Three core domains of delirium validated using exploratory and confirmatory factor analyses.

OBJECTIVE

To confirm the existence of the proposed three-core symptom domains in delirium by analyzing a dataset of nondemented adults using selected core symptoms as measured by the Delirium Rating Scale-Revised-98 (DRS-R98) scale.

METHODS

Exploratory factor analysis (EFA) and confirmatory factor analysis (CFA) of proposed delirium core symptoms were conducted in a pooled international dataset of 592 delirious and nondelirious patients using DSM-IV criteria from 14 studies with comparable methodologies. Using DRS-R98 categorization, 445 had either subsyndromal or full delirium and comprised the delirium group. The dataset was divided into three independent random subsamples to perform a stepwise analysis. First we performed EFA in 100 cases to delineate latent factor loadings of DRS-R98 items selected to represent the three-core domains (circadian, higher level thinking, and cognitive). These items were then assessed using CFA-modeling (n = 246) followed by a CFA-validation (n = 246). Reliability and goodness of fit of these two CFA were assessed statistically.

RESULTS

DRS-R98 items representing the proposed delirium core symptoms loaded onto one factor in the EFA, supporting their core nature. The two CFA confirmed the nature of this core factor as comprising three core domains where DRS-R98 items each loaded with high values (>0.7) onto their corresponding core domain (circadian, higher level thinking, and cognitive) with good fit and reliability. Attention was DRS-R98 item with the highest loading in CFA, followed by thought process, and then by sleep-wake cycle and motor behavior.

CONCLUSIONS

Our EFA and CFA confirm and validate the proposed three-core domains of delirium, where symptoms were highly related to the domain that they were hypothesized to represent. These domains are consistent with delirium being a state of impaired consciousness, and should be considered necessary to assess whether in clinical or research settings.