Diagnostic accuracy of the pediatric CAM-ICU, pre-school CAM-ICU, Pediatric Anesthesia Emergence Delirium and Cornell Assessment of Pediatric Delirium for detecting delirium in the pediatric intensive care unit: A systematic review and meta-analysis

Objectives: Delirium is a frequent complication of critical illness, affecting 34% of children admitted to pediatric intensive care units. The commonly used tools for detecting delirium in the pediatric intensive care unit include the Pediatric Confusion Assessment Method for the intensive care unit (ICU, pCAM-ICU), Preschool Confusion Assessment Method for the ICU (psCAM-ICU), Pediatric Anesthesia Emergence Delirium and the Cornell Assessment of Pediatric Delirium. Data sources: We searched four electronic databases for relevant articles from inception to March 1, 2023. Study selection: All full-text observational studies examining the sensitivity and specificity of the four tools for screening delirium in the pediatric intensive care units were included. Data extraction: Two researchers independently identified articles, extracted data, and retrieved the diagnostic accuracy parameters of the pediatric CAM-ICU, pre-school CAM-ICU, Pediatric Anesthesia Emergence Delirium, and Cornell Assessment of Pediatric Delirium relative to standard references. A bivariate diagnostic statistical analysis with a random-effects model was performed. Data synthesis: Four, five, three and seven studies on the pediatric CAM-ICU, pre-school CAM-ICU, Pediatric Anesthesia Emergence Delirium, and Cornell Assessment of Pediatric Delirium, respectively, were identified. Due to the limited number of Pediatric Anesthesia Emergence Delirium articles included, no pooled diagnostic accuracy was produced. The pooled sensitivity was 0.73, 0.84, and 0.95 for the pediatric CAM-ICU, pre-school CAM-ICU, and Cornell Assessment of Pediatric Delirium, respectively, whereas the pooled specificity was 0.98, 0.90, and 0.81, respectively. The Cornell Assessment of Pediatric Delirium had greater sensitivity compared to both the pediatric CAM-ICU and pre-school CAM-ICU (both p = 0.04) and lower specificity than the pediatric CAM-ICU did (p < 0.001). Age, sample size, and mechanical ventilation use were significant moderators of the specificity of the pediatric CAM-ICU (p < 0.001, <0.001, and = 0.001, respectively). Conclusions: Our data indicate that the Cornell Assessment of Pediatric Delirium is a more dependable instrument than the pediatric CAM-ICU and pre-school CAM-ICU for detecting pediatric intensive care delirium occurrence. More studies on the Pediatric Anesthesia Emergence Delirium are warranted. Implications for clinical practice: Healthcare providers are suggested adopting the Cornell Assessment of Pediatric Delirium into daily routine for the early detection of delirium in pediatric intensive care units.