1424: PERFORMANCE OF CORNELL ASSESSMENT OF PEDIATRIC DELIRIUM SCALE IN MECHANICALLY VENTILATED CHILDREN

Abstract

Learning Objectives: Delirium is a common problem in the pediatric intensive care unit (PICU). Accurate diagnosis is limited by difficulty in communicating with mechanically ventilated patients of differing developmental abilities. The purpose of this study was to examine the performance of Cornell Assessment of Pediatric Delirium (CAPD) scale in mechanically ventilated children. Methods: This is a single-center, prospective study of all patients admitted to the PICU between August 2017 and February 2018. Patients who required sedation and mechanical ventilation for >48 hours and received delirium assessment by CAPD scale were included in the study. Each CAPD assessment was followed by clinical assessment by a physician. All CAPD scores were compared with clinical diagnosis of delirium (considered gold standard). Sensitivity and specificity of the scale were calculated for overall population and also stratified based on presence or absence of developmental delay. To improve accuracy of the CAPD scale, we excluded 4 questions which are not directly linked to DSM criteria for delirium diagnosis. To define the best sensitivity and specificity of the tool, univariate logistic regression was performed with receiver operating characteristic (ROC) curves. Results: 837 paired assessments were completed for 95 mechanically ventilated patients. Delirium prevalence was 19%. Overall, CAPD score >9 had a sensitivity of 81.8% and specificity of 44.8%. PPV and NPV were 25.8% and 91.3% respectively. On analysis, area under the ROC curve was 0.66 with the best sensitivity and specificity at a score of 11 for overall data (73.6% and 54.9% respectively). Although patients with normal development had best sensitivity and specificity at a score of 9, patients with developmental delay had the best sensitivity and specificity at a score of 17 (74.4% and 63.2% respectively). By excluding 4 nonDSM questions, the sensitivity and specificity were much higher for both developmentally delayed (83.7% and 47.4% respectively) and normally developed children (81.9% and 70.5% respectively). Conclusions: This study shows that there was a much higher false positive rate of delirium diagnosis using CAPD in patients with developmental delay. Different cut-off points for patients with and without developmental delay may be needed and exclusion of non-DSM questions may improve the high false positive rate. This highlights the need for a new delirium screening tool for critically ill children, particularly for those who are mechanically ventilated.