In Reply We would like to thank Kalakoti and colleagues1 for their insightful comments and careful review of our article, in which we acknowledge a variety of potential sources of bias. We further recognize that a complete case analysis may also introduce bias, and we appreciate the opportunity to clarify our approach to missing data. For most of the variables considered, information was available for all participants. If no indication of an event or injury was present in the National Trauma Data Bank data set, we assumed that it did not occur. This practice is common in other risk models, such as the fracture risk assessment tool developed by the World Health Organization.2 Therefore, complete information on venous thromboembolism (VTE), intubation, intensive care unit admission, transfusion, central venous catheter, pelvic or lower-extremity fracture, major surgery, and patient age was available (only patients who were documented to be younger than 18 years of age were included).
Connelly CR, Laird A, Watters JM. Predicting Venous Thromboembolism in Pediatric Trauma Patients—Reply. JAMA Surg. 2016;151(9):882. doi:10.1001/jamasurg.2016.0482