Does health IT save money and lives? New evidence from vendor heterogeneity and product innovation

This paper examines the extent to which differences among Electronic Medical Record (EMR) vendors and differences in the timing of adoption affect the financial and clinical performance in hospitals. Under the 2009 Health Information Technology for Economic and Clinical Health (HITECH) Act, the government has provided $23.5 billion in subsidies to eligible hospitals and physicians to adopt certified EMRs. However, prior research has found little systematic econometric evidence that the technology is producing the anticipated effects. Prior research has also largely neglected the innovative nature of the industry and assumed that the products made by different vendors were homogeneous. Such assumptions may not be warranted in the research-intensive, competitive health IT industry.

Variation in research investments made by different EMR vendors might contribute to heterogeneity among products. We hypothesize that such product heterogeneity between vendors could impact product performance and help shed more light on the impact of EMRs on hospitals.

The innovative nature of the product may also have implications for assessing how the timing of adoption may impact hospital performance. On one hand, the latest systems might be more innovative. Hospitals that adopt more recent products (late adopters) may experience better performance than hospitals that adopt older systems (early adopters). We call it the innovation effect. On the other hand, given EMRs are very complex, more experience with a particular system may lead to improved performance. Consequently, early adopters may perform better than late adopters. We call it the experience effect. Our paper tests for the influences of innovativeness and experience with EMR systems on hospital performance.

We use Medicare inpatient claims data to construct our measures of costs and patient outcomes. We complement that data with hospital characteristics and IT adoption data from 2006 to 2010. The financial outcomes include total charges, Medicare payments, and other charges of various services. The patient outcomes include 30-day re-admission rates, 30-day mortality rates, and patient length of stay. We use a diff-in-diff estimation strategy. Our key identifying assumption is that the decision to adopt and the choice of EMR vendor are not driven by unobservable differences in costs or patient outcomes that are trending differently for adopters with different vendors and non-adopters prior to the adoption decision.

We find that EMR adoption leads to a reduction in hospital costs for Medicare inpatients, but that the reduction varies substantially by vendor. Our results suggest important difference among the EMRs offered by different vendors and those differences matter in terms of understanding whether a hospital will see reductions in costs as a consequence of adoption. We also find evidence to support both an innovation effect and an experience effect.

From a policy perspective, our results show that not all certified EMRs lead to the anticipated outcomes and suggest that the government's requirements for "meaningful use" (to be eligible for subsidies) could be too general.