对比天冬氨酸氨基转移酶-血小板比率指数（APRI）、血小板计数与脾脏直径比值（PC/SD）、血清腹水白蛋白梯度（SAAG）3种不同肝病模型诊断肝硬化食管胃静脉曲张（GOV）的准确性，并与增强CT比较，以期为临床筛选出诊断肝硬化GOV的最佳肝病模型。

选取2014年5月至2018年3月因肝硬化失代偿腹水收入常熟市第一人民医院住院的126例患者纳入研究。对患者行胃镜检查确诊是否存在GOV，并依据结果分为GOV组（51例）和非GOV组（75例）。收集患者相关检查数据并通过公式计算3种肝病模型APRI、PC/SD、SAAG的值，采用单因素分析（独立样本t检验）比较GOV组和非GOV组患者APRI、PC/SD、SAAG。将单因素分析差异有统计学意义的肝病模型进行多因素分析，通过基于偏最大似然估计的前进法二分类逐步logistic回归分析，筛选诊断GOV的有效肝病模型，并与增强CT比较受试者工作特征（ROC）曲线的曲线下面积（AUC），使用Youden指数确定最佳判定点。

GOV组患者与非GOV组患者比较，APRI、SAAG值明显升高［（0.72±0.31）vs（0.54±0.18），t=4.15，P<0.01；（21.62±3.14）g/L vs（15.87±1.78）g/L，t=13.08，P<0.01］，PC/SD值明显降低［（847.26±260.92）vs（1298.20±213.51），t=-10.23，P<0.01］，差异具有统计学意义。经多因素回归分析，PC/SD、SAAG是诊断GOV的独立危险因素，建立回归方程：logistic（P）=-5.266＋0.707 SAAG–0.007 PC/SD。通过ROC曲线对比，PC/SD（0.86，95%CI=0.79~0.92）、SAAG（0.89，95%CI=0.83~0.94）比增强CT（0.73，95%CI=0.65~0.81）有着更大的AUC，即更高的诊断准确性，差异具有统计学意义（P均<0.01）；但PC/SD和SAAG之间AUC对比差异无统计学意义（P>0.05）。使用Youden指数确定最佳判定点发现，PC/SD<1123.4具有对肝硬化GOV最高的诊断准确性（敏感度92.2%、特异度66.7%），SAAG>18.2g/L具有对肝硬化GOV最高的诊断准确性（敏感度86.3%、特异度85.3%）。

多种肝病模型中PC/SD、SAAG诊断肝硬化GOV准确性较高，且均优于增强CT，可在临床上作为胃镜发现肝硬化GOV的有效补充手段。

To compare the accuracy of three different liver disease models of aspartate aminotransferase-platelet ratio index (APRI), platelet count and spleen diameter ratio (PC/SD), and serum-ascites albumin gradient (SAAG) in the diagnosis of cirrhosis with GOV and to compare the results with those of enhanced CT, in order to find the best liver disease model for diagnosing cirrhotic GOV.

A total of 126 patients with decompensated ascites due to cirrhosis were enrolled. The patients were diagnosed with or without GOV by electronic gastroscopy. The relevant data were collected and APRI, PC/SD, and SAAG were calculated using formulae. Single-factor and multi-factor analyses were used to screen the effective liver disease model for GOV. These models were compared with enhanced CT with regard to the area under the receiver operating characteristic curve (AUC), and the Youden index was used to determine the best cut-off point.

Single-factor and multivariate regression analyses showed that PC/SD and SAAG were independent risk factors for the diagnosis of GOV. The regression equation was established as Logistic (P)=-5.266+ 0.707 SAAG -0.007 PC/SD. PC/SD and SAAG had larger AUC and higher diagnostic accuracy than enhanced CT [SAAG: 0.89 (95%CI: 0.83-0.94); PC/SD: 0.86 (95%CI: 0.79-0.92); enhanced CT: 0.73 (95%CI: 0.65-0.81), P<0.01] , but there was no significant difference in AUC between PC/SD and SAAG (Z=0.849, P=0.936). Using the Youden index to determine the best cut-off point, PC/SD<1123.4 had the highest diagnostic accuracy for cirrhotic GOV (sensitivity, 92.2%; specificity, 66.7%), and SAAG>18.2 g/L had the highest diagnostic accuracy for cirrhotic GOV (sensitivity, 86.3%; specificity, 85.3%).

PC/SD and SAAG for diagnosis of cirrhosis with GOV are more accurate, and they perform better than enhanced CT. PC/SD and SAAG can be used as an effective supplement to gastroscopy for the diagnosis of GOV.