肝硬化患者院内感染预测模型的建立及验证

Establishment and verification of the prediction model for nosocomial infection in patients with cirrhosis

目的研究肝硬化并发院内感染独立危险因素,并运用危险因素建立预测模型。方法回顾性分析2010年1月-2015年10月在恩施土家族苗族自治州中心医院住院肝硬化患者的临床资料,运用Logistic多因素分析,建立预后模型。用独立的临床病例资料进行验证评估模型判断能力。结果 532例肝硬化患者,其中院内感染68例(12.78%)。Logistic回归分析结果显示年龄、并发症、住院时间、肝功能分级、低蛋白血症、侵入性操作是肝硬化院内感染的独立危险因素。对构建的预测模型的预测能力进行评估发现:模型预测的灵敏度为84.00%,特异度为91.18%,总正确率为89.25%。结论年龄、并发症、住院时间、肝功能分级、低蛋白血症、侵入性操作是肝硬化医院内感染的独立危险因素,本研究构建的预测模型能够较为准确地预测肝硬化患者是否发生院内感染。

Objective To investigate independent risk factors of nosocomial infection in patients with cirrhosis and to establish and verify predictive model. Methods The clinical data of 532 patients with cirrhosis were retrospectively analyzed from Jan. 2010 to Oct. 2015. Logistic regression analysis was used to explore independent risk factors for nosocomial infection and to establish predictive model. Finally,validation set was used to verify model for predicting nosocomial infection. Results Among the 532 patients,68 patients had nosocomial infection（12. 78%）. Logistic regression analysis identified that age,complication,length of hospital stay,liver function classification,low serum albumin,surgical or invasive operation were independent factors of nosocomial infection. Sensitivity,specificity,total accuracy rate for model of predicting nosocomial infection were 84. 00%,91. 18%,89. 25%. Conclusion Age,complication,length of hospital stay,liver function classification,low serum albumin,surgical or invasive operation were independent factors of nosocomial infection,and they can be used to establish model for predicting nosocomial infection.