Background The clinical decision supports system (CDSS) is the centerpiece of the electronic health record (EHR) incentive program to enhance patient care and prevent medication errors. Recent studies suggested that medication-related CDS alerts were commonly inappropriately overridden and rate of adherence was usually low. The rate of medication errors for patients admitted to the intensive care unit (ICU) is higher than that for other patients and inappropriately overridden alerts may affect patient care. The aim of this study was to evaluate the embedded CDS alerts and assess physicians’ compliance with medication-related CDS alerts in the ICU, by measuring the appropriateness of interruptive overrides of major severity alerts.

Methods This retrospective study was done by chart review of adults admitted to ICUs between January 2017 and December 2017 at a tertiary-care institution. The numbers and types of medication-related CDS alerts in adult ICUs were determined and physicians’ compliance with the alerts was assessed by measuring the appropriateness of the level 1 interruptive overrides.

Results A total of 42,883 CDA alerts were fired in the adult ICUs: 7.5% of alerts were severity level 1 (indicates a major severity alert), 20.70% were level 2 (indicates a moderate severity alert); and 71.8% were level 3 (indicates a minor severity alert). A total of 3200 overridden major severity alters (level 1) were included for evaluation of physician compliance. An overall appropriateness rate for overridden alerts was 49.9% and the significance varied by alert category (drug allergy: 66.7%; drug- drug interactions: 59.7%; drug disease: 55.4%; drug dose screening: 29%).

Conclusion Almost more than 80% of the CDS alerts were warned of potential significance in patient harm and others had little clinical impact. However, almost 50.1% were inappropriately overridden and further efforts modification should be focused to improve the CDS alert system, and an uninformative alert must turn off. A future investigation is important to assess why physicians have a low adherence rate for following the recommendations of CDS alerts.