影响慢性肝衰竭患者生存时间的独立危险因素及其预测模型

Clinical study of survival time for chronic liver failure

目的探讨影响慢性肝衰竭患者生存时问的独立危险因素，并建立预测模型公式。方法采用回顾性病例分析方法，选择天津市第三中心医院2002年5月2007年5月行人工肝治疗的慢性肝功能衰竭、病历资料完整的患者362例，收集其临床资料，并使用SPSS13．0进行统计学分析，其中定量资料使用t检验和秩和检验，定性资料使用卡方检验，并进一步使用Cox回归分析筛选出对生存时间有影响的独立危险因素。根据Cox回归模型筛选出的独立危险因素建立适合我国慢性肝衰竭的预测模型公式。结果经Cox回归筛选影响生存时间的独立危险因素中有统计学差异的有：Child-Pugh评分、血钠值、是否有胆酶分离、是否有腹水、年龄、酪氨酸、精氨酸。本研究建立的Cox模型（包括Child-Pugh评分、血钠值、是否有胆酶分离、是否有腹水、精氨酸、酪氨酸、年龄）公式为：Cox公式评分=0．31×年龄6．62X合并腹水分数（是为1分，否为2分）＋10．16×胆酶分离分数（是为2分，否为1分）＋4．65×Child-Pugh分值0．39×血钠值＋0．05X酪氨酸值＋0．04×精氨酸值。该公式的AuC值为0．782，最佳临界值为27．69。结论Child-Pugh评分、血钠值、是否有胆酶分离、是否有腹水、年龄、酪氨酸、精氨酸是影响生存时间的独立危险因素。本研究建立的Cox模型公式对慢性肝衰竭患者的生存时间有较好的预测性。

Objective To identify independent risk factors influencing the survival time of patients with chronic liver failure and construct a predictive model. Methods Retrospective analysis was applied to clinical data of 362 patients with chronic liver failure treated with artificial liver in Tianjin third centre hospital between May 2002 and May 2007. Data were analyzed with SPSS 13.0 statistic software, t test and rank test were used on quantitative data, chi-square test was used on qualitative data, Cox regression analysis was used to select the independent risk factors influencing the survival time. According to independent risk factors from Cox regression model, a prognostic model was established. Results 1. Independent risk factors （P 〈0.05） influencing the survival time were： Child-Pugh score, bilirubin separation ALT, ascites, arginine, age, tyrosine and serum sodium. 2. By receiver operating characteristic curves （ROC） analysis, the area under ROC （AUR） to predict the outcome of chronic liver failure patients was 0.782, and the cutoff score was 27.69. Conclusions 1. Child-Pugh score, bilirubin separation ALT, ascites, arginine, age, tyrosine and serum sodium are independent risk factors affecting survival time of patients with chronic liver failure. 2. Cox model we constructed can reliably predict the survival time of patients with chronic liver failure.