Letter by Kuno et al Regarding Article, “Heart Team/Guidelines Discordance Is Associated With Increased Mortality: Data From a National Survey of Revascularization in Patients With Complex Coronary Artery Disease”
Letter by Kuno et al Regarding Article, “Heart Team/Guidelines Discordance Is Associated With Increased Mortality: Data From a National Survey of Revascularization in Patients With Complex Coronary Artery Disease”
复制标题
Kuno 等人关于文章“心脏团队/指南不一致与死亡率增加相关：来自复杂冠状动脉疾病患者血运重建全国调查的数据”的信函
DOI：
10.1161/circinterventions.121.010649
复制
发表时间：
2021
期刊：
影响因子：
--
通讯作者：
Kohsaka Shun
中科院分区：
文献类型：
--
作者：
Kuno Toshiki;Ikemura Nobuhiro;Kohsaka Shun
我们饶有兴趣地阅读了Witberg等人在2021年1月出版的《循环：心血管介入》杂志上发表的论文，1关于心脏团队或临床实践指南不一致与死亡率增加之间的关系。2在他们的分析中，作者利用考克斯比例回归分析来比较心脏团队或临床实践指南一致组与不一致组的死亡率。然而，图1中2组的Kaplan-Meier曲线相互交叉，这表明违反了比例风险假设。在这种情况下，汇总风险比（HR）为2.05意味着生存曲线上的平均HR，忽略了特定时期的HR 3;与不一致组相比，一致组的事件发生率更高，特别是在大约前180天，可能是由于冠状动脉旁路移植术后早期死亡率高。4因此，图示结果强调了特定时期HR的本质差异，作者无法分析。作为替代方案，显示特定时期的HR（例如，里程碑方法）可能更好地证明他们的发现。3，5此外，差异也可以计算为研究期随访开始至结束期间2个生存曲线之间的面积（限制平均生存时间），这不需要比例风险或任何其他模型假设。限制性平均生存时间的比较信息量更大; X年生存率增加平均比较心脏团队/指南一致性与不一致组。限制性平均生存时间可以使用SAS中的RMSTEG proc或R中的survRM 2进行分析。
We read with interest the paper in the January 2021 issue of Circulation: Cardiovascular Interventions by Witberg et al, 1 regarding the association between heart team or clinical practice guidelines discordance with increased mortality. 2 Within their analyses, the authors leveraged the Cox proportional regression analysis to compare the mortality of the heart team or clinical practice guidelines agreement versus discordance group. However, the Kaplan-Meier curves of the 2 groups in Figure 1 have crossed each other, which indicates the violation of the proportional hazard assumption. In such a case, the summary hazard ratio (HR) of 2.05 would mean that the average HRs over the survival curves, and ignores the period-specific HRs3; there was a higher rate of events in the agreement arm compared with the discordance arm especially in the approximately first 180 days, presumably due to high proportions of early mortality after coronary artery bypass grafting. 4 Hence, the illustrated result underscores an essential difference in the period-specific HRs, which could not be analyzed by the authors. As the alternatives, showing period-specific HRs (eg, the landmark methods) might better demonstrate their findings. 3, 5Further, the difference can also be calculated as the area between the 2 survival curses during the beginning to the end of follow-up of the study period (restricted mean survival time), which does not require the proportional hazard or any other model assumptions. Comparison of restricted mean survival times is more informative; X-year survival gain in average comparing heart team/guidelines agreement versus discordance arm. Restricted mean survival time can be analyzed using the RMSTREG proc in SAS or survRM2 in R.