Impact of implementing electronic health records on the occurrence of adverse medical events in hospitals

Background: The Institute of Medicine in 1999 estimated that 44,000-98,000 Americans die each year as a result of preventable medical errors. The expanded use of health information technology (HIT) would lend great opportunities in improving the quality of health care, but there has been little information to empirically assess the relationship between HIT use and medical errors.

Objective: The goal of this study is to understand the impact of electronic health record (EHR) implementation on the likelihood of adverse medical events (AMEs).

Methods: The analytic dataset was compiled with discharge data from 50 ambulatory surgery centers, hospitalizations from 137 hospitals, outpatient discharges from 96 hospitals, and facility characteristics from the Annual Cooperative Hospital Survey in the state of Oklahoma. The study population included 529,332 patients. Hierarchical generalized linear modeling estimated with Generalized Estimating Equations that allowed for the clustering of observations within hospitals was used to model the likelihood of AMEs as a function of HIT implementation, controlling for patient- and hospital-level characteristics. AMEs were defined using ICD-9 codes.

Results: Sixty-percent of the study population were female, 13.5% were age 19 and younger, 23.4% were between 20-44, 26.5% were between 45-64, and 35.6 were 65 and older. Forty-percent of the hospitals were for-profit, 28.8 were nonprofit, and 32.2% were government controlled. Patients who received care from facilities with a fully implemented EHR were less likely to have an AME (OR=0.41, 95% CI: 0.20, 0.83), comparing to those who were in facilities with a partially implemented HIT or paper-based system. Patient-level variables such as age, comorbidity, and length of stay, and facility-level characteristics such as profit status and governance, were also positively correlated with AME occurrence.

Conclusion: Findings showed that AMEs were less likely to occur with fully implemented EHRs. Although high capital costs have often been cited as a primary barrier for HIT implementation, the reduction in AMEs presents a business case for return on investments in support of wider HIT adoption and use.