PURPOSE: No method exists for predicting which child with vesicoureteral reflux (VUR) will have surgery. Our goals were to analyze practice patterns at a major pediatric center and to identify factors predisposing children to surgery for specific indications. MATERIALS AND METHODS: We analyzed a cohort of 3738 children presenting with primary VUR (1996-2005). Surgical indications included: 1) breakthrough urinary tract infection (UTI), 2) non-resolution over 3 years, 3) renal scan abnormality and 4) parent/surgeon preference. Logistic regression was applied to a random 60% subset of children. Validation in the remaining 40% was done using receiver operating characteristic curve analysis and the Hosmer and Lemeshow goodness-of-fit test. RESULTS: Independent predictors of surgery included higher age at presentation, antenatal hydronephrosis (ANH), bilateral VUR and VUR grade. Predictors of surgery for a breakthrough UTI included female gender, increasing age, and bilateral and high-grade VUR. Girls were less likely than boys to be operated for renal scan abnormality or parent/surgeon preference. ANH was a predictor of surgery for decreased function and parent/surgeon preference. The model had fair discrimination (c-statistic = 0.68-0.76) and high calibration (p ≥ 0.24). Probabilities of surgery were calculated. CONCLUSIONS: Higher age at presentation, being followed for ANH, and bilateral and high-grade VUR are independent predictors of VUR-corrective surgery. Predictors of surgery vary with indication. Our methods allow comparison of urological practice patterns and outcomes between institutions by taking into account indications for surgery.
PURPOSE: No method exists for predicting which child with vesicoureteral reflux (VUR) will have surgery. Our goals were to analyze practice patterns at a major pediatric center and to identify factors predisposing children to surgery for specific indications. MATERIALS AND METHODS: We analyzed a cohort of 3738 children presenting with primary VUR (1996-2005). Surgical indications included: 1) breakthrough urinary tract infection (UTI), 2) non-resolution over 3 years, 3) renal scan abnormality and 4) parent/surgeon preference. Logistic regression was applied to a random 60% subset of children. Validation in the remaining 40% was done using receiver operating characteristic curve analysis and the Hosmer and Lemeshow goodness-of-fit test. RESULTS: Independent predictors of surgery included higher age at presentation, antenatal hydronephrosis (ANH), bilateral VUR and VUR grade. Predictors of surgery for a breakthrough UTI included female gender, increasing age, and bilateral and high-grade VUR. Girls were less likely than boys to be operated for renal scan abnormality or parent/surgeon preference. ANH was a predictor of surgery for decreased function and parent/surgeon preference. The model had fair discrimination (c-statistic = 0.68-0.76) and high calibration (p ≥ 0.24). Probabilities of surgery were calculated. CONCLUSIONS: Higher age at presentation, being followed for ANH, and bilateral and high-grade VUR are independent predictors of VUR-corrective surgery. Predictors of surgery vary with indication. Our methods allow comparison of urological practice patterns and outcomes between institutions by taking into account indications for surgery.