Development of a calibrated logistic regression clinical risk score for predicting postoperative sepsis following retrograde intrarenal surgery
World Journal of Urology, cilt.44, sa.1, 2026 (SCI-Expanded, Scopus)
- Yayın Türü: Makale / Tam Makale
- Cilt numarası: 44 Sayı: 1
- Basım Tarihi: 2026
- Doi Numarası: 10.1007/s00345-026-06393-2
- Dergi Adı: World Journal of Urology
- Derginin Tarandığı İndeksler: Science Citation Index Expanded (SCI-EXPANDED), Scopus, BIOSIS, EMBASE, Gender Studies Database, MEDLINE, Academic Search Ultimate (EBSCO), Biomedical Reference Collection: Corporate Edition (EBSCO), Health Research Premium Collection (ProQuest), Pharma Collection (ProQuest)
- Anahtar Kelimeler: Machine learning, Postoperative sepsis, Retrograde intrarenal surgery (RIRS), Risk stratification, Urosepsis
- Sağlık Bilimleri Üniversitesi Adresli: Evet
Özet
Objective: To identify perioperative predictors of postoperative sepsis following retrograde intrarenal surgery (RIRS) and to develop a calibrated logistic regression–based clinical risk score for early clinical stratification. Methods: This retrospective study included 430 patients undergoing RIRS for renal stones between January 2021 and May 2025. Preoperative evaluation consisted of demographic, clinical, laboratory, and radiological assessments. Postoperative sepsis was defined according to Sepsis-2 criteria. Six clinically relevant variables were incorporated into a penalized logistic regression model: female sex, diabetes mellitus, recurrent urinary tract infection history, urinary nitrite positivity, residual stone fragments, and stone volume. Model performance was evaluated using cross-validated discrimination and calibration metrics. A Platt-scaled version of the model was generated to improve probability calibration, and SHAP analysis was used to assess feature contributions. An integer-based clinical scoring system was derived from calibrated coefficients. Results: Postoperative sepsis occurred in 20 patients (4.7%). All six variables were independently associated with sepsis. The calibrated model demonstrated strong performance (AUC 0.854) with markedly improved calibration compared with the uncalibrated version. SHAP analysis confirmed residual fragments, female sex, diabetes mellitus, and urinary nitrite as dominant contributors. The derived scoring system stratified patients into low (2.8% observed risk), intermediate (7.9%), and high-risk groups (15.9%). Decision curve analysis indicated superior net benefit over treat-all and treat-none strategies across clinically relevant thresholds. Conclusion: A calibrated, interpretable risk model using six readily obtainable perioperative variables accurately estimated sepsis risk after RIRS and enabled practical stratification into clinically meaningful risk categories. This tool may support targeted perioperative management and early postoperative monitoring. External validation is needed prior to routine implementation.