BackgroundDelirium screening instruments (DSIs) should be used to detect delirium, but they only show moderate sensitivity in patients with neurocritical illness. We explored whether, for these patients, DSI validity is impacted by patient-specific covariates.
MethodsData were prospectively collected in a single-center quality improvement project. Patients were screened for delirium once daily using the Intensive Care Delirium Screening Checklist (ICDSC) and the Confusion Assessment Method for the Intensive Care Unit (CAM-ICU). Reference was the daily assessment using criteria from the Diagnostic and Statistical Manual, 4th Edition, Text Revision (DSM-IV-TR). In a two-step receiver operating characteristics regression analysis adjusting for repeated measurements, the impact of acute diagnosis of stroke or transient ischemic attack (TIA), neurosurgical intervention, Richmond Agitation Sedation Scale, and ventilation status on test validity was determined.
ResultsOf 181 patients screened, 101 went into final analysis. Delirium incidence according to DSM-IV-TR was 29.7%. For the first complete assessment series (CAM-ICU, ICDSC, and DSM-IV-TR), sensitivity for the CAM-ICU and the ICDSC was 73.3% and 66.7%, and specificity was 91.8% and 94.1%, respectively. Consideration of daily repeated measurements increased sensitivity for the CAM-ICU and ICDSC to 75.7% and 73.4%, and specificity to 97.3% and 98.9%, respectively. Receiver operating characteristics regression revealed that lower Richmond Agitation Sedation Scale levels significantly impaired validity of the ICDSC (p = 0.029) and the CAM-ICU in its severity scale version (p = 0.004). Neither acute diagnosis of stroke or TIA nor neurosurgical intervention or mechanical ventilation significantly influenced DSI validity.
ConclusionsThe CAM-ICU and ICDSC perform well in patients requiring neurocritical care, regardless of the presence of acute stroke, TIA, or neurosurgical interventions. Yet, even very light or moderate sedation can significantly impair DSI performance.