A new prediction rule can help identify critically ill patients at high risk for delirium.
To create a prediction rule for the development of delirium during intensive care unit (ICU) stays, researchers assessed predictors among 1613 adult medical and surgical ICU patients at a single institution. They validated the prediction rule in 1443 other patients at the same hospital and at four other institutions in the Netherlands. They also compared the rule’s performance against prediction of delirium by ICU nurses and physicians. To identify delirium, ICU nurses assessed patients three times daily using a standard, valid assessment tool with good interrater reliability.
Delirium developed at a rate of 30% in the cohort as a whole. The prediction rule consisted of 10 characteristics readily available on admission to the ICU. The rule stratified patients into four groups: low risk (0%–20% risk for delirium), moderate risk (>20%–40%), high risk (>40%–60%), and very high risk (>60%). It performed much better than prediction by ICU nurses and physicians. The characteristics with the biggest impact were older age, coma, admission for a neurological or neurosurgical condition, and the use of sedation.
Comment: This study was well-designed and well-executed. The rate of delirium in this cohort was lower than rates in other studies of adult ICU patients (typically, 60%–80%); such high rates generally obviate the need for a prediction rule (JAMA 2004; 291:1753). A lower incidence may have been observed here because not all the patients in this study required mechanical ventilation and because fewer than half received sedation. Testing this prediction rule in other centers and countries will help clarify this discrepancy.
The rule has three major drawbacks. First, it requires a computer interface to be calculated in clinical practice. Second, its negative predictive value is poor. In the lowest risk group, up to 20% of patients still develop delirium. Third, dementia and alcohol misuse were not included in the rule because the incidence of these two risk factors was low among patients in the study; these patients should be considered at high risk for delirium regardless of other predictors. Nevertheless, application of this prediction rule has the potential to augment clinical judgment in identifying ICU patients at highest risk for delirium, for whom preventive measures may be appropriate: e.g., reorientation, early mobilization, maintenance of sleep–wake cycles, cognitive stimulation, and preferential use of dexmedetomidine for sedation.
Source: Journal Watch Neurology