DYNAMIC-ICU — DYNAMIC-ICU delirium prediction rule

What this tool is and how to interpret its evidence

DYNAMIC-ICU is a clinical prediction rule intended to estimate and update the risk of developing delirium during an intensive-care admission. It is not a current-delirium screen. The rule combines seven predictors: chronic-disease history, hearing impairment, infection, admission APACHE II score, exposure to sedatives or analgesics, an indwelling catheter and sleep disturbance. Points sum to 0-33 and define low risk at 0-9, moderate risk at 10-17 and high risk at 18-33. Because several predictors can change during admission, the score was conceived as dynamic rather than as a one-off admission measure. Formal completion time was not reported.

Fan and colleagues prospectively enrolled 560 adults across four ICUs and allocated 336 to derivation and 224 to validation. Delirium was assessed repeatedly with the Chinese CAM-ICU. Discrimination was high in the derivation analyses and internal bootstrap checks, and AUROC was .900 in the held-out validation cohort. In the derivation cohort, delirium occurred in 2.8%, 16.8% and 75.9% of the three score groups. Higher risk also tracked mortality and longer hospital stay.

The validation was internal to the same study and healthcare context, not broad external validation across countries or different ICU systems. Predictor definitions, clinical practices and delirium ascertainment may therefore affect transportability. DYNAMIC-ICU may help target prevention and repeated assessment, but a high score is not a diagnosis and a low score cannot rule out delirium.