Establishment of risk nomogram prediction model for venous catheter thrombosis.

Abstract

Venous catheters are widely used in clinical practice, but a drawback of their usage is the increased risk of thrombosis. The current study explored the risk factors affecting the formation of thrombosis following venous catheterization and establishes a risk nomogram prediction model for catheter-related thrombosis. Univariate and multivariate logistic regression analyses were carried out to identify the independent factors involved in venous catheter thrombosis. These factors were included in the construction of a nomogram. Finally, the C-index and calibration curves were used to validate the nomogram. A total of 146 cases were included in the sample, of which 36 were cases of thrombosis. The results of the univariate logistic regression analysis showed that the following were significant factors: age, Acute Physiology and Chronic Health Evaluation scoring system (APACHE II) score, white blood cell (WBC), hematocrit (HCT), international normalized ratio (INR), fibrinogen (FIB), and D-dimer. Multivariate logistic regression analysis was performed, which confirmed that the factors of age (AUC: 0.677, 95% CI: 0.564-0.790), APACHE II score (AUC: 0.746, 95% CI: 0.656-0.837), INR (AUC: 0.743, 95% CI: 0.636-0.849), and D-dimer (AUC: 0.826, 95% CI: 0.750-0.902) were independent variables. Next, a nomogram was constructed using these independent variables for predicting venous catheter thrombosis. Favorable results with C-indexes (0.816; 95% CI: 0.780-0.882) and calibration curves closer to ideal curves indicated the accurate predictive ability of this nomogram. The individualized nomogram demonstrated effective prognostic prediction for patients with venous thrombosis.