Project Details

Febrile infection in VesicoUreteral Reflux (FeVUR) study: a multi-institutional risk stratification for urinary tract infections in VUR with machine learning

Adree Khondker, MD

Summary

BACKGROUND: Vesicoureteral reflux (VUR) is a common pediatric condition, predisposing children to urinary tract infections (UTI), urosepsis, and renal insult. Despite guideline recommendations, the decision to observe, provide antibiotic prophylaxis, or offer surgery is heterogenous. GAP: There is a paucity in accurate prediction tools for determining the risk of infection to guide therapy, due to the lack of real-world risk stratification in this population and subjectivity in VUR severity. HYPOTHESIS: We hypothesize clinical risk factors, such as female sex, circumcision status, concurrent voiding dysfunction, are predictive of UTI-risk and that novel metrics for VUR severity are superior to traditional VUR grading. METHODS: The proposed study is a large multi-institutional retrospective cohort study to identify high-risk patients with VUR. The primary outcome with be febrile UTI (as defined by the AAP guidelines) in CAP-naïve children and breakthrough UTI in children on CAP. Secondary outcomes will be symptomatic UTI, renal scarring, and need for surgery. We will collect clinical factors including age, sex, VCUGs (with severity by traditional VUR grade and qVUR), circumcision status, history of UTIs, concurrent bladder/bowel dysfunction, and indications for surgery. Clinical risk models will be developed with regression analyses and machine learning approaches, to determine risk factors and predict a personalized estimate for UTI-risk. RESULTS: Pending. IMPACT: This study will personalize VUR management with more reliable metrics. For instance, if the model predicts a child’s UTI-risk as 5% UTI on CAP despite grade 4 VUR, then guideline-suggested surgery may be unnecessary. Conversely, a child with grade 2 VUR but with other clinical risk factors resulting in an 80% UTI risk should receive CAP or upfront surgery.