Electronic Medical Records and Medical Procedure Choice: Evidence from Cesarean Sections

The over use of medical technology in the U.S. is often considered a key factor in the rapid growth of healthcare expenditures. In addition to diffusion of medical technologies, healthcare providers have also increased their use of health information technology (HIT) in recent decades. HIT is intended to act as a complement to medical technology, and proponents expect that further implementation of HIT is likely to improve healthcare quality while reducing costs. This paper explores the interaction between HIT and the utilization of medical treatment technology in the context of Cesarean sections for childbirth. We ask whether hospital EMR adoption can improve providers’ ability to match patients with their most appropriate treatment, thus decreasing the use of treatments for patients with little or no benefit. More appropriate patient-procedure matching could improve health outcomes and reduce treatment costs if it reduces the prevalence of C-sections among lower-acuity mothers.  Our paper provides a unique contribution by tying the literature on the effects of EMR adoption to the literature on the utilization of expensive medical technology.

We describe a conceptual model in which EMR adoption impacts how likely a hospital is to prescribe a C-section delivery to a low risk patient and impacts the slope of the relationship between patient-risk and C-section provision. This model implies an empirical specification in which we exploit within hospital variation over time in EMR adoption and estimate the heterogeneous impact of EMR adoption on C-sections as patient risk level varies. We also examine whether effects of EMR adoption differ by a hospital’s pre-adoption level of C-section intensity. Finally, we consider the fact that changes in C-section rates could have ambiguous welfare effects and examine how EMR adoption changes patient health outcomes that reflect the tradeoff of bad outcomes as patients shift between delivery types.

We find consistent evidence that Computerized Practitioner Order Entry decreases the use of C-sections among low risk women in hospitals that were already low-intensity. We find some evidence that Obstetric EMRs decreases C-section rates among low-risk women, but this effect occurs with a lag. We also find that Obstetric EMRs in low-intensity hospitals increase C-section rates for medium risk women, the group with the most uncertainty. We do however find that these shifts are accompanied by increases in bad outcomes, suggesting that the marginal patient is potentially being shifted to a less-appropriate procedure. We find no evidence of EMRs changing procedure use in high intensity hospitals or for high-risk women.

Our results suggest that hospitals more likely to already be less intensively treating lower risk patients use EMRs as a tool to further decrease C-section rates. However, the welfare effects of these changes are unclear given our findings of increases in some bad outcomes. These findings suggest that future research should explore more comprehensive cost and health outcomes for mothers and infants, both in the hospital and after the hospital stay.