【摘要】 目的 探讨高龄住院老年患者产超广谱β-内酰胺酶(ESBL)大肠埃希菌定植的影响因素并构建预测模型。方法 选取2022年8月—2023年11月长治市人民医院收治的140例高龄患者划分至建模队列，另选取同期本院收治的63例高龄患者为验证队列，将建模队列中检出产ESBL大肠埃希菌定植的患者划分至暴露组，未检出产ESBL大肠埃希菌定植的患者划分至非暴露组，收集两组基线数据、抗菌药物使用情况、合并症情况，对存在差异的指标行多因素logistic回归分析，将多因素logistic回归分析结果构建列线图模型，通过C指数评定列线图辨识度，同时采取ROC曲线对内部验证结果进行评估，绘制ROC曲线评价预测模型的价值，构建校准曲线与决策曲线。结果 建模队列与验证队列的基线数据比较，差异均无统计学意义(P>0.05)。建模队列中，35例检出产ESBL大肠埃希菌定植者为暴露组，105例未检出者为非暴露组，两组的第三代头孢菌素使用、尿管留置、胃管留置、联用抗生素及住院时间比较，差异均有统计学意义(P<0.05)。多因素logistic回归分析结果显示：第三代头孢菌素使用、尿管留置、胃管留置、联用抗生素及住院时间是影响产ESBL大肠埃希菌定植的独立危险因素(P<0.05)。用验证队列对构建的预测模型进行内部验证，结果显示在建模队列和验证队列中模型均表现出较高的区分度和校准度，Hosmer-Lemeshow检验显示验证队列和建模队列预测概率与实际概率差异均无统计学意义(P>0.05)。以多因素logistic回归分析结果建立的列线图预测模型，经验证队列验证，ROC曲线显示AUC为0.917(95%CI:0.861～0.973),表明预测模型准确性较高。结论 第三代头孢菌素使用、住院时间、尿管留置、胃管留置及联用抗生素均为患者产ESBL大肠埃希菌定植的重要影响因素，以此为基础构建的列线图模型具有较高的临床价值。更多还原

【Abstract】 Objective To explore the influencing factors of Escherichia coli(E. coli) colonization in the production of extended spectrum β-lactamase(ESBL) in elderly hospitalized patients and establish a prediction model. Methods A total of 140 elderly patients admitted to Changzhi People’s Hospital from August 2022 to November 2023 were selected and divided into the modeling cohort, and another 63 elderly patients admitted to our hospital during the same period were selected as the validation cohort. Patients with ESBL-producing E. coli colonization detected in the modeling cohort were divided into the exposed group, and patients without ESBL-producing E. coli colonization were divided into the non-exposed group. Baseline data, antibacterial drug use, and comorbidities were collected from the two groups. Multivariate logistic regression analysis was performed for the indicators with differences. Multivariate logistic regression analysis results were used to construct a nomogram model. Column identification was assessed by C-index, and ROC curve was used to assess the internal validation results. Receiver operating characteristic curve(ROC curve) was drawn to evaluate the value of the prediction model and construct a calibration curve and decision curve were constructed. Results There were no significant differences in baseline data between the modeling cohort and the validation cohort(P>0.05). In the modeling cohort, 35 patients with ESBL-producing E. coli colonization detected were in the exposed group and 105 patients without ESBL-producing E. coli colonization were in the non-exposed group, and there were significant differences in the use of third-generation cephalosporins, urinary catheter indwelling, gastric tube indwelling, combined antibiotics, and length of hospital stay between the two groups(P<0.05). The results of multivariate logistic regression analysis showed that the use of third-generation cephalosporins, urinary catheter indwelling, gastric tube indwelling, combined antibiotics and length of hospital stay were independent risk factors affecting ESBL-producing E. coli colonization(P<0.05). The constructed prediction models were internally validated with the validation cohort, and the results showed that the models showed high discrimination and calibration in both the modeling cohort and the validation cohort, and the Hosmer Lemeshow test showed that there were no significant differences between the predicted probabilities and the actual probabilities in the validation cohort and the modeling cohort(P>0.05). The nomogram prediction model developed from the results of multivariate logistic regression analysis, validated by the validated cohort, showed an AUC of 0.917(95% CI: 0.861-0.973) by ROC curve, indicating a high accuracy of the prediction model. Conclusion The use of third-generation cephalosporins, length of hospital stay, urinary catheter indwelling, gastric tube indwelling and combined antibiotics are all important influencing factors of ESBL-producing E. coli colonization in patients, and the nomogram model constructed based on this has high clinical value.更多还原