Computer model predicting breakthrough febrile urinary tract infection in children with primary vesicoureteral reflux.

INTRODUCTION AND OBJECTIVE: Factors influencing the decision to surgically correct vesicoureteral reflux (VUR) include risk of breakthrough febrile urinary tract infection (fUTI) or renal scarring, and decreased likelihood of spontaneous resolution. Improved identification of children at risk for recurrent fUTI may impact management decisions, and allow for more individualized VUR management. We have developed and investigated the accuracy of a multivariable computational model to predict probability of breakthrough fUTI in children with primary VUR.

STUDY DESIGN: Children with primary VUR and detailed clinical and voiding cystourethrogram (VCUG) data were identified. Patient demographics, VCUG findings including grade, laterality, and bladder volume at onset of VUR, UTI history, presence of bladder-bowel dysfunction (BBD), and breakthrough fUTI were assessed. The VCUG dataset was randomized into a training set of 288 with a separate representational cross-validation set of 96. Various model types and architectures were investigated using neUROn++, a set of C++ programs.

RESULTS: Two hundred fifty-five children (208 girls, 47 boys) diagnosed with primary VUR at a mean age of 3.1 years (±2.6) met all inclusion criteria. A total 384 VCUGs were analyzed. Median follow-up was 24 months (interquartile range 12-52 months). Sixty-eight children (26.7%) experienced 90 breakthrough fUTI events. Dilating VUR, reflux occurring at low bladder volumes, BBD, and history of multiple infections/fUTI were associated with breakthrough fUTI (Table). A 2-hidden node neural network model had the best fit with a receiver operating characteristic curve area of 0.755 for predicting breakthrough fUTI.

DISCUSSION: The risk of recurrent febrile infections, renal parenchymal scarring, and likelihood of spontaneous resolution, as well as parental preference all influence management of primary VUR. The genesis of UTI is multifactorial, making precise prediction of an individual child's risk of breakthrough fUTI challenging. Demonstrated risk factors for UTI include age, gender, VUR grade, reflux at low bladder volume, BBD, and UTI history. We developed a prognostic calculator using a multivariable model with 76% accuracy that can be deployed for availability on the Internet, allowing input variables to be entered to calculate the odds of an individual child developing a breakthrough fUTI.

CONCLUSIONS: A computational model using multiple variables including bladder volume at onset of VUR provides individualized prediction of children at risk for breakthrough fUTI. A web-based prognostic calculator based on this model will provide a useful tool for assessing personalized risk of breakthrough fUTI in children with primary VUR.