三种评分系统在肝硬化食管胃底静脉曲张破裂出血患者风险评估中的应用

Risk assessment of cirrhosis patients with esophageal and gastric variceal bleeding by three scoring systems

摘要:

目的 探讨MELD(model for end-stage liver disease)、GBS(glasgow-blatchford score)、AIMS65评分系统在肝硬化食管胃底静脉曲张破裂出血（esophageal and gastric variceal bleeding, EGVB）患者风险评估中的临床应用价值。 方法 对天津医科大学总医院消化内科2015年1月1日至2018年3月1日入院的182例肝硬化EGVB患者进行回顾分析，依据MELD、GBS、AIMS65评分系统标准针对每例患者进行评分，评估各评分系统正确将肝硬化EGVB归为“高风险患者”的能力，并绘制受试者工作特征曲线（receiver-operating characteristic curve，ROC），采用曲线下面积（area under curve,AUC）评估各评分系统针对不同临床结局（输血、再出血、住院死亡）的预测能力，AUC＞0.7认为有较高准确性。 结果 临床结局包括输血113例（62.1%）、再出血31例（17.0%）、死亡11例（6.0%）。MELD评分得分为7～25分，其中得分＜9分4例（2.2%）；GBS评分得分为3～16分；AIMS65评分得分为0～3分，其中得分0～1分139例（76.4%，0分68例、1分71例）。MELD、GBS、AIMS65评分系统预测输血的AUC分别为0.514（95%CI：0.439～0.589）、0.681（95%CI：0.608～0.748）、0.669（95%CI：0.596～0.737）；预测再出血的AUC分别为0.525（95%CI：0.449～0.599）、0.528（95%CI：0.453～0.602）、0.580（95%CI：0.505～0.652）；预测住院死亡的AUC分别为0.642（95%CI：0.567～0.711）、0.581（95%CI：0.505～0.653）、0.786（95%CI：0.719～0.843），AIMS65优于MELD（P=0.083 6）和GBS（P=0.047 0）。 结论 GBS能正确将肝硬化EGVB患者归类为“高风险人群”，优于AIMS65和MELD评分系统。对于肝硬化EGVB患者，3种评分系统对输血和再出血的预测价值均不高，AIMS65对住院死亡有较高的预测价值。

Abstract:

Objective To compare the risk assessment capability of model for end-stage liver disease (MELD), glasgow-blatchford score (GBS), and the AIMS65 scoring system for liver cirrhosis patients with esophageal and gastric variceal bleeding (EGVB). Methods A retrospective analysis was made on data of 182 cirrhosis patients with EGVB admitted to the Department of Gastroenterology, General Hospital of Tianjin Medical University from January 1, 2015 to March 1, 2018. According to the MELD, GBS and AIMS65 scoring system, the corresponding scores of each patient were calculated to evaluate the ability of the three scoring systems to correctly classify EGVB as a "high-risk patient". The receiver operating characteristic curve was drawn to compare the predictive value of three scoring systems for different clinical outcomes (blood transfusion, rebleeding, and death). The area under curve (AUC)＞0.7 was believed to have higher accuracy. Results The clinical outcomes of 182 patients included blood transfusion in 113 (62.1%) cases, rebleeding in 31 (17.0%) cases, and death of 11 (6.0%) cases. The MELD score was 7-25, GBS was 3-16, and AIMS65 score was 0-3. There were 4 (2.2%) patients with MELD score < 9, 139 (76.4%) patients with AIMS65 score 0-1, including 68 patients with AIMS65 score of 0 and 71 patients with AIMS65 score of 1. The AUC of MELD, GBS and AIMS65 for predicting blood transfusion was 0.514 (95%CI: 0.439-0.589), 0.681 (95%CI: 0.608-0.748), and 0.669 (95%CI: 0.596-0.737), respectively. When predicting rebleeding, the AUC of MELD, GBS and AIMS65 was 0.525 (95%CI: 0.449-0.599), 0.528 (95%CI: 0.453-0.602) and 0.580 (95%CI: 0.505-0.652), respectively. When predicting in-hospital mortality, the AUC of MELD, GBS and AIMS65 was 0.642 (95%CI: 0.567-0.711), 0.581 (95%CI: 0.505-0.653) and 0.786 (95%CI: 0.719-0.843), respectively. AIMS65 was superior to MELD (P=0.083 6) and GBS (P=0.047 0). Conclusion GBS can correctly classify cirrhosis patients with EGVB as "high-risk group", and is better than AIMS65 and MELD scoring system. MELD, GBS and AIMS65 all have poor accuracy in predicting blood transfusion and rebleeding, AIMS65 has a higher predictive value for death.