Venous thromboembolic risk stratification in pediatric trauma: A Pediatric Trauma Society Research Committee multicenter analysis

BACKGROUND Venous thromboembolism (VTE) in injured children is rare, but its consequences are significant. Several risk stratification algorithms for VTE in pediatric trauma exist with little consensus, and all are hindered in development by relying on registry data with known inaccuracies. We performed a multicenter review to evaluate trauma registry fidelity and confirm the effectiveness of one established algorithm across diverse centers. METHODS Local trauma registries at 10 institutions were queried for all patients younger than 18 years admitted between 2009 and 2018. Additional chart review was performed on all "VTE" cases and random non-VTE controls to assess registry errors. Corrected data were then applied to our prediction algorithm using 10 real-time variables (Glasgow Coma Scale, age, sex, intensive care unit admission, transfusion, central line placement, lower extremity/pelvic fracture, major surgery) to calculate VTE risk scores. Contingency table classifiers and the area under a receiver operator characteristic curve were calculated. RESULTS Registries identified 52,524 pediatric trauma patients with 99 episodes of VTE; however, chart review found that 13 cases were misclassified for a corrected total of 86 cases (0.16%). After correction, the algorithm still displayed strong performance in discriminating VTE-fated encounters (sensitivity, 69%; area under the receiver operating characteristic curve, 0.96). Furthermore, despite wide institutional variability in VTE rates (0.04-1.7%), the algorithm maintained a specificity of >91% and a negative predictive value of >99.7% across centers. Chart review also revealed that 54% (n = 45) of VTEs were directly associated with a central line, usually femoral (n = 34, p < 0.001 compared with upper extremity), and that prophylaxis rates were underreported in the registries by about 50%; still, only 19% of the VTE cases had been on prophylaxis before diagnosis. CONCLUSION The VTE prediction algorithm performed well when applied retrospectively across 10 diverse pediatric centers using corrected registry data. These findings can advance initiatives for VTE screening/prophylaxis guidance following pediatric trauma and warrant prospective study.