探讨急性生理学和慢性健康状况（APACHE）Ⅳ评分用于预测肝移植患者术后住院期间病死率和入住重症监护病房（ICU）时间的价值。

回顾性研究2006年2月至2009年7月在中山大学附属第三医院肝移植中心接受同种异体原位肝移植的195例患者临床资料。其中男171例，女24例；年龄（48±11）岁。所有患者均签署知情同意书，符合医学伦理学规定。收集患者APACHE Ⅳ评分所需要的参数、实际入住ICU时间、住院期间死亡情况，并计算患者APACHE Ⅳ评分分值、实际病死率、预测入住ICU时间、预测病死率，并绘制其预测患者死亡的受试者工作特征（ROC）曲线。观察患者住院期间存活和死亡情况及其APACHE Ⅳ评分。应用ROC曲线研究APACHE Ⅳ评分预测患者住院期间死亡能力。采用Hosmer-Lemeshow拟合优度检验校准度分析APACHE Ⅳ评分系统预测患者住院期间死亡的准确性。住院期间存活者的实际入住ICU时间与预测入住ICU时间的比较采用Wilcoxon秩和检验，两者相关性研究采用Spearman秩相关分析。

本研究195例患者中，存活168例，死亡27例，病死率为13.8%；总体患者APACHE Ⅳ评分为（41±22）分，其中存活患者的APACHEⅣ评分为（36±16）分，死亡患者为（75±25）分。APACHE Ⅳ评分预测患者住院期间死亡的ROC曲线下面积（A）为0.937，其对患者住院期间死亡的预测切点为56分，灵敏度为0.85，特异度为0.91，Youden指数为0.76。患者实际病死率为13.8%，APACHE Ⅳ评分预测患者病死率为3.8%。APACHE Ⅳ总体评分及30~60分、>60分组实际病死率的预测存在低估情况。经Hosmer-Lemeshow拟合优度检验发现APACHE Ⅳ评分对患者住院期间死亡预测拟合良好，校准度高（χ²=1.568，P>0.05）。患者实际入住ICU时间中位数为3.7（2.4，5.5）d，预测入住ICU时间为3.2（2.4~4.8）d。实际入住ICU时间明显长于预测入住ICU时间，两者差异有统计学意义（Z=3.760，P<0.05）。Spearman秩相关分析显示两者呈正相关（r=0.467，P<0.05）。

APACHE Ⅳ评分对肝移植术后患者住院期间病死率有一定的预测价值，但对患者病死率和入住ICU时间存在低估情况。

To explore the value of acute physiology and chronic health evaluation (APACHE) Ⅳ in predicting mortality during hospital stay and intensive care unit (ICU) length of stay (LOS) of patients after orthotopic liver transplantation (OLT).

Clinical data of 195 patients [171 males, 24 females, mean age of (48±11) years old] who underwent OLT from February 2006 to July 2009 in Liver Transplantation Center, the Third Affiliated Hospital of Sun Yat-sun University were studied retrospectively.The informed consents of all patients were obtained and the ethical committee approval was received. The required parameters for APACHE IV, actual ICU LOS and mortality during hospital stay were collected. The APACHE Ⅳ scores, actual mortality, estimated ICU LOS, estimated mortality were calculated and receiver operating characteristic (ROC) curve of predicting the patients' mortality was drawn. The survival, mortality and APACHE Ⅳ scores of patients during hospital stay were observed. The ability of APACHE Ⅳ in predicting mortality during hospital stay was analyzed using ROC curve, and the accuracy of APACHE Ⅳ in predicting mortality during hospital stay was analyzed using Hosmer-Lemeshow goodness-of-fit test. The actual ICU LOS and estimated ICU LOS of the survivals during hospital stay were compared using Wilcoxon rank sum test. The correlation between them was studied using Spearman rank correlation analysis.

There were 168 survivals and 27 deaths with the mortality of 13.8%. The mean APACHE Ⅳ score was (41±22) for all the patients, was (36±16) for the survivals and(75±25) for the death. The area under the curve(A) value of APACHE Ⅳ in predicting mortality during hospital stay was 0.937. The predictive cutoff point was 56 with the sensitivity 0.85, specificity 0.91 and the Youden index 0.76. The actual mortality of the patients was 13.8% and the estimated mortality by APACHE Ⅳ was 3.8%. The actual mortality was underestimated by APACHE Ⅳ in the overall score, 30-60 scores and >60 scores groups. The Hosmer-Lemeshow goodness-offit of APACHE Ⅳ in predicting mortality during hospital stay was good with a high calibration (χ²=1.568, P>0.05) . The median of actual ICU LOS was 3.7(2.4, 5.5)d, and the median of estimated ICU LOS was 3.2(2.4, 4.8)d. The actual ICU LOS was evidently longer than the estimated ICU LOS, and significant difference was observed between them (Z=3.760, P<0.05). There was positive correlation between the actual ICU LOS and the estimated ICU LOS by Spearman rank correlation analysis (r=0.467, P<0.05).

APACHE Ⅳ has a certain value in predicting mortality of patients after OLT during hospital stay, but it underestimates the mortality and ICU LOS.