Objective: To compare validity and reliability of three instruments for detection and assessment of delirium in intensive care unit (ICU) patients. Delirium in critically ill patients is associated with higher mortality, prolonged duration of ICU stay, and greater healthcare costs. Currently, there are several assessment tools available for detection of delirium, but only a few of these assessment systems are developed specifically to screen for delirium in ICU patients.
Design: Prospective cohort study.
Setting: ICU at a university hospital.
Patients: A total of 156 surgical patients aged > or = 60 yrs consecutively admitted to the ICU, with a length of stay of at least 24 hrs.
Measurements and main results: This study was approved by the institutional ethics committee. Trained staff members performed daily and independently the Confusion Assessment Method for the ICU (CAM-ICU), the Nursing Delirium Screening Scale (Nu-DESC), and the Delirium Detection Score (DDS). These evaluations were compared against the reference standard conducted by a delirium expert (blinded to the study), who used delirium criteria from the Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition (DSM-IV). Of 156 patients, 63 (40%) were identified as delirious by the reference standard during the study. Using the CAM-ICU and the Nu-DESC, we measured comparable sensitivities (CAM-ICU, 81%; Nu-DESC, 83%). The specificity of the CAM-ICU was significantly higher than that of the Nu-DESC (96% vs. 81%, p < .01). In contrast, the DDS showed poor sensitivity (30%), whereas the specificity was significantly higher compared with the Nu-DESC (DDS, 91%; Nu-DESC, 81%, p < .05). The interrater reliability was "almost perfect" for the CAM-ICU (kappa = 0.89) and "substantial" for DDS and Nu-DESC (kappa = 0.79, 0.68).
Conclusion: The CAM-ICU showed the best validity of the evaluated scales to identify delirium in ICU patients. The Nu-DESC might be an alternative tool for detection of ICU delirium. The DDS should not be used as a screening tool.