个体化预测老年创伤性四肢骨折患者术后切口感染风险的列线图模型构建

Construction of a Nomogram Model for Individualized Prediction of the Risk of Postoperative Incision Infection in Elderly Patients with Traumatic Limb Fracture

目的探讨老年创伤性四肢骨折患者术后切口感染的影响因素,并构建列线图模型。方法选取2021年5月-2023年9月成都医学院第一附属医院收治的360例接受手术治疗的老年创伤性四肢骨折患者作为研究对象,采用随机数字表法分为建模组252例与验证组108例。采用LASSO回归、logistic回归分析术后切口感染的影响因素,根据影响因素构建列线图,通过ROC曲线、校准曲线评估风险预测模型的预测性能及效果。结果27个研究变量纳入LASSO回归模型中进行筛选,基于最小值与标准化构建2个模型。模型1共纳入12个变量,分别为体质量指数(BMI)、使用钢板、合并糖尿病、引流管放置时间、切口类型、手术时间、切口长度、术后3天外周血C反应蛋白与白蛋白比值(CAR)、血清降钙素原(PCT)/白蛋白(ALB)、白细胞介素-6(IL-6)、中性粒细胞与淋巴细胞比值(NLR)及血清红细胞沉降率(ESR);模型2共纳入7个变量,分别为合并糖尿病、引流管放置时间及术后3天外周血CAR、PCT/ALB、IL-6、NLR及ESR。以术后切口是否发生感染为因变量,模型1与模型2中经LASSO回归筛选出的变量作为自变量,分别构建logistic回归分析模型,模型1结果显示:BMI≥28 kg/m 2、合并糖尿病、切口类型Ⅲ型、手术时间、切口长度、使用钢板、引流管放置时间≥6 d及术后3天外周血CAR、PCT/ALB、IL-6、NLR、ESR均为老年创伤性四肢骨折患者术后切口感染的独立危险因素(P<0.05);模型2结果显示:合并糖尿病、引流管放置时间≥6 d及术后3天外周血CAR、PCT/ALB、IL-6、NLR及ESR均为老年创伤性四肢骨折患者术后切口感染的独立危险因素(P<0.05)。模型1在建模组与验证组中的ROC曲线下面积分别为0.949(95%CI:0.898~1.000)、0.981(95%CI:0.956~1.000),且模型1在建模组与验证组中的Hosmer-Lemeshow检验结果显示该模型未偏离完美拟合,校准曲线一致性良好。模型2在建模组与验证组中的ROC曲线下面积分别为0.882(95%CI:0.787~0.977)、0.921(95%CI:0.843~1.000),且模型2在建模组与验证组中的Hosmer-Lemeshow检验显示该模型偏离完美拟合,校准曲线一致性不佳。结论BMI、合并糖尿病、切口类型、手术时间、切口长度、使用钢板、引流管放置时间及术后3天外周血CAR、PCT/ALB、IL-6、NLR、ESR是引发老年创伤性四肢骨折患者术后切口感染的影响因素。

Objective To explore the influencing factors of incision infection in elderly patients with traumatic limb fracture after surgery,and to construct a nomographic model.Methods A total of 360 elderly patients with traumatic limb fracture who received surgical treatment in the First Affiliated Hospital of Chengdu Medical College from May 2021 to September 2023 were selected as the study subjects and divided into the modeling group(n=252)and the verification group(n=108)by the computer random number table.LASSO regression and logistic regression were used to analyze the influencing factors of postoperative incision infection,and column line diagrams were constructed according to the influencing factors,and the predictive performance and effect of the risk prediction model were evaluated by ROC curve and calibration curve.Results Twenty-seven research variables were included in the LASSO regression model for screening,and two models were constructed based on minimum values and standardization.Model 1 included 12 variables,including body mass index(BMI),use of steel plate,diabetes,drainage tube placement time,incision type,operation time,incision length,the ratio of peripheral blood C-reactive protein to albumin(CAR),serum procalcitonin(PCT)/albumin(ALB),interleukin-6(IL-6),neutrophil to lymphocyte ratio(NLR),and serum erythrocyte sedimentation rate(ESR)three days after operation.Seven variables were included in model 2,including diabetes,drainage tube placement time,and peripheral blood CAR,PCT/ALB,IL-6,NLR,ESR 3 days after surgery.The logistic regression analysis models were constructed with the dependent variable of whether the incision infection occurred after surgery,and the variables screened by LASSO regression in model 1 and model 2 as independent variables.The results of model 1 showed that BMI≥28 kg/m 2,combined with diabetes,incision typeⅢ,operation time,incision length,use of steel plate,placement time of drainage tube≥6 days,and peripheral blood CAR,PCT/ALB,IL-6,NLR,ESR 3 days after surgery were independent risk factors for postoperative incision infection in elderly patients with traumatic limb fracture(P<0.05).The results of model 2 showed that the combination of diabetes,the placement time of drainage tube≥6 days and the peripheral blood CAR,PCT/ALB,IL-6,NLR,ESR 3 days after operation were independent risk factors for postoperative incision infection in elderly patients with traumatic limb fracture(P<0.05).The area under the ROC curve of Model 1 in the modeling and validation groups was respectively 0.949(95%CI:0.898-1.000)and 0.981(95%CI:0.956-1.000),respectively.The results of the Hosmer Lemeshow test for Model 1 in the modeling and validation groups showed that the model did not deviate from a perfect fit,and the consistency of the calibration curve was good.The area under the ROC curve of Model 2 in the modeling and validation groups was respectively 0.882(95%CI:0.787-0.977)and 0.921(95%CI:0.843-1.000),respectively.Furthermore,the Hosmer Lemeshow test of Model 2 in the modeling and validation groups showed that the model deviated from a perfect fit and the consistency of the calibration curve was poor.Conclusion BMI,combined diabetes,incision type,operation time,incision length,use of steel plate,placement time of drainage tube,CAR,PCT/ALB,IL-6,NLR and ESR of peripheral blood 3 days after operation are the influencing factors of postoperative incision infection in elderly patients with traumatic limb fracture.