PRE-DELIRIC — PRE-DELIRIC delirium prediction model

What this tool is and how to interpret its evidence

PRE-DELIRIC is a multivariable model for estimating an adult ICU patient's probability of developing delirium during the admission. It uses ten risk factors available within the first 24 hours, spanning age, acute illness severity, admission circumstances, coma, infection, metabolic disturbance, selected sedative or opioid exposure and renal function. The model returns a probability rather than a diagnosis. It was designed to support preventive planning and systematic observation in patients who are not already delirious; an assessment tool is still needed when current delirium is suspected.

The original multicentre study included 3,056 patients. Development used 1,613 patients from one hospital, temporal validation used another 549 there, and external validation used 894 patients in four other Dutch hospitals. Reported areas under the receiver-operating-characteristic curve were 0.87 in development, 0.89 in temporal validation and 0.84 externally, with a pooled value of 0.85. Later evaluations show that discrimination and especially calibration can vary by population and implementation.

PRE-DELIRIC is completed after information from the first ICU day is available, so it is not the earliest admission model. Predictor definitions, medication practice, case mix and delirium ascertainment affect transportability. No single probability threshold is appropriate for every service. Local teams should validate calibration and decide what action a risk category triggers. A high probability should prompt prevention and repeated assessment, not treatment for presumed delirium; a low probability does not rule out later delirium.