Risk Factors for Venous Thromboembolism in Pediatric Trauma Patients and Validation of a Novel Scoring System: The Risk of Clots in Kids With Trauma Score.

Yen, Jennifer; Van Arendonk, Kyle J; Streiff, Michael B; McNamara, LeAnn; Stewart, F Dylan; Conner, Kim G; Thompson, Richard E; Haut, Elliott R; Takemoto, Clifford M

OBJECTIVES Identify risk factors for venous thromboembolism and develop venous thromboembolism risk assessment models for pediatric trauma patients. DESIGN Single institution and national registry retrospective cohort studies. SETTING John Hopkins level 1 adult and pediatric trauma center and National Trauma Data Bank. PATIENTS Patients 21 years and younger hospitalized following traumatic injuries at John Hopkins (1987-2011). Patients 21 years and younger in the National Trauma Data Bank (2008-2010 and 2011-2012). INTERVENTIONS None. MEASUREMENTS AND MAIN RESULTS Clinical characteristics of Johns Hopkins patients with and without venous thromboembolism were compared, and multivariable logistic regression analysis was used to identify independent venous thromboembolism risk factors. Weighted risk assessment scoring systems were developed based on these and previously identified factors from National Trauma Data Bank patients (2008-2010); the scoring systems were validated in this cohort from Johns Hopkins and a cohort from the National Trauma Data Bank (2011-2012). Forty-nine of 17,366 pediatric trauma patients (0.28%) were diagnosed with venous thromboembolism after admission to our trauma center. After adjusting for potential confounders, venous thromboembolism was independently associated with older age, surgery, blood transfusion, higher Injury Severity Score, and lower Glasgow Coma Scale score. These and additional factors were identified in 402,329 pediatric patients from the National Trauma Data Bank from 2008 to 2010; independent risk factors from the logistic regression analysis of this National Trauma Data Bank cohort were selected and incorporated into weighted risk assessment scoring systems. Two models were developed and were cross-validated in two separate pediatric trauma cohorts: 1) 282,535 patients in the National Trauma Data Bank from 2011 to 2012 and 2) 17,366 patients from Johns Hopkins. The receiver operating curve using these models in the validation cohorts had area under the curves that ranged 90-94%. CONCLUSIONS Venous thromboembolism is infrequent after trauma in pediatric patients. We developed weighted scoring systems to stratify pediatric trauma patients at risk for venous thromboembolism. These systems may have potential to guide risk-appropriate venous thromboembolism prophylaxis in children after trauma.