Derivation and validation of an improved pediatric shock index for predicting need for early intervention and outcomes in pediatric trauma.

BACKGROUND:Shock index, pediatric age adjusted (SIPA), has been widely applied in pediatric trauma but has limited precision because of the reference ranges used in its derivation. We hypothesized that a pediatric shock index (PSI) equation based on age-based vital signs would outperform SIPA. METHODS:A retrospective cohort of trauma patients aged 1 to 18 years from Trauma Quality Programs - Participant Use File 2010 to 2018 was performed. A random 70% training subset was used to derive Youden index-optimizing shock index (SI) cutoffs by age for blood transfusion within 4 hours. We used linear regression to derive equations representing the PSI cutoff for children 12 years or younger and 13 years or older. For children 13 years or older, the well-established SI of 0.9 remained optimal, consistent with SIPA and other indices. For children 12 years or younger in the 30% validation subset, we compared our age-based PSI to SIPA as predictors of early transfusion, mortality, pediatric intensive care unit admission, and injury severity score of ≥25. For bedside use, a simplified "rapid" pediatric shock index (rPSI) equation was also derived and compared with SIPA. RESULTS:A total of 439,699 patients aged 1 to 12 years met the inclusion criteria with 2,718 (1.3% of those with available outcome data) requiring transfusion within 4 hours of presentation. In the validation set, positive predictive values for early transfusion were higher for PSI (8.3%; 95% confidence interval [CI], 7.5-9.1%) and rPSI (6.3%; 95% CI, 5.7-6.9%) than SIPA (4.3%; 95% CI, 3.9-4.7%). For early transfusion, negative predictive values for both PSI (99.3%; 95% CI, 99.2-99.3%) and rPSI (99.3%; 95% CI, 99.2-99.4%) were similar to SIPA (99.4%; 95% CI, 99.3-99.4%). CONCLUSION:We derived the PSI and rPSI for use in pediatric trauma using empiric, age-based SI cutoffs. The PSI and rPSI achieved higher positive predictive values and similar negative predictive values to SIPA in predicting the need for early blood transfusion and mortality. LEVEL OF EVIDENCE:Prognostic/Epidemiological; level III.