Document Type


Publication Date


Publication Title

Journal of clinical anesthesia


STUDY OBJECTIVE: Our goal was to determine when postoperative delirium first occurs, and to assess evaluation strategies that reliably detect delirium with lowest frequency of testing.

DESIGN: This was a retrospective study that used a database from a five-center randomized trial.

SETTING: Postoperative cardiothoracic ICU and surgical wards.

PARTICIPANT: Adults scheduled for elective coronary artery bypass and/or valve surgery.

INTERVENTION AND MEASUREMENTS: Postoperative delirium was assessed using CAM-ICU questionnaires twice daily for 5 days or until hospital discharge. Data were analyzed using frequency tables and Kaplan-Meier time-to-event estimators, the latter being used to summarize time to first positive CAM-ICU over POD1-5 for all patients for various evaluation strategies, including all assessments, only morning assessment, and only afternoon assessments. Sensitivity for various strategies were compared using McNemar's test for paired proportions.

MAIN RESULTS: A total of 95 of 788 patients (12% [95% CI, 10% to 15%]) had at least 1 episode of delirium within the first 5 postoperative days. Among all patients with delirium, 65% were identified by the end of the first postoperative day. Delirium was detected more often in the mornings (10% of patients) than evenings (7% of patients). Compared to delirium assessments twice daily for five days, we found that twice daily assessments for 4 days detected an estimated 97% (95% CI 91%, 99%) of delirium. Measurements twice daily for three days detected 90% (82%, 95%) of delirium.

CONCLUSIONS: Postoperative delirium is common, and CAM-ICU assessments twice daily for 4 days, versus 5 days, detects nearly all delirium with 20% fewer assessments. Four days of assessment may usually be sufficient for clinical and research purposes.

PubMed ID




First Page


Last Page