P.V. Menon1, N. Georgette3, R.C. Keskey2, J. Bruce4, H. Wain4, A. Laytin1, D. Rhee1, I. Nasr1, E.G. Saoud1, M.B. Slidell1, D.L. Clarke4 1Johns Hopkins University School Of Medicine, Pediatric Surgery, Baltimore, MD, USA 2University Of Chicago, Surgery, Chicago, IL, USA 3Boston Children’s Hospital, Emergency Medicine, Boston, MA, USA 4University of KwaZulu-Natal, Pietermaritzburg, South Africa, Pietermaritzburg, KWAZULU-NATAL, South Africa
Introduction:
Low- and middle-income countries (LMICs) bear a disproportionately high burden of injury, with pediatric trauma contributing to 40% of childhood deaths. Accurate trauma scoring is essential for quantifying injury severity and improving care. However, the resource constraints in LMICs necessitate scores based on simple clinical observations. One such score is the Pediatric Shock Index (PSI), derived from age, heart rate, and blood pressure. PSI has demonstrated superior performance over the Shock Index, Pediatric Age-Adjusted (SIPA) in high-income countries like the United States. We hypothesized that the simplicity of PSI would enable it to also outperform SIPA in an LMIC context when the two were compared using data from a middle-income country (MIC) trauma registry.
Methods:
Retrospective study of children (1-18 years) admitted to a single South African trauma center from Dec 2012 to July 2024. We excluded patients with missing or likely spurious emergency department vital signs (HR/SBP of <0.2 or >5), or missing outcome data for mortality. Trauma scores were assigned retrospectively. For bedside use, the simplified “rapid” pediatric shock index (rPSI) equation was also evaluated. We compared PSI and rPSI to SIPA as predictors of in-hospital mortality, blood transfusion, pediatric intensive care unit (PICU) admission, mechanical ventilation, and need for surgery. The sensitivity and specificity were compared using McNemar's test. The PPVs and NPVs were compared using weighted generalized score statistics.
Results:
Of 2636 injured children, 2,232 (85%) met the inclusion criteria. Median age was 8 (IQR: 5, 11), and 68% were male. In-hospital mortality was 1% (n = 23). For all outcomes, PSI achieved the highest specificity followed by rPSI and then SIPA. SIPA achieved the highest sensitivity, which was expected given that the rPSI and PSI cutoffs exceed those of SIPA for most ages. For blood transfusion, PICU admission, mechanical ventilation, and surgery, the positive predictive values (PPV) were higher for PSI and rPSI than SIPA. For all outcomes, negative predictive values (NPV) for both PSI and rPSI were similar to SIPA.
Conclusion:
The PSI and rPSI achieved higher specificity, PPVs, and similar NPVs to SIPA in predicting the need for blood transfusion, PICU admission, mechanical ventilation, surgery, and mortality. None of the scoring systems achieve high enough sensitivity and specificity to be solely applied for clinical decisions. However, the empiric age-based cutoffs of PSI better represent the injured child's initial hemodynamic state. Additionally, the rPSI is quick to apply. With further evaluation, PSI and rPSI may be promising for risk stratification in an LMIC context.