Risk Factors and Prediction of Chronic Postsurgical Pain Among Patients with Distal Lower fracture: Cohort Analysis

Background: The surgical interventions aimed at fracture repair can paradoxically lead to chronic postoperative pain (CPSP), which is associated with depression, impaired quality of life, and increased societal burden. This phenomenon is particularly understudied in young patients with distal lower extremity fracture. Developing a scalable and accurate predictive model could revolutionize postoperative care by enabling early detection of high-risk patients and guiding personalized pain management strategies. Methods: This study collected in-hospital medical records and conducted follow-up for all patients over a one-year period. We developed a predictive model through a three-stage approach involving Least Absolute Shrinkage and Selection Operator (LASSO) regression, information gain analysis, and multivariate logistic regression, followed by model validation. Using the Shinyapps.io platform to build a webpage risk calculator for the final prediction model. Results: The final cohort included 818 patients: 38.39% of whom experienced CPSP, and 18.15% experienced neuropathic pain. There are six independent variables associated with CPSP: postoperative analgesic technique, fixation type, preoperative clinical management, and NRS score on the day of the visit and postoperative day 1. The optimism-corrected area under the receiver operating curve for the development cohort and validation cohort were 0.872 and 0.838, respectively and this model demonstrated good calibration and clinical utility. A web-based predictive nomogram was established by integrating machine learning-based big data variable screening with the interpretability of traditional logistic regression. Conclusions: This study demonstrates that pain management strategies, surgical approaches, and patient psychological factors collectively influence the development of CPSP. By integrating machine learning-based big data variable screening with the interpretability of traditional logistic regression, we developed a web-based predictive nomogram capable of identifying early CPSP risk at hospital discharge, thereby improving accessibility to transitional pain care interventions.