Using a Real-Time Risk Prediction Model to Predict Pediatric Venous Thromboembolism (VTE) Events

Overview

The study will evaluate the effectiveness of a novel, real-time risk prediction model for identifying pediatric patients at risk for developing in-hospital blood clots (or venous thromboembolism \[VTE\]) based on data easily extracted from the electronic medical record. The study will assess whether using the risk percentages for developing VTE derived from the model increases the number of high-risk patients screened by the pediatric hematology team, which may may lead to an overall reduction in the number of pediatric VTEs seen at Monroe Carell Jr. Children's Hospital at Vanderbilt.

VTE risk factors in adult hospitalized patients are well established and prevention strategies have been implemented for many years. Unfortunately, VTE prevention guidelines are not well established in children, and the pathophysiology of pediatric VTE is sufficiently different from adults that adult studies cannot be extrapolated to pediatrics. There are no randomized trials in pediatrics to determine whether a risk prediction model helps prevent pediatric VTEs. A risk prediction model was developed that can be applied at admission and updated daily to predict pediatric patients at higher risk for developing a VTE. This model was developed from electronically extracted data from all admissions to the Monroe Carell Jr. Children's Hospital at Vanderbilt from January 1, 2010 to October 31, 2017. Cases were identified based on ICD-9/10 codes. Potential covariates were identified from previous studies and known risk factors for VTE development. The variables with the highest adjusted odds ratio (OR) for developing VTE were history of thrombosis (OR 8.7, 95% confidence interval (CI) 6.6-11.3, p\<0.01), presence of a central venous line (OR 4.9, 95%CI 4.0-5.8, p\<0.01), and cardiology consultation (OR 4.0, 95%CI 3.3-4.8, p\<0.01). Additional significant variables include whether a blood gas was performed, infectious disease consultation, diagnosis of cancer, age, mean corpuscular hemoglobin concentration (MCHC), red cell distribution width (RDW), lactate, and whether surgery was performed. There have been several smaller pediatric VTE risk prediction models that have been developed and published. However, none of these have been evaluated for efficacy in a prospective trial, and none of these studies have used a randomized trial approach to evaluate benefit in identifying pediatric patients at high risk for developing VTE. Therefore, the investigators are performing a randomized, pragmatic trial to evaluate the pediatric VTE risk prediction model and its efficacy at predicting pediatric patients at higher risk for developing a VTE.