The Impact of Electronic Health Record Adoption and Integration on Physician Productivity and Health Outcomes

We investigate the impact of the installation and integration of an outpatient electronic health record (EHR) on physician productivity and health outcomes. In contrast to previous studies, we use a mixed methods approach and analyze a direct measure of physician productivity: relative value units.  We use data from a large health care network in Eastern Pennsylvania, collected between July 2007 and December 2012.  This time period spans the staggered installation of a new outpatient EHR at four outpatient OB/GYN practices, and its integration with the hospital’s information system. Using risk adjusted physician- and time-fixed effects models with practice-specific linear and quadratic time trends, we investigate the impact of the following three stage of EHR installation and integration on relative value units and a measure of adverse birth outcomes collected via chart review: 1) Initial installation; 2) Transmission of a discharge summary from the inpatient Labor and Delivery Triage Unit to the practices; 3) Transmission of discrete clinical data elements from the Triage Unit to the practices. In addition, we analyze qualitative data from multiple rounds of interviews with physicians and medical staff in order to identify the underlying changes in work processes that are responsible for the EHR effects derived from our empirical model. Our estimates indicate that the EHR installation led to a four-year reduction in productivity, but that it also reduced the severity of adverse birth outcomes.  In addition, we find evidence that the EHR influenced clinical practice norms.  In particular, the rate of Caesarean sections increased during the time period when productivity was falling, and induced births increased after the transmission of discrete clinical data from the Triage Unit to the OB/GYN practices. Qualitative data from physician interviews suggest that much of the productivity reduction resulted from additional documentation requirements associated with the new system, as well as from information overload caused by the integration of information into a single record. Eventually, new policies were developed to address these issues, allowing productivity to return to pre-installation levels. Our results indicate that it may take a longer time period than initially appreciated for physicians to recover from productivity shocks associated with the installation of EHRs and similar technologies, and that complementary changes in work processes are need to avoid significant productivity declines. Nonetheless, these new systems also have the potential to improve patient health.