骨科创伤患者术后多重耐药菌感染风险Nomogram模型构建与验证

Establishment and validation of nomogram model for the risk of multidrug- resistant organism infection in patients after orthopedic trauma surgery

摘要:

目的 探讨骨科创伤患者术后多重耐药菌(MDRO)感染的危险因素及风险列线图(Nomogram)预测模型的构建。 方法 回顾性分析2019年1月—2021年1月某院重症监护病房(ICU)的458例骨科创伤住院患者的临床资料，分别使用单因素和多因素lasso logistic回归分析骨科创伤住院患者术后MDRO感染的独立危险因素。最后纳入筛选出的独立危险因素建立Nomogram预测模型，另选取2021年2—10月224例骨科创伤患者建立验证组，对建模组数据进行验证。 结果 458例创伤患者为建模组，其中创伤合并MDRO感染114例，感染率24.89%；非MDRO感染组患者344例。两组患者临床资料进行lasso logistic回归分析，APACHE Ⅱ评分≥20分(OR=2.567，95%CI: 1.593~4.178)、发热日数≥3 d(OR=2.656，95%CI: 1.631~4.361)、住院日数≥10 d(OR=3.563，95%CI: 2.207~5.808)、使用抗菌药物(OR=2.314，95%CI: 1.415~3.848)、联合使用抗菌药物日数≥7 d(OR=2.114，95%CI: 1.329~3.624)是创伤患者术后MDRO感染的独立危险因素。基于回归分析结果中的5项独立危险因素，建立预测创伤术后MDRO感染的Nomogram模型，验证结果显示：Bootstrap内部验证，预测值与实测值基本一致，建模组预测曲线下面积(AUC)为0.877 8，验证组AUC为0.871 8。说明预测模型具有较好的预测能力。决策分析曲线显示该模型有较高的获益性。 结论 Nomogram预测模型的建立有利于早期识别创伤患者术后MDRO感染的高危因素，及时采取防控措施，从而降低MDRO医院感染率。

Abstract:

Objective To explore the risk factors and establishment of risk nomogram model for multidrug-resistant organism (MDRO) infection in patients undergoing orthopedic trauma surgery. Methods Clinical data of 458 patients who undergoing orthopedic trauma surgery in an intensive care unit of a hospital from January 2019 to Ja-nuary 2021 were retrospectively analyzed, univariate and lasso logistic regression analysis were used to analyze independent risk factors for MDRO infection in hospitalized patients after orthopedic trauma surgery. The screened independent risk factors were included to establish nomogram prediction model, in addition, 224 orthopedic trauma patients from February to October 2021 were selected to establish a verification group to verify the data of modeling group. Results A total of 458 trauma patients were as the modeling group, including 114 trauma patients complicated with MDRO infection, with an infection rate of 24.89%; there were 344 patients in non-MDRO infection group. Clinical data of two groups of patients were conducted lasso logistic regression analysis, APACHE Ⅱ score≥20 points (OR=2.567, 95%CI: 1.593-4.178), fever days ≥ 3 days (OR=2.656, 95%CI: 1.631-4.361), hospitalization days≥10 days (OR=3.563, 95%CI: 2.207-5.808), antimicrobial use (OR=2.314, 95%CI: 1.415-3.848), combined antimicrobial use days ≥7 days (OR=2.114, 95%CI: 1.329-3.624) were independent risk factors for MDRO infection in patients after orthopedic trauma surgery. Based on the five independent risk factors in the regression analysis, a nomogram model for predicting MDRO infection after trauma surgery was established, the verification results showed that the predicted value was basically consistent with the measured value in the internal verification of Bootstrap, the area under the prediction curve (AUC) of modeling group and verification group were 0.877 8 and 0.871 8 respectively, which showed that the prediction model had good prediction ability. The decision curve analysis (DCA) showed that the model had high benefit. Conclusion The establishment of nomogram prediction model is conducive to early identification of the high-risk factors of post-operative MDRO infection in trauma surgery patients, and timely take prevention and control measures, so as to reduce MDRO healthcare-associated infection.