Construction and evaluation of a predictive model for catheter dysfunction in cancer patients with implanted vascular access ports: A cross-sectional study

Background: Catheter dysfunction in implanted vascular access ports (ports) is a prevalent complication disrupting cancer treatment, yet existing evidence lacks risk prediction models specifically for cancer populations and standardized dysfunction assessment. Methods: A cross-sectional study enrolled 413 cancer patients from a Chinese Tertiary A hospital (Jan–Jun 2024). Catheter function was assessed using the Catheter Injection and Aspiration Scheme (CINAS). Univariate/multivariate logistic regression identified predictors, and a nomogram was constructed. Model validation included Hosmer-Lemeshow test, AUC, Cox & Snell R ² , Nagelkerke R ² , and 1,000-bootstrap resampling. Results: Dysfunction incidence was 16.7% (69/413), with 88.4% resolved post-intervention. Three independent predictors emerged: port retention time >12 months (OR=5.105, P<0.001), history of catheter dysfunction (OR=30.672, P<0.001), and lower CPT-SMS self-management score (OR=0.936 per point, P<0.001). The nomogram assigned CPT-SMS the highest weight (53.5%). The model showed excellent discrimination (AUC=0.910, 95% CI:0.879–0.941), calibration (P=0.577), and minimal overfitting (0.002). Conclusions: A predictive model was developed based on retention time, history of catheter dysfunction, and CPT-SMS score to estimate the risk of catheter dysfunction in cancer patients. The validated model exhibited good discrimination and accuracy. Its implementation could support effective patient assessment, enabling early identification and professional nursing interventions to prevent and resolve most catheter dysfunctions.