The Validity and Reliability of the Japanese Version of the Cornell Assessment of Pediatric Delirium

Abstract

Objectives: 

To examine validity and reliability of the Japanese version of the Cornell Assessment of Pediatric Delirium.

Design: 

Prospective double-blinded observational cohort study.

Setting: 

Eight-bed mixed PICU (post-surgical and internal medicine) from May 2017 to June 2018.

Patients: 

All children between the ages of 0–13 years who were admitted to the PICU for at least 24 hours were eligible for inclusion, as long as the child was arousable to verbal stimulation.

Interventions: 

Two nurses simultaneously and independently assessed each patient for pediatric delirium with the Japanese version of the Cornell Assessment of Pediatric Delirium; this was compared to the gold standard of psychiatric diagnosis based on Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition standards.

Measurements and Main Results: 

Forty-one children were enrolled and 92 assessments (ranging from one to four per subject) were completed. Congenital heart disease patients accounted for 73 enrollees (79%). Forty-three percent of observations were performed in children on invasive mechanical ventilation. Pediatric delirium prevalence (as determined by psychiatric diagnosis) was 53%. The Japanese version of the Cornell Assessment of Pediatric Delirium demonstrated an optimal scoring cutpoint of 9. Overall, area under the curve was 92%, sensitivity 90% (95% CI, 79–97%), specificity 88% (95% CI, 75–96%), positive predictive value of 90% (95% CI, 79–97%), negative predictive value of 91% (95% CI, 80–97%), and a Cohen’s κ of 0.89 (95% CI, 0.8–0.98). In children on invasive mechanical ventilation, the Japanese version of the Cornell Assessment of Pediatric Delirium maintained an area under the curve 87%, sensitivity 97%, and specificity 64%.

Conclusions: 

The Japanese version of the Cornell Assessment of Pediatric Delirium is a valid and reliable tool for use in Japanese PICUs. This will allow for detection of delirium in real-time and may lead to better identification of the population and risk factors for appropriate management and therapeutic and preventative interventions.