The construction of a predictive model for postoperative bladder paralysis in patients with cervical cancer based on logistic regression and nomogram

Objective To construct a predictive model for postoperative bladder paralysis in patients with cervical cancer (CC) based on Logistic regression and nomogram. Methods The clinical data of 220 patients with CC admitted to our hospital from December 2019 to September 2022 were retrospectively collected as the modeling cohort. In addition, the clinical data of 130 patients with CC admitted to the same hospital during a later time period, from October 2022 to September 2025, were collected as the temporal validation cohort. According to whether the patients in the modeling and validation cohorts developed bladder paralysis after surgery, they were separated into the bladder paralysis groups and the non-bladder paralysis groups. Results Logistic regression analysis of the variables selected by LASSO regression showed that clinical staging, number of lymph nodes removed, extent of hysterectomy, width of parametrial tissue resection, and vaginal stump length were risk factors for postoperative bladder paralysis in patients with CC (P < 0.05). The AUC in the modeling cohort was 0.899, and 0.899, and the H-L test yielded χ 2 = 7.405 (P = 0.734). DCA showed a favorable net clinical benefit across threshold probabilities of 0.14–0.82. In the temporal validation cohort,the AUC was 0.944, and the H-L test yielded χ 2 = 7.234 (P = 0.725). DCA showed a favorable net clinical benefit across threshold probabilities of 0.19–0.84. Conclusion Clinical staging, number of lymph nodes removed, extent of hysterectomy, width of parametrial tissue resection, and vaginal stump length were risk factors for postoperative bladder paralysis in patients with CC. The nomogram model constructed based on Logistic regression has good consistency and discrimination.