Validation of a chart review method for identifying delirium in the emergency department

To validate the Chart-based Delirium Identification Instrument (CHART-DEL) which was developed in the inpatient setting, for identifying delirium in the emergency department (ED).

This retrospective observational study included ED patients aged 75 and older who presented between 2021 and 2023 and were screened for delirium utilizing a validated two-step tool which included the Delirium Triage Screen (DTS) and the Brief Confusion Assessment Method (bCAM). A stratified random sample of 600 medical records were reviewed using the CHART-DEL method. Inter-rater agreement and diagnostic test accuracy were calculated. We followed standardized guidelines for reporting (STROBE).

A total of 27,082 visits were included. Median age was 83 years, 51.8 % were women. Twenty percent of patients had known dementia and/or cognitive impairment, and 48.8 % were admitted to the hospital. There were 1197 patients (4.4 %) with delirium in the ED. When compared to the ED assessment, the chart review correctly identified 474/600 cases (accuracy 79.0 %, 95 % CI: 75.5-82.1 %), including 223/300 with delirium (sensitivity 74.3 %, 95 % CI: 68.9-79.1 %), and 251/300 without delirium (specificity 83.7 %, 95 % CI: 78.9-87.6 %). Agreement between the reviewers was 91.0 %. Weight calculations to approximate the original population resulted in an accuracy of 86.4 % (95 % CI: 86.0-86.8 %), sensitivity of 74.2 % (95 % CI: 71.6-76.6 %) and specificity of 87.0 % (95 % CI: 86.6-87.4 %).

The CHART-DEL can be applied to the ED setting, but performed better at correctly identifying patients who do not have delirium. Automating this method could improve retrospective delirium detection in the ED and support future clinical applications of delirium screening tools.