传统肾脏替代治疗开始指标在判断重症急性肾损伤患者预后的价值

The clinical data used for conventional indications of initiating renal replacement therapy and for predicting the prognosis of critically ill patients with acute kidney injury

目的 探讨血清肌酐（serum creatinine，SCr）或尿素氮（blood urea nitrogen，BUN）水平、急性肾损伤（acute kidney injury，AKI）分级等传统肾脏替代治疗（renal replacement therapy，RRT）开始指标在判断重症AKI患者预后的价值。方法 采用回顾性分析方法，选择在2011年1月至2015年1月期间入住浙江省人民医院重症监护病房（intensive care unit，ICU）诊断为AKI并接受连续性肾脏替代治疗（continuous renal replacement therapy， CRRT）治疗的258例成年患者，根据出院时预后分为存活组（n=104）和死亡组（n=154），比较两组患者在一般情况、疾病组成，CRRT治疗前肾功能情况（尿量、SCr、BUN和AKI分级）、内环境稳态（酸碱平衡和电解质水平）和疾病严重程度（APACHE Ⅱ评分和SOFA评分）等指标上的差异。同时采用多因素Cox比例风险模型和ROC曲线分析，筛选影响重症AKI患者预后的危险因素。结果 两组患者在性别、年龄、原发疾病性质、AKI病因组成、CRRT治疗前APACHE Ⅱ评分、肾功能情况（AKI分级、尿量、BUN和SCr）、血钾和血磷水平等因素上均未存在显著差异（均P＞0.05）；但死亡组中严重脓毒症患者比例更高（31.17% 比19.23%，P=0.033）、pH值更低（7.27±0.34 比7.41±0.34，P=0.024）、乳酸水平更高（3.97±2.87比2.64±2.30，P=0.006）；采用多因素Cox比例风险模型检验分析后发现，仅血磷水平（P=0.043）和乳酸水平（P=0.009）为影响重症AKI患者预后的独立危险因素，而SCr、BUN、AKI分级、尿量、pH值、碳酸氢根水平（HCO3-）、血钾水平等传统RRT开始参考指标则与患者预后无显著相关（均P＞0.05）。因此，将pH值、HCO3-、血钾水平、血磷水平、尿量和AKI分级这六项传统指标进行综合，采用ROC曲线分析并比较综合指标和SCr、BUN、乳酸等因素在预测重症AKI患者院内死亡率的作用，结果显示综合指标和乳酸的曲线下面积（AUC）较高，分别为0.669[95%可信区间（CI）：0.577-0.762]和0.683（95%CI：0.590-0.777），而SCr和BUN这两项指标的AUC均＜0.5，分别为0.460（95% CI：0.358-0.562）和0.469（95% CI：0.366-0.571）。结论 在预测重症AKI患者预后的作用上，RRT治疗前的综合指标优于任一传统RRT开始参考指标。因此，临床上我们在判断重症AKI患者何时应开始RRT治疗时，应综合考虑，而不是根据某一指标。

Objective To investigate the value of clinical data used for conventional indications of initiating renal replacement therapy （RRT） such as serum creatinine （SCr） , blood urea nitrogen （BUN） and acute renal injury （AKI） stage and in estimating the prognosis of critically ill patients with AKI. Methods A retrospective analysis of 258 AKI adult inpatients treated with continuous renal replacement therapy （CRRT） in ICU from Jan. 2011 to Jan. 2015. According to the outcomes, all subjects were divided into survival group （ n = 104） and death group （ n = 154 ）. The general condition, AKI causes, results of renal function （ urine output, SCr, BUN and AKI stage）, homeostasis （acid-base balance and electrolyte level） , severity of disease （ APACHE Ⅱ score and SOFA score） and others were compared between two groups. Additionally, risk factors for the prognosis of critically ill patients with AKI were screened by the multivariate Cox＇s proportional hazard models and the receiver operating characteristic （ROC） curve. Results There were no significant differences in gender, age, primary disease, AKI causes, APACHE Ⅱ score, renal function （urine output, SCr, BUN and AKI stage）, serum potassium level and phosphorus level between two groups before CRRT （P 〉 0. 05）, but more patients in death group had severe sepsis （ 31.17% vs. 19.23% , P = 0. 033 ） , lower pH value [ （7. 27 ±0. 34）vs. （7.41±0. 34）, P =0. 024] and higher level of lactate [ （3.97 ±2. 87） vs. s （2. 64 ± 2. 30 ）, P = 0. 006 ]. After the analysis with multivariate Coxg proportional hazard models, it was found that the levels of serum phosphorus （ P = 0. 043 ） and lactate （ P = 0. 009 ） were the independent risk factors for prognosis of critically ill patients with AK1, and other conventional indications for initiating RRT such as SCr, BUN, AKI stage, urine output, pH, bicarbonate level or potassium level were not closely associated with the prognosis of patients （P 〉 0. 05 ）. Therefore, a composite of these six variables （pH, bicarbonate level, phosphorus level, potassium level, urine output and AKI stage） was analyzed. According to the analysis result of ROC curve, the diagnostic value of combined six different variables in predicting in-hospital mortality of AKI patients [area under the curve （AUC） 0. 669, 95% CI： 0. 577-0. 762] was almost as high as that of lactate （AUC： 0. 683,95% CI： 0. 590-0. 777） , and significantly higher than SCr （AUC： 0. 460, 95% CI： 0. 358- 0.562） , BUN （AUC： 0.469, 95% CI： 0.366-0.571）. Conclusions This composite of six different variables is more useful than any other conventional indications for initiating RRT in predicting post-AKI mortality. As a result, a composition of six different variables should be considered rather than any single variable alone for indication of initiating RRT in critically ill patients with AKI.