Screening for delirium with the Intensive Care Delirium Screening Checklist (ICDSC): a re-evaluation of the threshold for delirium.

BACKGROUND With its high incidence and subsequent adverse consequences in the intensive care setting, several instruments have been developed to screen for and detect delirium. One of the more commonly used is the Intensive Care Delirium Screening Checklist (ICDSC); however, the optimal cut-off score indicating delirium has been debated. METHODS In this prospective cohort study, the ICDSC threshold for delirium set at ≥3, ≥4, or ≥5 was compared with the DSM-IV-TR-determined diagnosis of delirium (used as standard), and with the Confusion Assessment Method for the ICU (CAM-ICU), with respect to their concurrent validity. RESULTS In total, 289 patients were assessed, including 122 with delirium. The cut-off score of ≥4 had several shortcomings: although 90% of patients with delirium were correctly classified, 23% remained undetected. The agreement with the DSM-IV-TR diagnosis of delirium was only moderate (Cohen's κ 0.59) and the sensitivity was only 62%. In contrast, when the cut-off was ≥3, 83% of patients with delirium were correctly classified and only 14.5% remained undetected. The agreement with DSM-IV-TR was substantial (Cohen's κ 0.68) and the sensitivity increased to 83%. The benefit of setting the cut-off at ≥5 was not convincing: although 90% of patients with delirium were correctly classified, 30% remained undetected. The concurrent validity was only moderate (Cohen's κ 0.44), and the sensitivity reached only 44%. Changing the ICDSC cut-off score did not strengthen the moderate agreement with the CAM-ICU (Cohen's κ 0.45-0.56). CONCLUSION In clinical routine, decreasing the ICDSC threshold for delirium to ≥3 increased the accuracy in detecting delirium at the cost of over-identification and is therefore recommended as the optimal threshold. Increasing the cut-off score to ≥5 decreased the concurrent validity and sensitivity; in addition, the under-detection of delirium was substantial.