上海交通大学学报（医学版）

上海交通大学学报（医学版）

• 论著（临床研究） • 上一篇    下一篇

乙型肝炎后肝硬化患者食管静脉曲张发生的无创性预测指标研究

赵丹，毛华，黄纯炽，张绍衡   

  1. 广东省南方医科大学附属珠江医院消化内科，广州 510580
  • 出版日期:2015-03-28 发布日期:2015-03-26
  • 通讯作者: 毛华, 电子信箱: huam@fimmu.com。
  • 作者简介:赵丹（1988—）， 女， 硕士生; 电子信箱: 623324331@qq.com。
  • 基金资助:

    广东省科技厅科技计划项目（2009B030801212）

Study on noninvasive prediction indicators for incidence of esophageal varices of patients with liver cirrhosis caused by hepatitis B

ZHAO Dan, MAO Hua, HUANG Chun-chi, ZHANG Shao-heng   

  1. Department of Grastroenterology, Zhujiang Hosipital, Southern Medical University, Guangzhou 510580, China
  • Online:2015-03-28 Published:2015-03-26
  • Supported by:

    Science and Technology Planning Project of Guangdong Province, 2009B030801212

PDF (PC)

1193

摘要：

目的 探讨肝硬化患者发生食管静脉曲张（EV）的无创性指标。方法 收集肝硬化住院患者274例，记录性别、年龄、发病原因、外周血小板计数（PLT）、总胆红素、白蛋白、凝血时间、国际标准化比值（INR）、粪便潜血（OB）、脾脏长径（SD）和厚度以及门静脉宽度。首先对150例肝硬化患者各指标分别进行方差分析和多因素Logistic回归分析，建立预测模型，分别用预测模型与单独PLT、PLT/SD值验证另外124例肝硬化患者有无EV发生，比较这些指标的受试者工作特征曲线下面积（AUCROC）、特异度及敏感度。结果 多因素Logistic回归分析显示：PLT、SD、INR及OB结果共同构建的模型可作为肝硬化患者EV的预测指标（P<0.05）；AUCROC的计算结果显示：预测模型AUCROC (95%CI)为0.925（0.881, 0.969），模型与PLT/SD（χ2=7.212，P=0.007）和PLT（χ2=9.412，P=0.002）有关，PLT/SD与PLT有关（χ2=0.186，P=0.666），提示预测模型对有无EV发生的预测更具优势。结论 PLT、SD、INR、OB实验结果建立的预测模型对肝硬化患者发生EV的风险有较好的预测作用。

关键词: 肝硬化, 食管静脉曲张, 预测模型, 无创性

Abstract:

Objective To investigate noninvasive indicators for the incidence of esophageal varices (EV) of patients with liver cirrhosis. Methods A total of 274 inpatients with liver cirrhosis were selected. The gender, age, etiology, platelet count (PLT), total bilirubin, albumin, coagulation time, international normalized ratio (INR), fecal occult blood (OB), spleen diameter (SD), spleen thickness, and width of portal vein of patients were recorded. The variance analysis and multivariate Logistic regression analysis were conducted for indicators of 150 patients with liver cirrhosis. The prediction model was established. The prediction model, PLT, and PLT/SD were used to verify whether other 124 patients with liver cirrhosis developed EV. AUCROC, sensitivity, and specificity of patients were compared. Results The multivariate Logistic regression analysis showed that the model that was established by test results of PLT, SD, INR, and OB could be used to predict the incidence of EV of patients with liver cirrhosis (P<0.05). Computation results of AUCROC showed that the AUCROC (95%CI) of prediction model was 0.925 (0.881, 0.969), the model was correlated with PLT/SD (χ2=7.212, P=0.007) and PLT (χ2=9.412, P=0.002); and PLT/SD was correlated with PLT (χ2=0.186, P=0.666), which indicated that the prediction model was better for predicting the incidence of EV. Conclusion The prediction model that is established by test results of PLT, SD, INR, and OB is helpful for predicting the incidence of EV of patients with liver cirrhosis.

Key words: liver cirrhosis, esophageal varices, predictive model, noninvasive indictor