Establishment and Clinical Application of a Risk Prediction Model for Postoperative Complications of Arm Ports in...

Long-term venous access is essential for administering chemotherapy in gynecological cancers, and the use of arm ports is becoming more common due to their practical advantages. However, postoperative complications remain a significant clinical concern. Currently, investigations specifically predicting the risk of arm port-related complications in gynecological oncology patients are limited. The advances of precision medicine warrant the development of individualized risk prediction tools to optimize clinical decision-making.

This retrospective analysis enrolled 476 patients who underwent arm ports (APs) implantation in the Gynaecology Department, The Fourth Hospital of Hebei Medical University, between July 2022 and October 2024. Patients were randomly divided into a training set (n = 334) and a validation set (n = 142) using computer-generated random numbers. Univariate and multivariate logistic regression analyses were used to identify independent risk factors, and the significant variables were incorporated into a prediction model. A nomogram of the prediction model was generated. Furthermore, the prediction model was internally validated using the Bootstrap method. The model's discriminative performance, calibration, and clinical utility were determined using the receiver operating characteristic (ROC) curve, calibration curve, and decision curve analysis (DCA), respectively.

The overall postoperative complication rate was found to be 12.18% (58/476). Multivariate logistic regression analysis identified preoperative neutrophil-to-lymphocyte ratio (NLR) (odds ratio [OR] = 1.53, 95% confidence interval [CI]: 1.20-1.96), history of ipsilateral arm surgery (OR = 5.02, 95% CI: 2.06-12.23), and planned chemotherapy cycles (OR = 1.52, 95% CI: 1.26-1.85) as independent risk factors of postoperative complications (all p < 0.05). The nomogram prediction model constructed based on these factors demonstrated superior performance, yielding an area under the curve (AUC) of 0.819 (95% CI: 0.756-0.882) in the training set and 0.869 (95% CI: 0.793-0.945) in the validation set. The calibration curve showed good agreement between predicted probabilities and the actual observed incidence of postoperative complications. DCA indicated that the model showed greater clinical benefit than either treating all or treating no patients across a moderate range of threshold probabilities.

The proposed prediction model demonstrates fair to good capability in assessing the risk of postoperative complications following AP implantation in gynecological oncology patients. It supports clinicians in identifying high-risk individuals before surgery and enabling the implementation of targeted preventive approaches.