Using a rule-based system to define error in the emergency department

The evaluation of peer-reviewed cases for error is key to quality assurance (QA) in emergency medicine, but defining error to ensure reviewer agreement and reproducibility remains elusive. The objective of this study was to create a consensus-based set of rules to systematically identify medical errors.

This is a prospective, observational study of all cases presented for peer review at an urban, tertiary care, academic medical center emergency department (ED) quality assurance (QA) committee between October 13, 2015, and September 14, 2016. Our hospital uses an electronic system enabling staff to self-identify QA issues for subsequent review. In addition, physician or patient complaints, 72-hour returns with admission, death within 24 hours, floor transfers to ICU < 24 hours, and morbidity and mortality conference cases are automatic triggers for review. Trained reviewers not involved in the patient's care use a structured 8-point Likert scale to assess for error and preventable or non-preventable adverse events. Cases where reviewers perceived a need for additional treatment, or that caused patient harm, are referred to a 20-member committee of emergency department leadership, attendings, residents, and nurses for consensus review. For this study, "rules" were proposed by the reviewers identifying the error and validated by consensus during each meeting. The committee then decided if a rule had been broken (error) or not broken (judgment call). If an error could not be phrased in terms of a rule broken, then it would not be considered an error. The rules were then evaluated by 2 reviewers and organized by theme into categories to determine common errors in emergency medicine.

We identified 108 episodes of rules broken in 103 cases within a database of 920 QA reviewed cases. In cases where a rule was broken and therefore an error was scored, the following 5 major themes emerged: (1) not acquiring necessary information (eg, not completing a relevant physical exam), N = 33 (31%); (2) not acting on data that were acquired (eg, abnormal vital signs or labs), N = 25 (23%); (3) knowledge gaps by clinicians (eg, not knowing to reduce a hernia), N = 16 (15%); (4) communication gaps (eg, discharge instructions), N = 17 (16%); and (5) systems issues (eg, improper patient registration), N = 17 (16%).

The development of consensus-based rules may result in a more standardized and practical definition of error in emergency medicine to be used as a QA tool and a basis for research. The most common type of rule broken was not acquiring necessary information. A rule-based definition of medical error in emergency medicine may identify key areas for risk reduction strategies, help standardize medical QA, and improve patient care and physician education.