Objectives: To determine risk factors and outcomes associated with delirium in PICU patients.
Design: Retrospective cohort study.
Setting: Thirty-two-bed PICU within a tertiary care academic children's hospital.
Patients: All children admitted to the PICU March 1, 2014, to October 1, 2016, with at least one Cornell Assessment of Pediatric Delirium score (n = 2,446).
Interventions: None.
Measurements and main results: Cornell Assessment of Pediatric Delirium score was performed twice daily as standard of care. We characterized delirium as: 1) presence of greater than or equal to 1 positive score (Cornell Assessment of Pediatric Delirium ≥ 9) and 2) number of days with a positive score. We built multivariable logistic and linear regression models using electronic medical records data. Many patients (n = 1,538; 63%) had a short length of stay (< 48 hr). Compared with patients with length of stay greater than or equal to 48 hours, fewer experienced delirium (30% vs 69%; p < 0.0001). Among 908 patients with length of stay greater than or equal to 48 hours, presence of delirium was independently associated with age less than 2 years old, baseline cognitive dysfunction, primary diagnosis, and duration of mechanical ventilation. Benzodiazepines demonstrated a dose-response effect (odds ratio for presence of delirium, 1.8 [p = 0.03], 3.4 [p < 0.001], and 9.7 [p = 0.005] for < 25th percentile, 25-75th percentile, and > 75th percentile of total dose, vs no exposure). In terms of outcomes, presence of delirium was independently associated with increased ICU length of stay (p < 0.001), whereas days of delirium were independently associated with decline in cognitive function from ICU admission to discharge (odds ratio, 1.06; p < 0.001), increased ICU (p < 0.001), and hospital length of stay (p < 0.001). Neither delirium presence nor total days were independently associated with mortality.
Conclusions: Delirium is common in the PICU, particularly among patients with length of stay greater than or equal to 48 hours. It is independently associated with patient characteristics and PICU exposures, including benzodiazepines. The role of delirium as an independent causal factor in patient outcome requires further investigation.