Fennema P, Lubsen J
Netherlands Institute for Health Sciences, Rotterdam, The Netherlands.
J Bone Joint Surg Br. 2010 May;92(5):701-6. doi: 10.1302/0301-620X.92B5.23470.
Survival analysis is an important tool for assessing the outcome of total joint replacement. The Kaplan-Meier method is used to estimate the incidence of revision of a prosthesis over time, but does not account appropriately for competing events which preclude revision. In the presence of competing death, this method will lead to statistical bias and the curve will lose its interpretability. A valid comparison of survival results between studies using the method is impossible without accounting for different rates of competing events. An alternative and easily applicable approach, the cumulative incidence of competing risk, is proposed. Using three simulated data sets and realistic data from a cohort of 406 consecutive cementless total hip prostheses, followed up for a minimum of ten years, both approaches were compared and the magnitude of potential bias was highlighted. The Kaplan-Meier method overestimated the incidence of revision by almost 4% (60% relative difference) in the simulations and more than 1% (31.3% relative difference) in the realistic data set. The cumulative incidence of competing risk approach allows for appropriate accounting of competing risk and, as such, offers an improved ability to compare survival results across studies.
生存分析是评估全关节置换术结果的重要工具。Kaplan-Meier方法用于估计假体随时间的翻修发生率，但未适当考虑妨碍翻修的竞争事件。在存在竞争性死亡的情况下，该方法会导致统计偏差，曲线将失去其可解释性。如果不考虑不同的竞争事件发生率，就无法对使用该方法的研究之间的生存结果进行有效比较。本文提出了一种替代且易于应用的方法——竞争风险累积发生率。使用三个模拟数据集以及来自406例连续非骨水泥型全髋关节假体队列的真实数据（随访至少十年），对这两种方法进行了比较，并突出显示了潜在偏差的程度。在模拟中，Kaplan-Meier方法高估了翻修发生率近4%（相对差异60%），在真实数据集中高估了超过1%（相对差异31.3%）。竞争风险累积发生率方法能够适当考虑竞争风险，因此，它提供了一种在不同研究之间比较生存结果的更好能力。
