【摘要】 目的 探究Logistic回归模型拟合临床因素、营养状况、炎症指标对维持性血液透析患者并发肺部感染的预测价值。方法 选取2018年8月—2020年8月维持性血液透析152例作为研究对象，根据是否并发肺部感染分为感染组32例与未感染组120例。比较2组临床因素、营养状况指标[白蛋白(Alb)、前白蛋白(PA)、血红蛋白(Hb)]、炎症指标[C反应蛋白(CRP)、肿瘤坏死因子-α(TNF-α)]水平，分析营养指标与炎症指标相关性，采用Logistic回归模型拟合临床因素、营养状况、炎症指标生成联合预测因子，评价各指标与联合预测因子对维持性血液透析并发肺部感染的预测价值，并进行个体值预测。结果 感染组年龄≥60岁、合并糖尿病、平均每周透析时间≥12 h及血清CRP、TNF-α水平高于未感染组，血清Alb、PA、Hb水平低于未感染组(P<0.05,P<0.01)。血清Alb、PA、Hb与CRP、TNF-α呈负相关(P<0.01)。年龄、合并糖尿病、平均每周透析时间、血清CRP、TNF-α是维持性血液透析并发肺部感染的独立危险因素，血清Alb、PA、Hb是维持性血液透析并发肺部感染的独立保护因素(P<0.01)。联合预测因子预测维持性血液透析并发肺部感染的预测价值高于各临床因素、营养状况、炎症指标单独预测；随机抽取1例患者，将各自变量代入概率预测方程得到概率值P=0.162,小于最佳临界值，故在预测准确率为82.24%的条件下，该患者在维持性血液透析期间不会并发感染。结论 Logistic回归模型拟合临床因素、营养状况、炎症指标生成的联合预测因子对维持性血液透析患者并发肺部感染具有可靠预测价值，能为临床及早防治提供可靠依据。更多还原

【Abstract】 Objective To explore the value of fitting Logistic regression model with clinical factors, nutritional status, and inflammatory indicators in prediction of pulmonary infection in patients undergoing maintenance hemodialysis(MHD). Methods A total of 152 patients undergoing MHD between August 2018 and August 2020 were selected and divided into infection group(n=32) and non-infection group(n=120) according to whether or not complicated by pulmonary infection. Levels of clinical factors, nutritional status indexes [albumin(ALB), prealbumin(PA), hemoglobin(Hb)], inflammatory indexes [C-reactive protein(CRP), tumor necrosis factor-α(TNF-α)] were compared between two groups, and correlations between nutritional indexes with inflammatory indexes were analyzed. Logistic regression model was used to fit clinical factors, nutritional status and inflammatory indexes to generate joint predictor, and values of each one index and joint predicting factor in prediction of patients undergoing MHD complicated by pulmonary infection were evaluated, and the individual value was predicted. Results In infection group, serum levels of CRP and TNF-α in patients aged more than 60 years old, having diabetes mellitus and the dialysis time of average weekly equal or more than 12 h were significantly higher, while serum levels of ALB, PA and Hb were significantly lower than those in non-infection group(P<0.05, P<0.01). Serum ALB, PA and Hb were negatively correlated with CRP and TNF-α(P<0.01). Age, complicated by diabetes mellitus, average weekly dialysis time, serum CRP and TNF-α were independent risk factors, while serum ALB, PA and Hb were independent protective factors for patients undergoing HMD complicated by pulmonary infection(P<0.01). The value of joint predictor in prediction of patients undergoing MHD complicated by pulmonary infection was higher than that by one index of clinical factors, nutritional status and inflammatory indexes alone. One patient was randomly selected, and their variables were substituted into the probability prediction equation to obtain the probability value(P=0.162), which was less than the optimal critical value. Therefore, the patient did not experience infection during MHD under the condition that the prediction accuracy was 82.24%. Conclusion The joint predictor generated by fitting Logistic regression model with clinical factors, nutritional status and inflammatory indexes has reliable predictive value for pulmonary infection in patients undergoing MHD, which may provide basis for early prevention and treatment in clinical practice.更多还原