Establishment and validation of a predictive model for peripherally inserted central catheter-related thrombosis in patients with liver cancer

Abstract

BACKGROUND Peripherally inserted central catheters (PICCs) are commonly used in hospitalized patients with liver cancer for the administration of chemotherapy, nutrition, and other medications. However, PICC-related thrombosis is a serious complication that can lead to morbidity and mortality in this patient population. Several risk factors have been identified for the development of PICC-related thrombosis, including cancer type, stage, comorbidities, and catheter characteristics. Understanding these risk factors and developing a predictive model can help healthcare providers identify high-risk patients and implement preventive measures to reduce the incidence of thrombosis. AIM To analyze the influencing factors of PICC-related thrombosis in hospitalized patients with liver cancer, construct a predictive model, and validate it. METHODS Clinical data of hospitalized patients with liver cancer admitted from January 2020 to December 2023 were collected. Thirty-five cases of PICC-related thrombosis in hospitalized patients with liver cancer were collected, and 220 patients who underwent PICC placement during the same period but did not develop PICC-related thrombosis were randomly selected as controls. A total of 255 samples were collected and used as the training set, and 77 cases were collected as the validation set in a 7:3 ratio. General patient information, case data, catheterization data, coagulation indicators, and Autar Thrombosis Risk Assessment Scale scores were analyzed. Univariate and multivariate unconditional logistic regression analyses were performed on relevant factors, and the value of combined indicators in predicting PICC-related thrombosis in hospitalized patients with liver cancer was evaluated using receiver operating characteristic (ROC) curve analysis. RESULTS Univariate analysis showed statistically significant differences (P < 0.05) in age, sex, Karnofsky performance status score (KPS), bedridden time, activities of daily living impairment, parenteral nutrition, catheter duration, distant metastasis, and bone marrow suppression between the thrombosis group and the non-thrombosis group. Other aspects had no statistically significant differences (P > 0.05). Multivariate regression analysis showed that age ≥ 60 years, KPS score ≤ 50 points, parenteral nutrition, stage III to IV, distant metastasis, bone marrow suppression, and activities of daily living impairment were independent risk factors for PICC-related thrombosis in hospitalized patients with liver cancer (P < 0.05). Catheter duration of 1-6 months and catheter duration > 6 months were protective factors for PICC-related thrombosis (P < 0.05). The predictive model for PICC-related thrombosis was obtained as follows: P predictive probability = [exp (Logit P )]/[1 + exp (Logit P )], where Logit P = age × 1.907 + KPS score × 2.045 + parenteral nutrition × 9.467 + catheter duration × 0.506 + tumor-node-metastasis (TNM) staging × 2.844 + distant metastasis × 2.065 + bone marrow suppression × 2.082 + activities of daily living impairment × 13.926. ROC curve analysis showed an area under the curve (AUC) of 0.827 (95%CI: 0.724-0.929, P < 0.001), with a corresponding optimal cut-off value of 0.612, sensitivity of 0.755, and specificity of 0.857. Calibration curve analysis showed good consistency between the predicted occurrence of PICC-related thrombosis and actual occurrence (P > 0.05). ROC analysis showed AUCs of 0.888 and 0.729 for the training and validation sets, respectively. CONCLUSION Age, KPS score, parenteral nutrition, TNM staging, distant metastasis, bone marrow suppression, and activities of daily living impairment are independent risk factors for PICC-related thrombosis in hospitalized patients with liver cancer, while catheter duration is a protective factor for the disease. The predictive model has an AUC of 0.827, indicating high predictive accuracy and clinical value.