Construction and verification of risk prediction model of constipation after stroke

MA Shasha; CHEN Xiaoying; LYU Hongxia; SONG Yanfen; LI Xuexin

doi:10.12102/j.issn.2095-8668.2025.20.017

Abstract：Objective: To construct and validate a predictive model for the risk of constipation after stroke. Methods: A convenience sampling method was used to prospectively select 546 patients with stroke who were treated in the Department of Neurosurgery at the Affiliated Hospital of Binzhou Medical University. Among them, 384 cases admitted from August 2021 to October 2022 were assigned to the modeling group, and 162 cases admitted from November 2022 to February 2023 were assigned to the validation group. A nomogram predicting the risk of constipation after stroke was constructed and validated, and the predictive performance of the model was evaluated. Results: In the modeling group of 384 stroke patients, 157 (40.89%) developed constipation. Multivariate logistic regression analysis showed that hospitalization duration, prophylactic use of laxatives, ambulation of ≥30 min/day, nasogastric tube placement method, fluid intake ≥1000 mL/day, urinary tract infection, and hypokalemia were independent risk factors for constipation after stroke (P < 0.05). Based on these factors, a nomogram was constructed. The Hosmer-Lemeshow goodness-of-fit test showed χ² = 7.034, P = 0.533. The concordance index (C-index) was 0.908, and the area under the receiver operating characteristic curve (AUC) was 0.908, with a sensitivity of 90.3%, specificity of 80.3%, and prediction accuracy of 85.2%. In the validation group, the AUC was 0.919, with a sensitivity of 79.3%, specificity of 90.0%, and prediction accuracy of 84.0%. Conclusion: This study constructed a nomogram based on risk factors for constipation after stroke and verified the predictive performance and clinical effectiveness of the model, which is convenient for early and accurate prediction of constipation in stroke patients and timely implementation of targeted interventions.