Validity and reliability of the DDS for severity of delirium in the ICU

Abstract

Introduction: Until now, there has been no gold standard for monitoring delirium in intensive care unit (ICU) patients. In this prospective cohort study, a new score, the Delirium Detection Score (DDS), for severity of delirium in the ICU was evaluated.

Methods: After ethical approval and written informed consent, intensive care doctors and nurses assessed 1073 consecutive patients in surgical ICUs using the DDS together with the Ramsay Sedation Scale (RSS). The DDS is composed of eight criteria (orientation, hallucination, agitation, anxiety, seizures, tremor, paroxysmal sweating, and altered sleep- wake rhythm). Additionally, intensive care doctors had to document the Sedation-Agitation Scale (SAS) combined with a defined clinical assessment. For interrater reliability, pair of evaluators assessed patients in a blinded fashion at the same time.

Results: RSS1 (9%) was associated with a significantly (p < 0.001) higher DDS than RSS levels 2-6. The DDS increased with the severity of delirium (p < 0.001). The receiver operating characteristics (ROC) for the differentiation between no delirium (SAS < 4) and symptoms of delirium at all (SAS 5-7) showed an area under the curve (AUC) of 0.802 (95% confidential interval (CI): 0.719-0.898; p < 0.001) and 69% sensitivity and 75% specificity was determined. For reliability, a Cronbach's alpha of 0.667 was calculated. The paired comparisons revealed an intraclass correlation between 0.642 and 0.758.

Conclusion: The DDS demonstrated good validity with excellent sensitivity and specificity for delirium. The severity of delirium can be more accurately estimated by the DDS. By its composition of several items, the DDS might help to start a symptom-guided therapy immediately.