Analysis of predictive factors of thrombosis in autogenous arteriovenous fistula

Abstract

Purpose: To investigate the predictors of early diagnosis of thrombus of autogenous arteriovenous fistula (aAVF). Methods: The included patients were divided into the thrombus group with aAVF failure or thrombosis and the control group with good internal fistula function. The general data of the patients, including age, sex, diabetes mellitus, were collected. Platelets (PLT), platelet crit (P-LCR), platelet distribution width (PDW), mean platelet volume (MPV), homocysteine (HCY), and other biochemical data were collected. The predictors of thrombus of aAVF were obtained by the t test and logistic regression analysis, and receiver operating characteristic (ROC) curve analysis was used to compare the area under the ROC curve (AUC) between the combined predictors and the original indicators. The optimal critical value was determined when the Youden index reached its maximum value, and the sensitivity, specificity, accuracy, diagnostic index, and so on were calculated. Finally, prediction was performed by substituting each value in individually. Results: PLT, PDW, P-LCR, MPV, and HCY showed significant differences between two groups ( p < 0.05). Logistic regression analysis showed that, for PLT (OR = 1.014, 95% CI 1.006–1.022, p = 0.01), PDW (OR = 1.295, 95% CI 1.009–1.661, p = 0.042), P-LCR (OR = 1.230, 95% CI 1.089–1.389, p = 0.001), MPV (OR = 1.696, 95% CI 1.101–2.613, p = 0.017), and HCY (OR = 1.332, 95% CI 1.182–1.502, p = 0.01), the difference was significant; PLT, PDW, P-LCR, MPV, and HCY were positively correlated with thrombogenesis ( p < 0.05). By logistic regression, a group of the five predictors of PLT, PDW, P-LCR, MPV, and HCY was obtained, and the combined predictors were 0.014*PLT + 0.258*PDW + 0.207*P-LCR + 0.528*MPV + 0.287*HCY. The area under the curve of the combined predictor was 0.933, the sensitivity was 92.4%, the specificity was 81.2%, the maximum diagnostic index was 0.736, the diagnostic cutoff point was 21.790, and the accuracy rate was 87%. Conclusion: PLT, PDW, P-LCR, MPV, and HCY are predictors of thrombus of aAVF. They can better predict thrombus of aAVF, and the combination of these five indicators is better than a single indicator.