目的:探讨踝关节骨折患者术后创伤性关节炎的危险因素。建立风险预测模型。
方法:选取2020年5月至2022年5月收治的550例踝关节骨折患者为研究对象,按照7:3分为建模组(385例)和验证组(165例)。在建模组中,根据手术后是否发生创伤性关节炎,将患者分为发生组(112例)和未发生组(273例)。年龄,体重指数(BMI),性别,吸烟史,糖尿病史,损伤类型,骨折类型,操作时间,体力劳动,开放性损伤,骨质疏松,减少不良,术后负重时间,血管损伤,采用单因素和多因素logistic回归分析踝关节骨折患者发生创伤性关节炎的危险因素;应用R软件建立线图预测模型;应用受试者工作特征(ROC)曲线和校准图验证模型的区分度和一致性。
结果:385例踝关节骨折患者中有112例发展为术后创伤性关节炎,275没有。单因素分析显示年龄差异显著,BMI,骨折类型,操作时间,以上的体力劳动,开放性损伤,两组间骨质疏松程度及减少程度差(P<0.05)。多因素Logistic回归分析显示,年龄(OR=2.887),BMI(OR=4.042),断裂类型(OR=4.244),操作时间(OR=2.665),Ⅱ级以上体力劳动(OR=5.099),骨质疏松症(OR=10.219),复位不良(OR=3.112)是踝关节骨折后创伤性关节炎的独立危险因素(P<0.05)。基于上述风险因素,建立了一个列线图模型来预测踝关节骨折患者术后创伤性关节炎的风险,并进行了内部和外部核查。结果表明,建模组和验证组的校正曲线与理想曲线吻合良好,表明模型预测的术后创伤性关节炎风险与实际风险基本一致。ROC曲线分析结果显示:0.867[(95CI(0.826,0.908)]和0.882[95CI(0.827,0.938)],分别,表明该预测模型具有良好的预测能力。
结论:年龄,BMI,骨折类型,操作时间,Ⅱ级以上的体力劳动,骨质疏松和复位不良都是踝关节骨折患者术后创伤性关节炎的危险因素。基于上述危险因素的预测模型可有效评估踝关节骨折患者术后创伤性关节炎的风险。
OBJECTIVE: To explore risk factors of post-operative traumatic arthritis in patients with ankle fracture,and to establish risk prediction model.
METHODS: Totally 550 patients with ankle fracture treated from May 2020 to May 2022 were selected as research objects and divided into modeling group (385 patients) and verification group (165 patients) according to 7:3. In modeling group,patients were classified as occurrence group (112 patients) and non-occurrence group (273 patients) according to whether traumatic arthritis occurred after opertaion. Age,body mass index(BMI),gender,smoking history,diabetes history,injury type,fracture type,operation time,manual labor,open injury,osteoporosis,poor reduction,postoperative weight-bearing time,vascular injury,and surgical method were recorded; risk factors of traumatic arthritis in ankle fracture patients were analyzed by single factor and multi factor logistic regression analyses; R software was used to build the prediction model of line graph;receiver operating characteristic (ROC) curve and calibration graph were applied to verify the discrimination and consistency of the model.
RESULTS: One hundred and twelve of 385 patients with ankle fracture were developed to post-operative traumatic arthritis,and 275 did not. Univariate analysis showed that there were significant differences in age,BMI,fracture type,operation time,physical labor aboveⅡ,open injury,osteoporosis and poor reduction between two groups (P<0.05). Multivariate Logistic regression analysis showed that age (OR=2.887),BMI (OR=4.042),fracture type (OR=4.244),operation time (OR=2.665),physical labor above gradeⅡ(OR=5.099),osteoporosis (OR=10.219),and poor reduction (OR=3.112) were independent risk factors for traumatic arthritis after ankle fracture (P<0.05). Based on the above risk factors,an nomogram model was established to predict the risk of postoperative traumatic arthritis in ankle fracture patients,and internal and external verification was conducted. The results showed calibration curve of modeling group and verification group showed a good fit between correction curve and ideal curve,indicating that the predicted risk of postoperative traumatic arthritis by the model was basically consistent with actual risk. Area runder ROC curve analysis results showed 0.867[(95%CI(0.826,0.908)] and 0.882 [95%CI(0.827,0.938)],respectively,indicating that the prediction model had good prediction ability.
CONCLUSIONS: Age,BMI,fracture type,operation time,physical labor above gradeⅡ,osteoporosis and poor reduction are all risk factors for post-operative traumatic arthritis in patients with ankle fracture. The prediction model based on the above risk factors could effectively evaluate risk of post-operative traumatic arthritis in patients with ankle fracture.