Risk Stratification for ESBL-Producing Enterobacterales in Elderly Diabetic Patients with Urinary Tract Infections: A Multicenter Model to Support Empirical Antibiotic Decision-Making

Tingting Huang,1 Jibao Qin,2 Xi Jiang,1 Ming Hu,1 Jiaping Wang,1 Kaiyong Chen,3 Fumeng Yang,4 Jin Gu,5 Huiyi Wu,1 Xinkuan Chen6 1Department of Laboratory Medicine, Donghai Hospital Affiliated to Kangda College of Nanjing Medical University / Donghai County People’s Hospital, Lianyungang, Jiangsu, People’s Republic of China; 2Department of Laboratory Medicine, Dongfang Hospital Affiliated to Kangda College of Nanjing Medical University, Lianyungang, Jiangsu, People’s Republic of China; 3Department of Laboratory Medicine, Guanyun County People’s Hospital, Lianyungang, Jiangsu, People’s Republic of China; 4Department of Laboratory Medicine, Lianyungang Second People’s Hospital Affiliated to Kangda College of Nanjing Medical University, Lianyungang, Jiangsu, People’s Republic of China; 5Department of Laboratory Medicine, Lianyungang Hospital Affiliated to Nanjing University of Traditional Chinese Medicine, Lianyungang, Jiangsu, People’s Republic of China; 6Department of Laboratory Medicine, First Hospital Affiliated to Kangda College of Nanjing Medical University / The First People’s Hospital of Lianyungang, Lianyungang, Jiangsu, People’s Republic of ChinaCorrespondence: Xinkuan Chen, Email 18961328118@189.cnObjective: Antimicrobial resistance among elderly diabetic patients with urinary tract infections (UTIs) poses a significant challenge for empirical antibiotic therapy. Delayed availability of microbiological susceptibility results often leads to treatment mismatch and inappropriate broad-spectrum antibiotic use. This study aimed to develop and validate a clinically interpretable risk stratification model to estimate the probability of extended-spectrum β-lactamase (ESBL)-producing Enterobacterales isolation prior to microbiological confirmation, thereby supporting early empirical antibiotic decision-making.Methods: We conducted a multicenter retrospective cohort study including elderly (≥ 60 years) patients with type 2 diabetes and positive urine cultures. Multivariable logistic regression was used to construct a prediction model for ESBL-positive isolation. Model discrimination and calibration were evaluated using the area under the receiver operating characteristic curve (AUC), Brier score, calibration plots, and bootstrap internal validation. Internal–external cross-validation and independent external validation were performed to assess model transportability. Decision curve analysis (DCA) was applied to evaluate clinical net benefit across threshold probabilities.Results: A total of 612 patients were included, of whom 364 (58.6%) had ESBL-positive isolates. Independent predictors included diabetes duration ≥ 10 years, HbA1c ≥ 8.5%, recent antibiotic exposure, urinary tract device use, and low-level pyuria (< 5 WBC/HPF). The model demonstrated stable discrimination (AUC 0.81 in the training set; 0.79 in internal validation; 0.81 in external validation) and good calibration (Brier score 0.18– 0.19). Decision curve analysis showed meaningful clinical net benefit across a wide range of threshold probabilities (10%– 65%). Exploratory analysis indicated that empirical therapy mismatch was associated with prolonged hospitalization and increased sepsis incidence, underscoring the potential clinical relevance of early resistance risk identification.Conclusion: This multicenter risk stratification model provides a practical tool for early estimation of ESBL-producing Enterobacterales risk in elderly diabetic patients with UTIs. By integrating routinely available clinical and laboratory variables, the model may support antimicrobial stewardship efforts and improve empirical antibiotic decision-making before susceptibility results become available. Prospective validation in diverse healthcare settings is warranted to confirm its clinical impact.Keywords: diabetes, bacteriuria, positive urine culture, ESBL, enterobacterales, nomogram, risk prediction