Construction of Lasso-Logistic prediction model for urinary tract infection after transurethral resection of bladder tumor

Objective To analyze the occurrence of urinary tract infection (UTI) after transurethral resection of bladder tumor and construct a Lasso-Logistic prediction model.

Methods A total of 920 bladder cancer patients with transurethral resection of bladder tumor in Beijing Friendship Hospital Affiliated to Capital Medical University from May 2022 to October 2023 were selected, and the incidence of postoperative UTI was recorded. Patients were divided into UTI and non-UTI groups based on the occurrence of UTI, and clinical materials were compared between the two groups. Lasso-Logistic regression analysis was used to identify the influencing factors of postoperative UTI in bladder cancer patients, and a Lasso-Logistic prediction model was constructed based on these factors. The prediction performance and clinical utility of the model were evaluated through the receiver operating characteristic (ROC) curve and decision curve analysis (DCA).

Results The incidence of UTI during hospitalization after transurethral resection of bladder tumor for bladder cancer patients was 12.50% (115/920). Lasso-Logistic regression analysis revealed that age, hypertension, diabetes, serum procalcitonin (PCT), interleukin-6 (IL-6), C-reactive protein (CRP), peripheral blood CD3⁺, CD4⁺/CD8⁺, immunoglobulin A (IgA), immunoglobulin M (IgM), urine matrix metalloproteinase-7 (MMP-7), surfactant protein A (SP-A), and surfactant protein D (SP-D) were independent influencing factors for postoperative UTI in bladder cancer patients (P < 0.05). The Lasso-Logistic prediction model was constructed based on above factors: Logit(P)=-2.516+1.109×age+1.002×diabetes+1.359×hypertension+1.496×CRP+1.726×PCT+1.562×IL-6-1.155×CD3⁺-1.280×CD4⁺/CD8⁺-1.032×IgA-1.411×IgM+1.589×MMP-7-0.843×SP-A-0.799×SP-D. The ROC curve analysis showed that the area under the curve (AUC) of the model for predicting postoperative UTI in bladder cancer patients was 0.944 (95%CI, 0.927 to 0.958), with sensitivity and specificity of 87.83% and 85.22% respectively. DCA results indicated that the model had significant positive net benefits.

Conclusion The incidence of UTI after transurethral resection of bladder tumor in bladder cancer patients is relatively high. The construction of a Lasso-Logistic prediction model based on influencing factors can provide a reliable reference for clinical prediction of UTI risk.