初次TKA术后早期患者满意度预测模型的构建、内部验证及比较

Construction,Internal Validation and Comparison of Satisfaction Prediction Models for Patient in the Early Stage After Primary TKA

目的 构建初次人工膝关节置换术(TKA)后早期患者满意度预测模型并进行内部验证。比较Van Onsem、Kunze模型与本预测模型的有效性。方法 前瞻性收集2021年1月-6月间因膝骨关节炎于天津医院关节外科行初次TKA手术患者277例。采集患者术前相关信息及术后6个月KSS满意度评分。以TKA术后6个月KSS满意度作为结局指标,TKA术前因素作为预测指标。通过单因素回归分析、Lasso回归进行变量筛选,利用多因素Logistic回归构建TKA术后早期满意度预测模型。计算预测模型的AUC、C指数、绘制Calibration曲线、DCA曲线。利用本中心的数据集对Kunze、Van Onsem模型进行外部验证,与本次构建的模型进行比较。结果 277例患者中术后6个月满意232例(83.75%),不满意45例(16.25%)。基于WOMAC评分、KSS功能评分、年龄、BMI、GAD-7、OARSI外侧关节间隙分级、OARSI股骨内侧骨赘分级7种因素构建TKA术后满意度预测模型。本预测模型的AUC为0.77,校正后的AUC为0.83(95%CI:0.763~0.833),Kunze模型的AUC值为0.61(95%CI:0.52~0.70),Van Onsem模型的AUC值为0.71(95%CI:0.63~0.79)。结论 本次构建的TKA术后早期满意度预测模型在中国北方群体中具有良好的准确性和区分度,临床实用价值较高。另外,Van Onsem与Kunze的预测模型不适用于本研究群体。

Objective To construct a predictive model of satisfaction in the early stage after primary total knee arthroplasty(TKA)and conduct internal validation.In addition,to compare the effectiveness of Van Onsem,Kunze model and this prediction model.Methods A total of 277 patients with knee osteoarthritis who underwent primary TKA in the Joint Surgery Department of Tianjin Hospital from January 2021 to June 2021 were prospectively collected.The preoperative information and KSS satisfaction score at 6 months after operation were collected.KSS satisfaction at 6 months after TKA was used as the outcome index,and preoperative factors of TKA were used as predictors.Univariate regression analysis and Lasso regression were used to screen variables,and multivariate logistic regression was used to construct a prediction model of early satisfaction after TKA.The AUC,C index,Calibration curve and DCA curve of the prediction model were calculated.The Kunze and Van Onsem models were externally validated using the data set of our center,and compared with the model constructed in this paper.Results Of the 277 patients,232 patients(83.75%)were satisfied and 45 patients(16.25%)were dissatisfied at 6 months after operation.Based on WOMAC score,KSS function score,age,BMI,GAD-7,OARSI lateral joint space classification and OARSI medial femoral osteophyte classification,a satisfaction prediction model after TKA was constructed,and the AUC of the prediction model was 0.77,the corrected AUC was 0.83(95%CI:0.763-0.833),the AUC of the Kunze model was 0.61(95%CI:0.52-0.70),and the AUC of the Van Onsem model was 0.71(95CI:0.63-0.79).Conclusion The prediction model of early satisfaction after TKA constructed in this study has good accuracy and discrimination in the northern Chinese population,and has high clinical practical value.In addition,Van Onsem and Kunze's prediction model does not apply to this study group.