牛津膝关节评分中「整体疼痛」「行走跛行」和「膝关节打软腿」是后续翻修的最强预测因素
该研究基于新西兰关节登记系统中1999至2019年所有初次全膝关节置换术(TKA)和单髁膝关节置换术(UKA)且术后6个月、5年或10年有牛津膝关节评分(OKS)的病例,采用逻辑回归和受试者工作特征分析评估各OKS问题对后续翻修的预测能力。结果显示,「整体疼痛」「行走跛行」「膝关节打软腿」三个问题组成的简化模型预测UKA翻修的诊断能力优于完整OKS,预测TKA翻修的能力与完整OKS相当;疼痛域在5年和10年预测两种手术后续翻修的能力更优。
要点
- 「整体疼痛」「行走跛行」「膝关节打软腿」是后续翻修的最强预测问题。
- 简化模型预测UKA翻修6个月AUC 0.80优于完整OKS 0.78(P<.01),5年0.81优于0.77(P=.02)。
- 疼痛域在5年和10年预测两种手术后续翻修的能力均优于功能域。
核心结论
- 「整体疼痛」「行走跛行」「膝关节打软腿」三个问题是后续翻修的最强预测因素。
- 随访中关注这些问题的低分可能有助于及时识别翻修风险最高的患者。
研究设计与样本
- 纳入新西兰关节登记系统中1999年至2019年所有初次全膝关节置换术(TKA)和单髁膝关节置换术(UKA)且有术后6个月、5年或10年牛津膝关节评分(OKS)的病例。
- 6个月时TKA 27,708例、UKA 8,415例;5年时TKA 11,519例、UKA 3,365例;10年时TKA 6,311例、UKA 1,744例。
- 采用逻辑回归和受试者工作特征分析评估预测模型。
主要发现
- 由「整体疼痛」「行走跛行」「膝关节打软腿」三个问题组成的简化模型预测UKA翻修的诊断能力优于完整OKS:6个月AUC 0.80 vs 0.78(P<.01),5年AUC 0.81 vs 0.77(P=.02)。
- 简化模型预测TKA翻修在6个月(AUC 0.77 vs 0.76)、5年(0.78 vs 0.75)、10年(0.76 vs 0.73)与完整OKS诊断能力相当(均无显著差异)。
- 简化模型预测UKA 10年翻修的AUC 0.80 vs 0.77,无显著差异。
- 疼痛域在5年和10年预测两种手术后续翻修的诊断能力均优于功能域。
关键数值与推荐
- 简化模型预测UKA翻修6个月AUC 0.80 vs 完整OKS 0.78(P<.01)。
- 简化模型预测UKA翻修5年AUC 0.81 vs 完整OKS 0.77(P=.02)。
- 简化模型预测TKA翻修6个月AUC 0.77 vs 0.76、5年0.78 vs 0.75、10年0.76 vs 0.73,均无显著差异。
- 简化模型预测UKA翻修10年AUC 0.80 vs 0.77,无显著差异。
临床怎么用
- 随访中关注「整体疼痛」「行走跛行」「膝关节打软腿」三个问题的低分可能有助于及时识别翻修风险最高的患者。
局限性
- 原文未报告局限性。
以上解读依据论文摘要(全文未开放获取或许可不允许翻译转载)。
原文
BACKGROUND: The Oxford Knee Score (OKS) is used to measure knee arthroplasty outcomes; however, it is unclear which questions are more relevant. Our aims were to (1) identify which OKS question(s) were the strongest predictors of subsequent revision and (2) compare the predictive ability of the "pain" and "function" domains. METHODS: All primary total knee arthroplasties (TKAs) and unicompartmental knee arthroplasties (UKAs) in the New Zealand Joint Registry between 1999 and 2019 with an OKS at 6 months (TKA n = 27,708; UKA n = 8,415), 5 years (TKA n = 11,519; UKA n = 3,365) or 10 years (TKA n = 6,311; UKA n = 1,744) were included. Prediction models were assessed using logistic regressions and receiver operating characteristic analyses. RESULTS: A reduced model with 3 questions ("overall pain," "limping when walking," "knee giving way") showed better diagnostic ability than full OKS for predicting UKA revision at 6 months (area under the curve [AUC]: 0.80 versus 0.78; P < .01) and 5 years (0.81 versus 0.77; P = .02), and comparable diagnostic ability for predicting TKA revision at all time points (6 months, 0.77 versus 0.76; 5 years, 0.78 versus 0.75; 10 years, 0.76 versus 0.73; all not significant), and UKA revision at 10 years (0.80 versus 0.77; not significant). The pain domain had better diagnostic ability for predicting subsequent revision for both procedures at 5 and 10 years. CONCLUSION: Questions on "overall pain", "limping when walking", and "knee giving way" were the strongest predictors of subsequent revision. Attention to low scores from these questions during follow-up may allow for prompt identification of patients most at risk of revision.
引用
- Tay ML, Monk AP, Frampton CM, et al. The Strongest Oxford Knee Score Predictors of Subsequent Revision are "Overall Pain," "Limping When Walking," and "Knee Giving Way". J Arthroplasty. 2023. PMID: 36898485. doi:10.1016/j.arth.2023.03.001
供执业医师工作参考,不替代临床判断;数值与分级以指南原文为准。