An evaluation of emergency physician selection of observation unit patients

Objective: Appropriate patient selection is critical for maximal observation unit (OU) effectiveness. We hypothesized emergency physicians underuse the OU for admitted patients and overuse the OU for patients who would otherwise be discharged. Methods: Treating emergency physicians were asked about patient suitability for admission to an OU at a busy, urban, academic emergency department (ED) as part of a prospective cohort study of ED patients who were admitted or had an ED length of stay exceeding 4 hours. The OU was closed for renovation during the 2-month study, so physician opinion could be compared with patient course in the absence of observation services. Two blinded emergency physicians reviewed charts using structured forms and explicit definitions to determine actual patient course. Hospitalized patients were considered potential OU candidates according to a priori criteria: (1) hospital length of stay less than 48 hours, (2) no procedure or diagnosis requiring hospitalization, and (3) no death. Results: Of 1747 enrolled patients, 131 were excluded with incomplete data. Median age was 45 years. Patients were 40% white and 48% men. Emergency physicians identified 363 (23%) patients as observation candidates. Of these, 182 (50%) were actually discharged directly. The remaining 181 (50%) were hospitalized; 101 (56%) were observation candidates based on chart review. Of 799 admitted patients not selected for observation, 232 (29%) were suitable for observation by chart review. Conclusions: Selection of patients for observation was suboptimal; emergency physicians routinely identified patients as OU candidates who were not ultimately admitted, and they missed many admitted patients who might have been appropriate OU candidates. Both over- and underuse should be addressed to maximize the effectiveness of OUs.