股骨转子间骨折内固定失效后的全髋关节置换:隐性失血的模型预测

Total hip arthroplasty after failure of internal fixation for intertrochanteric fractures:model prediction of occult blood loss

背景:股骨转子间骨折在老年人和骨质疏松患者中尤为多见,治疗常需内固定手术来稳定骨折并促愈合。但内固定手术偶可失效,引发骨折不愈合、畸形愈合或再骨折等严重问题。目的:探讨股骨转子间骨折内固定后失效行人工全髋关节置换的效果及术后隐性失血的影响因素。方法:收集2019年5月至2022年9月于阳泉市第一人民医院治疗且随访资料完整的股骨转子间骨折内固定失效患者86例,采用人工全髋关节置换治疗,记录患者相关临床指标,对患者手术前后的Harris评分、疼痛目测类比评分以及SF-36评分进行对比分析。对全髋关节置换后隐性失血的影响因素进行单变量和多变量Logistic回归分析,建立Logistic回归模型和神经网络模型预测隐性失血,通过绘制受试者工作特征曲线对比2种模型的预测效能和准确性。结果与结论:(1)86例股骨转子间骨折患者在内固定失效后,均通过人工全髋关节置换手术成功治疗;(2)与术前相比,患者在术后3,6,12个月的Harris评分、目测类比评分、SF-36评分及患髋主动活动幅度均显著改善(P<0.05);(3)高隐性失血组的年龄、高血压比例、糖尿病比例、骨质疏松比例、手术时间、总失血量、术中失血量、术后引流量及血红蛋白下降量均高于低隐性失血组(P <0.05);低隐性失血组的骨水泥型假体构成比、前方入路构成比及2-3级髋关节间隙高于高隐性失血组(P <0.05);两组在性别比例、体质量指数、慢性支气管炎比例、侧别比例、麻醉方式及髋臼杯直径方面差异无显著性意义(P> 0.05);(4)通过多因素Logistic回归分析,年龄、糖尿病、骨质疏松、假体类型、手术入路、髋关节间隙、总出血量、术中出血量、术后引流量及血红蛋白下降量是全髋关节置换后高隐性失血的危险因素(P <0.05);(5)将影响因素纳入到Logistic回归模型和神经网络模型预测隐性失血,绘制2个模型受试者工作特征曲线的曲线下面积分别为0.882和0.923,灵敏度分别为0.879和0.886,特异度分别为0.854和0.908,约登指数分别为0.733和0.794,准确性分别为0.867和0.897;2个模型的曲线下面积、约登指数和准确性均为神经网络模型更高;(6)提示全髋关节置换在治疗股骨转子间骨折内固定术后失效方面具有显著的临床效果,髋关节活动度恢复较好;年龄、糖尿病、骨质疏松、假体类型、手术入路、髋关节间隙、总出血量、术中出血量、术后引流量及血红蛋白下降量是人工全髋关节置换后高隐性失血风险的重要因素,基于危险因素构建的Logistic回归模型和神经网络模型对隐性失血的预测结果相差不大,神经网络模型更优。

BACKGROUND:Intertrochanteric fractures of the femur are particularly common in elderly and osteoporosis patients,and treatment often requires internal fixation surgery to stabilize the fracture and promote healing.However,internal fixation surgery may occasionally fail,leading to serious problems such as nonunion of fractures,malunion,or re-fracture.OBJECTIVE:To explore the effect of artificial total hip arthroplasty on postoperative failure of internal fixation for intertrochanteric fractures of the femur and the influencing factors of postoperative occult blood loss.METHODS:Totally 86 patients with failed internal fixation of intertrochanteric fractures of the femur who were treated in First People’s Hospital of Yangquan City from May 2019 to September 2022 and had complete follow-up data were collected.Using artificial total hip arthroplasty for treatment,relevant clinical indicators of the patient were recorded,and Harris score,visual analog scale score,and SF-36 score were compared and analyzed before and after surgery.Univariate and multivariate logistic regression analysis was conducted on the factors affecting occult blood loss after total hip arthroplasty.Logistic regression models and neural network models were established to predict occult blood loss.The predictive performance and accuracy of the two models were compared by drawing receiver operating characteristic curves.RESULTS AND CONCLUSION:(1)Totally 86 patients with intertrochanteric fractures of the femur were successfully treated with artificial total hip arthroplasty after internal fixation failure.(2)There were statistically significant differences in Harris score,visual analog scale score,SF-36 score,and hip active range of motion between patients before and 3,6,and 12 months after surgery(P<0.05).(3)The age,hypertension ratio,diabetes ratio,osteoporosis ratio,operation time,total blood loss,intraoperative blood loss,postoperative drainage volume,and hemoglobin decrease in the high occult blood loss group were significantly higher than those in the low occult blood loss group(P<0.05).The proportion of bone cement type prostheses,anterior approach,and 2-3 grade hip joint space in the low occult blood loss group were higher than those in the high occult blood loss group(P<0.05).There was no statistically significant difference between the two groups in terms of gender ratio,body mass index,proportion of chronic bronchitis,proportion of sides,anesthesia method,and acetabular cup diameter(P>0.05).(4)Through multivariate logistic regression analysis,age,diabetes,osteoporosis,prosthesis type,surgical approach,hip joint space,total blood loss,intraoperative blood loss,postoperative drainage volume,and hemoglobin decrease were the risk factors for high occult blood loss after total hip arthroplasty(P<0.05).(5)After incorporating influencing factors into the logistic regression model and neural network model for predicting occult blood loss,the receiver operating characteristic curves of the two models were plotted with area under curve values of 0.882 and 0.923,sensitivity values of 0.879 and 0.886,specificity values of 0.854 and 0.908,Youden index values of 0.733 and 0.794,and accuracy values of 0.867 and 0.897,respectively.The area under curve,Youden index,and accuracy of both models were higher than those of neural network models.(6)It is concluded that total hip arthroplasty has a significant clinical effect on treating postoperative failure of internal fixation for intertrochanteric fractures of the femur,with good recovery of hip joint range of motion.Age,diabetes,osteoporosis,prosthesis type,surgical approach,hip joint space,total blood loss,intraoperative blood loss,postoperative drainage volume,and hemoglobin decrease are important factors for the high risk of occult blood loss after total hip arthroplasty.Logistic regression model and neural network model based on risk factors have little difference in the prediction results of occult blood loss,and neural network model is higher.