Predictive model for assessment of ICU delirium among patients in critical care units: Prospective observational...

Delirium is a major neurocognitive/psychotic disorder that is secondary to another medical condition. However, its accurate prediction at the optimal time is often missed. The study predicted ICU delirium among patients admitted into intensive care units (ICU) by using PRE-DELIRIC and E-PRE-DELIRIC models and compared the performance of these models in identifying the predictors of this condition.

A prospective observational study was conducted with 350 patients admitted to ICUs who were aged more than 18 years. Using the E-PRE-DELIRIC and PRE-DELIRIC models, data were collected from consecutive patients within 24h of admission. The data were analyzed through descriptive and inferential statistics by using the Statistical Package for the Social Sciences (SPSS) 16.0 version.

The receiver operating characteristic curve (ROC) was plotted using the sample positive for delirium and belonging to the category of more than 20% chance of developing delirium according to the E-PRE-DELIRIC and PRE-DELIRIC models. The area under the curve (AUC) values of the E-PRE-DELIRIC and PRE-DELIRIC scores were 0.717 and 0.760, respectively. PRE-DELRIC had a sensitivity and specificity of 76.9% and 62.7%, respectively, with the cut-off value being 26.50 for predicting delirium. The E-PRE-DELIRIC had a sensitivity and specificity of 71.0% and 67.3%, respectively, with the cut-off value being 30 for predicting delirium. A statistically significant association was observed between gender, occupation, comorbidity, and ICU delirium.

With a higher AUC and better sensitivity, the PRE-DELIRIC model was a comparatively better predictor of true-positive delirium cases.