and Christopher Whaley for helpful comments. We would like to thank Claire Cravero, Quinn Johns, Alyssa Chen, Aleah Peffer, Isabel Francisco from the Covid-19 Research Database/Datavant for facilitating our access to the Connected Death Index data and providing a secure computing environment to link the EMR and deaths records. We would like to thank Matt Shober, Rose Baumgardner, Bob Suhendra and Cliff Cavanaugh from Healthjump for answering our many questions about the data. We would also like to thank Danny Chang, Vishal Singh, Vamsi Bushan, Naba Sahoo and Madeline Yozwiak for excellent research assistance. This research was made possible by partial funding from the National Institutes of Health Contract No. 75N95D20F4000 The views expressed herein are those of the authors and do not necessarily reflect the views of the National Bureau of Economic Research.
- A new study of COVID-19-related disruptions in health care sheds light on the value of nonurgent medical care. In Mortality Effects of...