BACKGROUND: Pediatric delirium (PD) is a severe neuropsychiatric disorder often seen at the pediatric intensive care unit (PICU). The Pediatric Anesthesia Emergence Delirium (PAED) scale assesses five behavioral items on a five-level severity scale, and is easily applicable in children. However, the five-level severity scales are rather arbitrarily anchored and subjective. This study aimed to pilot a practical and clinical improvement of the PAED by condensing the five-level scales of the five behavioral items to a more objectively anchored two- and three-point scale. METHODS: Post-hoc analysis of routine data in an eight-bed PICU in a tertiary university hospital. 144 critically ill, non-electively admitted patients, aged 1-18 years, were included between November 2006 and February 2010. Scales of the five PAED-items were condensed post-hoc from five to two- and three levels of severity. Five scale properties were analyzed: 1) internal consistency; 2) item-total score correlations; 3) inter-rater agreement; 4) sensitivity and specificity; and 5) discriminative diagnostic ability. RESULTS: Three-level PAED-items post-hoc displayed Cronbach's alpha of 0.86, and mean item-total score correlation was 0.71 (range 0.60 to 0.79). Inter-rater agreement was high (0.90). The most optimal cut-off was 8 (sensitivity=100%, specificity=96.7%) with an area under the curve (AUC) of 0.98. Likelihood ratio for a positive test result (LR+) was 30.3. CONCLUSION: A three-level severity scale for the five PAED-items may be optimal to diagnose PD. Further prospective research is required to determine whether a revised PAED has adequate psychometric properties and is applicable across different clinical settings.
BACKGROUND:Pediatric delirium (PD) is a severe neuropsychiatric disorder often seen at the pediatric intensive care unit (PICU). The Pediatric Anesthesia Emergence Delirium (PAED) scale assesses five behavioral items on a five-level severity scale, and is easily applicable in children. However, the five-level severity scales are rather arbitrarily anchored and subjective. This study aimed to pilot a practical and clinical improvement of the PAED by condensing the five-level scales of the five behavioral items to a more objectively anchored two- and three-point scale. METHODS: Post-hoc analysis of routine data in an eight-bed PICU in a tertiary university hospital. 144 critically ill, non-electively admitted patients, aged 1-18 years, were included between November 2006 and February 2010. Scales of the five PAED-items were condensed post-hoc from five to two- and three levels of severity. Five scale properties were analyzed: 1) internal consistency; 2) item-total score correlations; 3) inter-rater agreement; 4) sensitivity and specificity; and 5) discriminative diagnostic ability. RESULTS: Three-level PAED-items post-hoc displayed Cronbach's alpha of 0.86, and mean item-total score correlation was 0.71 (range 0.60 to 0.79). Inter-rater agreement was high (0.90). The most optimal cut-off was 8 (sensitivity=100%, specificity=96.7%) with an area under the curve (AUC) of 0.98. Likelihood ratio for a positive test result (LR+) was 30.3. CONCLUSION: A three-level severity scale for the five PAED-items may be optimal to diagnose PD. Further prospective research is required to determine whether a revised PAED has adequate psychometric properties and is applicable across different clinical settings.
Authors: Chani Traube; Gabrielle Silver; Julia Kearney; Anita Patel; Thomas M Atkinson; Margaret J Yoon; Sari Halpert; Julie Augenstein; Laura E Sickles; Chunshan Li; Bruce Greenwald Journal: Crit Care Med Date: 2014-03 Impact factor: 7.598