The additive impact of the distal ureteral diameter ratio in predicting early breakthrough urinary tract infections in children with vesicoureteral reflux.

ABSTRACT

INTRODUCTION: Many factors influence patient and provider decisions to surgically correct vesicoureteral reflux (VUR), including risk of breakthrough febrile urinary tract infections and likelihood of spontaneous resolution. Ureteral diameter ratio has been shown in several studies to be more predictive than reflux grade with regard to breakthrough urinary tract infection (UTI). We developed and investigated the accuracy of a computational model for predicating febrile breakthrough urinary tract infection within 13 months of starting prophylactic antibiotics in children with VUR. OBJECTIVE: The aim of this study was to validate a model for evaluating the impact of distal ureteral diameter ratio (UDR) in predicting early breakthrough urinary tract infections in children with VUR. STUDY DESIGN: Following a retrospective review, we recorded patient demographics, presenting symptoms, VUR grade, laterality, VUR during filling or voiding, initial bladder volume at the onset of VUR, ureteral duplication, voiding dysfunction, distal ureteral diameter ratio, and number of UTIs prior to VUR diagnosis. NeUROn++, a set of C++ programs, was used to model each data set using logistic regression and neural networks with different architectures. RESULTS: After exclusions, 136 children (93 girls and 43 boys) diagnosed with primary VUR had detailed VCUG and UDR data available. Fourteen children (10.3%) experienced breakthrough febrile UTI events within 13 months of VUR diagnosis. There was a significant association with UDR and breakthrough UTI (p = 0.008). Various computational prediction models for the outcome of breakthrough UTI were developed and evaluated. The computational model that fit best was a model using all variables with an ROC of 0.802. DISCUSSION AND CONCLUSIONS: Clinicians and parents often opt for intervention based on likelihood of spontaneous resolution of VUR as well as clinical course, thereby placing an emphasis on the ability to predict likelihood of breakthrough UTI infections. Our statistical analysis and prediction models further confirm UDR as an important variable predictive of breakthrough UTIs within the first 13 months of beginning prophylactic antibiotics. Furthermore, we developed a neural network model incorporating UDR and grade with an ability to yield the greatest accuracy of any breakthrough UTI predictive calculator to date at 80%.