# Development of a prediction model for urinary tract infection risk after open reimplantation in children with primary unilateral vesicoureteral reflux: A multicentre study

## TL;DR

This study created a model to predict febrile UTI risk in children after VUR surgery, using factors like surgery time and antibiotic use.

## Contribution

A novel nomogram was developed to predict postoperative febrile UTI risk in children with unilateral VUR.

## Key findings

- The 3-year cumulative incidence of postoperative febrile UTI was 27.2%.
- Operative time and prior injection were significant predictors of UTI risk.
- The nomogram showed good performance with a C-index of 0.743.

## Abstract

To develop a predictive model for symptomatic postoperative febrile urinary tract infections (UTIs) in children undergoing open reimplantation for vesicoureteral reflux (VUR) and evaluate the association with VUR recurrence.

This multicentre retrospective study included children with unilateral VUR (grades III–V) who underwent open Cohen or Lich‐Gregoir reimplantation (2010–2022), had recurrent febrile UTIs, and ≥1 year follow‐up. Analyses used 10‐pooled multiple imputation, with complete case for sensitivity. Full and Least Absolute Shrinkage and Selection Operator (LASSO) Weibull regression models with centre clustering, bootstrapping, and 10‐fold cross‐validation identified predictors. Prediction used demographic, clinical, procedural and antibiotic factors. The non‐scaled LASSO model informed the nomogram, evaluated using C‐indices, calibration and decision curve analysis (DCA). UTI and VUR recurrence were analysed via cumulative incidence.

A total of 404 children (median age 8 (6–9) years; follow‐up 2.3 (2.0–3.3) years, 233 complete‐case) were analysed. Median preoperative febrile UTIs were four, 74.5% had antibiotic resistance and median postoperative prophylaxis was two days. The 3‐year cumulative incidence of postoperative UTI was 27.2% (95% CI: 22.9–31.6). LASSO‐significant predictors included operative time (HR 1.10, 95% CI 1.03–1.16); in sensitivity analyses, prior injection (HR 2.08, 95% CI 1.88–2.30) and postoperative antibiotic duration (HR 0.81, 95% CI 0.69–0.97) were also significant. The nomogram included preoperative fever, antibiogram resistance, renal defect, VUR phase, prior injection, surgical indication, catheterization, hospitalization and stenting. The model performed well (C‐indices = 0.743; calibration slope = 1), with DCA supporting clinical utility for 10–40% predicted risk. Recurrent VUR grade ≥II after 12 months (3.3%–12.7% at 1–3 years, n = 273) did not increase UTI risk.

Children with unilateral dilating VUR remained at risk of postoperative febrile UTIs. The nomogram can assist in identifying high‐risk children for targeted interventions, but requires external validation and refinement.