Computerized Medication Alerts and Prescriber Mental Models: Observing Routine Patient Care

Abstract

Computerized medication alerts (e.g., drug-drug interactions, drug-allergy interactions), which are intended to protect patient safety, need to match the mental models of medication prescribers in order to aid medication ordering. To maximally protect patient safety, the programmer mental model, system image, and prescriber mental model should work seamlessly together to fully support prescriber decision-making. In this study, we examined prescribing processes in the context of routine patient care to understand how the design of medication alerts can be enhanced for prescribers. We shadowed prescribers, including physicians, pharmacists, and nurse practitioners, across five outpatient primary care clinics at a large Veterans Affairs Medical Center (VAMC). In addition, prescribers were opportunistically interviewed as they ordered mediations via a computerized order entry system and resolved any subsequent medication alerts. This investigation is one of the few to examine medication alerts by directly observing prescribers during patient care. Altogether, 191 medication alerts occurred across 63.5 total hrs of observation, 19 prescribers, and 86 patients during routine patient care tasks. Results reveal problematic system images and mismatches between programmer and prescriber mental models. Findings can help inform medication alert redesigns, which may promote safer, more effective prescribing practices.