Identification of risk factors and construction of a nomogram for predicting postoperative stone recurrence in patients with infective upper urinary tract stones

Infection-related upper urinary tract stones (struvite and/or carbonate apatite) are prone to early postoperative recurrence. This study aimed to identify predictors of 6-month recurrence and develop a pragmatic risk-prediction nomogram to support individualized follow-up. A retrospective single-center study was conducted in 148 patients with infection-related stones confirmed by infrared spectroscopy who underwent surgical removal between January 2019 and December 2023. Patients were randomly allocated to a modeling cohort (n = 104) and a validation cohort (n = 44) using a 7:3 non-stratified split (internal split-sample validation only). Candidate predictors and outcomes were complete for all patients, and no imputation was required. Recurrence within 6 months was defined as new stone formation or a > 2-fold increase in residual fragment size on ultrasound or plain computed tomography (CT). Multivariable logistic regression was used to identify independent predictors and to construct a nomogram. Model performance was assessed using discrimination, calibration, and decision curve analysis. In the modeling group, the stone recurrence rate was 44.2% (n = 46/104), while in the validation group it was 70.5% (n = 31/44). Compared to non-recurrence cases, patients in the recurrence group had significantly higher rates of residual stones (72% vs. 22%, P < 0.001), renal atrophy (33% vs. 9%, P = 0.002), and positive urine cultures at six-month follow-up (39% vs. 3%, P = 0.008), as well as higher median CT values 821 HU (IQR 592–1003) vs. 511 HU (IQR 416–864), P = 0.004. Multivariate logistic regression identified residual stones (OR = 6.19, 95% CI: 2.12–18.08, P = 0.001), positive urine culture during follow-up (OR = 9.71, 95% CI: 1.73–54.41, P = 0.010), and CT value (OR = 1.02, 95% CI: 1.00–1.04, P = 0.026) as independent predictors of recurrence. A nomogram constructed using these three factors showed strong predictive performance, with an area under curve of 0.86 (95% CI: 0.77–0.93) in the modeling group and 0.75 (95% CI: 0.59–0.91) in the validation group. Calibration curves demonstrated good agreement between predicted and observed recurrence risk, and decision curve analysis indicated favorable clinical utility of the model. Residual stones, positive follow-up urine culture, and higher CT attenuation values were identified as independent predictors of early recurrence after surgery for infection-related upper urinary tract stones. The nomogram showed acceptable internal performance but requires external validation.