Reliability and validity of a scoring instrument for clinical performance during Pediatric Advanced Life Support simulation scenarios

Abstract

Aim: To assess the reliability and validity of scoring instruments designed to measure clinical performance during simulated resuscitations requiring the use of Pediatric Advanced Life Support (PALS) algorithms.

Methods: Pediatric residents were invited to participate in an educational trial involving simulated resuscitations that employ PALS algorithms. Each subject participated in a session comprised of four scenarios (asystole, dysrhythmia, respiratory arrest, shock). Video-recorded sessions were independently reviewed and scored by four raters using instruments designed to measure performance in terms of timing, sequence, and quality. Validity was assessed by two-factor analysis of variance with postgraduate year (PGY-1 versus PGY-2) as an independent variable. Reliability was assessed by calculation of overall interrater reliability (IRR) as well as a generalizability study to estimate variance components of individual measurement facets (scenarios, raters) and associated interactions.

Results: 20 subjects were scored by four raters. Based on a two-factor ANOVA, PGY-2s outperformed PGY-1s (p<0.05); significant differences in difficulty existed between the four scenarios, with dysrhythmia scores being the lowest. Overall IRR was high (0.81) and most variance could be attributed to subject (17%), scenario (13%), and the interaction between subject and scenario (52%); variance attributable to rater was minimal (1.4%).

Conclusions: The instruments assessed in this study measure clinical performance during PALS scenarios in a reliable and valid manner. Measurement error could be minimized further through the use of additional scenarios but additional raters, for a given scenario, would not improve reliability. Further studies should assess validity of measurement with respect to actual clinical performance during resuscitations.