DOI: 10.29188/2222-8543-2026-19-1-130-139
For citation:
Pavlov V.N., Vorobyov V.A., Ananyev V.A., Shcherbatykh A.V., Lelyavin K.B., Iosifov D.A., Kiselev K.V. Building a prognostic outcome model for severe pyelonephritis. Experimental and Clinical Urology 2026;19(1):130-139; https://doi.org/10.29188/2222-8543-2026-19-1-130-139
Pavlov V.N., Vorobyov V.A., Ananyev V.A., Shcherbatykh A.V., Lelyavin K.B., Iosifov D.A., Kiselev K.V.
Information about authors:
- Pavlov V.N. – Dr. Sci., Professor, Academician of the Russian Academy of Sciences, Head of the Department of Urology, Rector of the Bashkir State Medical University of the Ministry of Health of the Russian Federation, Ufa, Russia; https://orcid.org/0000-0003-2125-4897
- Vorobev V.A. – Dr. Sci., Professor of the Department of Faculty Surgery and Urology at the Irkutsk State Medical University of the Ministry of Health of the Russian Federation, Irkutsk, Associate Professor of the Department of Urology and Oncology at the Bashkir State Medical University of the Ministry of Health of the Russian Federation, Ufa, Russia; https://orcid.org/0000-0003-3285-5559
- Ananiev V.A. – PhD, Head of the Urology Department No 2, Regional Clinical Hospital, Barnaul, Russia; https://orcid.org/0000-0002-1636-3151
- Shcherbatykh A.V. – Dr. Sci., Professor, Head of the Department of Faculty Surgery and Urology, Rector of the Irkutsk State Medical University, Irkutsk, Russia; https://orcid.org/0000-0003-1990-1207
- Lelyavin K.B. – Dr. Sci., Associate Professor of the Department of Emergency Medical Care and Disaster Medicine at the Irkutsk State Medical Academy of Postgraduate Education – a branch of the Federal State Budgetary Educational Institution of Additional Professional Education Russian Medical Academy of Continuous Professional Education, Irkutsk, Russia; https://orcid.org/0000-0001-9278-9739
- Iosifov D.A. – 2-year resident specializing in urology, Bratsk City Hospital No 1, Bratsk, Russia; https://orcid.org/0009-0005-0026-2144
- Kiselev K.V. – 2-year resident specializing in urology, Bratsk City Hospital No 3, Bratsk, Russia; https://orcid.org/0009-0007-7090-2172
12
Introduction. Severe acute pyelonephritis (SAP) is frequently complicated by systemic infection and carries a considerable risk of in-hospital death. Robust prognostic markers and early risk stratification tools are urgently needed.
Objective. To identify factors associated with in-hospital mortality in SAP, estimate survival using Kaplan–Meier analysis, and develop an out-come-prediction model based on neural-network logistic analysis.
Materials and methods. We retrospectively reviewed 67 consecutive patients (mean age 55±17 years; 65.7% female) admitted to a single urology department with destructive SAP. Demographic, clinical, laboratory and therapeutic variables, complications and outcomes were extracted from medical records. Statistical work-up comprised descriptive statistics, χ² and Mann–Whitney tests, multivariate logistic regression, Kaplan–Meier survival curves with log-rank test, and construction of an artificial neural network (six input features, one hidden layer) for individual outcome prediction.
Results. Overall in-hospital mortality was 19.4%. Independent predictors of death were age > 60 years (OR 5.1; p=0.03), male sex (OR 4.0; p=0.04) and severe clinical status at admission (OR 6.8; p<0.01); sepsis increased the risk 7.5-fold (p<0.001). The Kaplan–Meier curve revealed a steep decline in survival during the first 12 days, reaching a plateau of 66 % thereafter. Patients > 60 years exhibited significantly lower 14-day survival than younger individuals (57% vs 85%; p=0.02). The neural-network model achieved an AUC of 0.89, accuracy 85%, sensitivity 77% and specificity 90%. Age, baseline severity and leukocytosis were the most influential features.
Conclusions. Advanced age, male sex and systemic infectious complications are key determinants of mortality in SAP. Combining classical biostatistics with machine-learning techniques enables accurate early outcome prediction and may optimise triage and management of high-risk patients.
