Development and validation of a nomogram model for predicting prognosis in conservatively treated patients with lateral epicondylitis

Lateral epicondylitis is a common cause of elbow pain; however, a small proportion of patients fail to respond to conservative treatment. Early prediction of these outcomes is crucial for personalized management, yet existing models often lack standardized imaging parameters. This study aimed to develop and validate a nomogram model to predict the failure of conservative treatment of lateral epicondylitis. A total of 500 patients with lateral epicondylitis, categorized into those who failed conservative management (operative group, n = 145) and those who responded successfully (conservative group, n = 355), were retrospectively enrolled. The cohort was partitioned into a training set ( n = 301), an internal validation set ( n = 130), and an external validation set ( n = 69). The clinical and magnetic resonance imaging (MRI) variables were performed univariable and multivariable regression to build Model 1. Two different MRI classifications for assessing lesion signal change were then added to Model 1 to form Models 2 and 3. Area under the curve (AUC), calibration curves, and decision curve analysis (DCA) were used to assess the predictive power, accuracy, and clinical decision-making ability of the models. The sex, duration of pain, size of common extensor tendon (CET) tear, and muscle oedema are associated with the failure of conservative treatment. Model 3 of the external validation set has a significantly higher AUC than Model 1( P = 0.046). The AUC of Model 3 in the training set, internal validation set, and external validation set was 0.848 [95% CI: 0.798–0.898], 0.809[95% CI: 0.725–0.893], and 0.845[95% CI: 0.739–0.950]. The calibration curve and DCA indicated that Model 3 had high stability and clinical utility. We developed and validated a model based on clinical and MRI features to stratify patients at the highest risk of conservative treatment failure, aiding in personalized surgical decision-making. In addition, we identified new MRI classifications that significantly improved the predictive power of the model. Further large-scale, multi-center external validation studies are warranted to confirm the generalizability and clinical utility of these findings.