Validation of Electronic Health Record Detection of Patient Safety Outcomes

Abstract

Background: Adverse events (AE) are common for hospitalized Canadians, and lead to worse patient outcomes. Monitoring patient safety is logistically challenging. We aimed to validate the use of our electronic health record (EHR) to monitor important patient safety. Methods: EHR data was used to identify patients who were exposed to one of four high-priority safety outcomes: venous thromboembolism, dysglycemia, Clostridium difficile(C. difficile) infection, and prolonged nil per os(NPO) orders. A manual chart review was performed to determine the sensitivity and specificity of the EHR for each patient safety outcome. Results: The sensitivity and specificity ranged from 38.4% to 78.1% and 88.1 to 99.2%, respectively, for the prespecified patient safety outcomes. Conclusion: The EHR is reasonably sensitive and specific to monitor rates of dysglycemia, C. difficileinfection, and prolonged NPO in medical inpatients, but does not have adequate sensitivity to be used to capture venous thromboembolism safety outcomes.