Abstract: Objective　To explore the risk factors for sepsis-induced left ventricular diastolic dysfunction （LVDD），and to develop and validate a cardiac marker-based diagnostic model for this disease.Methods　We enrolled sepsis or septic shock from Emergency Department，Emergency Intensive Care Unit，Intensive Care Unit and Respiratory Intensive Care Unit，the First Affiliated Hospital of Xinjiang Medical University from August 2015 to August 2017，including 177 with normal cardiac function was normal cardiac function group，and 66 with sepsis-induced LVDD was sepsis-induced LVDD group.We collected their clinical data and laboratory findings.Co-predictor variable set pre1 （containing the associated variables for sepsis-induced LVDD included in the multivariate Logistic regression model） was used to construct a diagnostic model for sepsis-induced LVDD.Results　Two groups showed significant differences in the age，systolic blood pressure，diastolic blood pressure，mean arterial pressure （MAP），heart rate，respiratory frequency，hemoglobin，INR，partial thromboplastin time （PTT），blood lactate，NT-proBNP，and heart-type fatty acid-binding protein （H-FABP），new arrhythmia （P<0.05）.Multivariate Logistic regression analysis demonstrated that age，INR，PTT，NT-proBNP，H-FABP and new arrhythmia were associated with sepsis-induced LVDD （P<0.05）.Multivariate Logistic regression equation was established：logit（P）=3.231+0.046age-0.344MPA+0.104NT-proBNP+0.065H-FABP+0.744new arrhythmia.Age，NT-proBNP，H-FABP and new arrhythmia were independent factors of sepsis-induced LVDD （P<0.05）.For predicting sepsis-induced LVDD，the AUC of co-predictor variable set pre1 was larger than that of age，MAP，NT-proBNP，H-FABP，new arrhythmia （Z=1.798，P=0.040；Z=3.124，P=0.001；Z=1.894，P=0.032；Z=1.671，P=0.050；Z=1.901，P=0.026）.Eighty patients with sepsis were randomly selected，the sensitivity，specificity，coincidence rate，positive likelihood ratio，and negative likelihood ratio of co-predictor variable set pre1 in predicting sepsis-induced LVDD was 78.12%，89.58%，85.00%，7.49，4.09，respectively.Conclusion　Age，NT-proBNP，H-FABP and new arrhythmia are independent factors of sepsis-induced LVDD.The co-predictor variable set is superior to each of the variable alone in the diagnosis of sepsis-induced LVDD.The cardiac marker-based diagnostic model played a role in predicting sepsis-induced LVDD.

Key words: Sepsis, Biological markers, Ventricular dysfunction

摘要： 目的　探讨脓毒症舒张性心功能障碍的相关影响因素，构建脓毒症舒张性心功能障碍联合诊断模型并验证。方法　纳入2015年8月—2017年8月入住新疆医科大学第一附属医院急诊科、急诊重症监护室、重症监护室及呼吸重症监护室脓毒症和脓毒性休克患者，选取其中心功能正常177例（心功能正常组）和舒张性心功能障碍66例（舒张性心功能障碍组）。记录患者一般资料及实验室检查指标，采用多因素Logistic分析模型构建联合诊断模型，生成联合预测因子pre1。结果　心功能正常组和舒张性心功能障碍组患者年龄、收缩压、舒张压、平均动脉压、心率、呼吸频率、血红蛋白、国际标准化比值（INR）、部分凝血活酶时间、血乳酸、N末端脑钠肽前体（NT-proBNP）、人心型脂肪酸结合蛋白（H-FABP）水平、新发心律失常发生率比较，差异均有统计学意义（P<0.05）。多因素Logistic回归模型结果显示，年龄、INR、部分凝血活酶时间、NT-proBNP、H-FABP、新发心律失常与脓毒症舒张性心功能障碍有回归关系（P<0.05）。建立多因素Logistic回归方程：logit（P）=3.231+0.046×年龄-0.344×平均动脉压+0.104×NT-proBNP+0.065×H-FABP+0.744×新发心律失常。年龄、NT-proBNP、H-FABP、新发心律失常是脓毒症舒张性心功能障碍的独立影响因素（P<0.05）。联合预测因子pre1诊断脓毒症舒张性心功能障碍的AUC大于年龄、平均动脉压、NT-proBNP、H-FABP、新发心律失常（Z=1.798，P=0.040；Z=3.124，P=0.001；Z=1.894，P=0.032；Z=1.671，P=0.050；Z=1.901，P=0.026）。简单随机选取80例脓毒症患者，代入联合诊断模型灵敏度为78.12%，特异度为89.58%，符合率为85.00%，阳性似然比为7.49，阴性似然比为4.09。结论　年龄、NT-proBNP、H-FABP、新发心律失常是脓毒症舒张性心功能障碍的独立影响因素，联合预测因子诊断脓毒症舒张性心功能障碍优于单项指标，构建联合诊断模型具有临床应用价值及意义。
脓毒症；生物学标记；心室功能障碍