Drug-related emergency department visits: external validation of an assessment tool in a general emergency department population

Background The process of identifying drug-related hospitalisations is subjective and time-consuming. Assessment tool for identifying hospital admissions related to medications (AT-HARM10) was developed to simplify and objectify this process. AT-HARM10 has not previously been externally validated, thus the predictive precision of the tool is uncertain. Aim To externally validate AT-HARM10 in adult patients admitted to the emergency department (ED). Method This retrospective cross-sectional study investigated 402 patients admitted to the ED, Diakonhjemmet Hospital, Oslo, Norway. A trained 5th-year pharmacy student used AT-HARM10 to assess all patients and to classify their ED visits as possibly or unlikely drug-related. Assessment of the same patients by an interdisciplinary expert panel acted as the gold standard. The external validation was conducted by comparing AT-HARM10 classifications with the gold standard. Results According to AT-HARM10 assessments, 169 (42%) patients had a possible drug-related ED visit. Calculated sensitivity and specificity values were 95% and 71%, respectively. Further, positive and negative predictive values were 46% and 98%, respectively. Adverse effects/over-treatment and suboptimal treatment were the issues most frequently overestimated by AT-HARM10 compared with the gold standard. Conclusion AT-HARM10 identifies drug-related ED visits with high sensitivity. However, the low positive predictive value indicates that further review of ED visits classified as possible drug-related by AT-HARM10 is necessary. AT-HARM10 can serve as a useful first-step screening that efficiently identifies unlikely drug-related ED visits, thus only a smaller proportion of the patients need to be reviewed by an interdisciplinary expert panel.