论文部分内容阅读
目的:评价简易无创模型AAR(AST-to-ALT ratio)、APRI(AST-to-platelet ratio index)、SPRI(spleen-to-platelet ratio index)、API(age-platelet index)、ASPRI(age-spleen-to-platelet ratio index)预测乙型肝炎相关肝硬化的临床价值。方法:慢性乙型肝炎170例,其中病理诊断为非肝硬化138例,肝硬化32例。参照原始文献构建预测肝纤维化程度的简易无创模型。统计分析采用SPSS 13.0软件。简易无创模型在非肝硬化与肝硬化患者之间的比较采用两独立样本的t检验。简易无创模型预测肝硬化的评价采用二分类Logistic逐步回归分析。结果:肝硬化患者的平均AAR、SPRI、API、ASPRI显著大于非肝硬化患者(P=0.000,0.009,0.000,0.005),平均APRI相近于非肝硬化患者(P=0.223)。只有AAR和API符合Logistic回归模型纳入自变量标准(P<0.05)。基于简易无创模型建立的Logistic回归模型预测肝硬化的灵敏度、特异度、阳性预测值、阴性预测值、准确度分别为0.22,0.99,0.70,0.84,0.84。结论:AAR和API可能是预测乙型肝炎相关肝硬化较可靠的简易无创模型。基于简易无创模型建立的预测乙型肝炎相关肝硬化的Logistic回归模型有一定的临床实践效能。
Objective: To evaluate the value of AST-to-ALT ratio, APRI, SPR, age-platelet index (APRI) spleen-to-platelet ratio index to predict the clinical value of hepatitis B-related cirrhosis. Methods: 170 cases of chronic hepatitis B, pathological diagnosis of 138 cases of non-cirrhosis, 32 cases of cirrhosis. Constructing a simple and non-invasive model for predicting the degree of liver fibrosis with reference to the original literature. Statistical analysis using SPSS 13.0 software. A simple, non-invasive model comparing t-tests with non-cirrhotic and cirrhotic patients using two independent samples. A simple noninvasive model for predicting cirrhosis was dichotomized Logistic stepwise regression analysis. Results: The average AAR, SPRI, API, ASPRI in cirrhotic patients were significantly higher than those in non-cirrhotic patients (P = 0.000,0.009,0.000,0.005). The average APRI was similar to that in non-cirrhotic patients (P = 0.223). Only AARs and APIs met Logistic regression models incorporating the independent variable criteria (P <0.05). Logistic regression model based on simple noninvasive model predicts the sensitivity, specificity, positive predictive value, negative predictive value and accuracy of liver cirrhosis were 0.22,0.99,0.70,0.84,0.84 respectively. Conclusion: AAR and API may be simple noninvasive models to predict the reliability of hepatitis B-related cirrhosis. Logistic regression models based on a simple non-invasive model for predicting hepatitis B-related cirrhosis have some clinical efficacy.