Unrecognized delirium in ED geriatric patients

Abstract

To determine the sensitivity of an emergency physician's conventional evaluation compared with the validated Confusion Assessment Method (CAM) regarding the recognition of acute confusional states (delirium) in elderly Emergency Department (ED) patients, a cohort of 385 patients presenting to an urban teaching hospital ED was systematically assembled. Patients had to be conscious, able to speak and older than 64 years of age. After the ED physician had examined the patient and test results had been obtained, a series of geriatric assessment results, including one for the likely presence of delirium, was made available to the ED physician; however, no result was specifically highlighted. All patients were assessed by an attending ED physician in the customary fashion. In addition, a study nurse interviewed patients using the CAM and followed patient outcomes for three months. The ED record for all patients with delirium or “probable” delirium, as determined by the CAM, were reviewed for physician diagnosis and disposition to determine how often delirium had been recognized by the emergency physician. Thirty-eight of the 385 patients screened (10%) met criteria for delirium or “probable” delirium; ED charts were complete for 35 of these, which constituted the study sample. The ED diagnosis included delirium or an acceptable synonym in 6 (17%) of these patients. In the 21 patients (62%) admitted to the hospital, the most common ED diagnosis was infection “rule out sepsis” (n = 7). Six of 13 patients discharged (46%) were diagnosed as “status post fall” without evidence of significant injury. The 3-month mortality rate for patients with delirium or “probable” delirium was 14% versus 8% for the non-delirium group (P = .20). These results suggest that the diagnosis of delirium may frequently be missed by the use of a conventional work-up in elderly patients who present to the ED. Educational efforts and/or use of formal assessment instruments may improve diagnostic sensitivity; however, formal evaluations of these strategies will be required.