Development and validation of a nomogram for predicting the likelihood of undergoing surgical intervention in ruptured corpus luteum

OBJECTIVE: To investigate clinical factors associated with surgical intervention in patients with ruptured corpus luteum (CL) and to develop a prediction model for estimating the likelihood of undergoing surgical intervention. METHODS: This retrospective study enrolled 341 patients with ruptured CL, who were allocated to surgical (n = 106) or conservative (n = 235) group. Baseline demographic, laboratory, and imaging data were collected and compared. Multivariable logistic regression analysis was employed to identify independently associated factors, which were subsequently used to develop a nomogram prediction model. Model performance was evaluated using the area under the receiver operating characteristic curve (AUC) for discriminative ability, calibration curves for accuracy, and decision curve analysis (DCA) for clinical utility. RESULTS: Abdominal rebound tenderness (OR = 1.363, 95% CI: 1.078-1.724), lower admission hemoglobin level (OR = 0.891, 95% CI: 0.862-0.921), longer prothrombin time (OR = 1.445, 95% CI: 1.019-2.048), larger maximum CL cyst diameter (OR = 2.340, 95% CI: 1.821-3.006), and greater pelvic fluid depth (OR = 1.482, 95% CI: 1.154-1.903) were independent factors associated with undergoing surgical intervention. The nomogram demonstrated good discriminative ability (AUC = 0.884, 95% CI: 0.847-0.921), with a bias-corrected AUC of 0.876 after Bootstrap validation. The calibration curve indicated good agreement between predicted probabilities and actual observations. DCA showed a favorable net clinical benefit across a wide range of threshold probabilities. CONCLUSION: This study developed a nomogram incorporating five independent predictors: abdominal rebound tenderness, admission hemoglobin, prothrombin time, maximum corpus luteum cyst diameter, and pelvic fluid depth. Although internal validation demonstrated good discrimination (AUC = 0.884) and potential clinical utility, inherent treatment-selection bias renders this model strictly exploratory. Pending prospective external validation, it serves merely as a preliminary decision-support reference.