To the Editor I read with great interest the article by Zander et al,¹ which aimed to present a prediction model for edema resolution in patients with Fuchs dystrophy. However, some of their findings seem to rely on statistical assumptions that may not be justified in this study.

In Figure 2,¹ the lower plots, which are Bland-Altman plots, and in the related text, the 95% limits of agreement between predicted and observed edema are shown (orange lines) with reference to the 1986 publication by Bland and Altman² regarding this method of assessing agreement. However, to conclude 95% limits of agreement using the Bland-Altman method, the differences (y-axis) must be normally distributed.² In this study, the authors do not indicate whether this was the case. The differences may not be normally distributed, noting in Figure 2A¹ that there are 5 dots above the higher orange line and 2 below the lower one. Since this derivation cohort consisted of 100 eyes, assuming normality of the differences, no more than 5 eyes should lie outside the 95% limits of agreement, whereas in this study, a total of 7 dots were outside the limits and do not appear to have been divided evenly between the lower and upper margins as one would expect in a normal distribution.

致编辑，我非常感兴趣地阅读了 Zander 等人的文章¹ ，该文章旨在提出 Fuchs 营养不良患者水肿消退的预测模型。然而，他们的一些发现似乎依赖于在本研究中可能不合理的统计假设。

在图 2, ¹中，下图是 Bland-Altman 图，在相关文本中，参照 Bland 和 Altman ²的 1986 年出版物，显示了预测水肿和观察到的水肿之间 95% 的一致性极限（橙色线）关于这种评估协议的方法。但是，要使用 Bland-Altman 方法得出 95% 的一致性限制，差异（y 轴）必须呈正态分布。²在这项研究中，作者没有说明是否是这种情况。差异可能不是正态分布的，请注意图 2A ¹在较高的橙色线上方有 5 个点，在较低的橙色线下方有 2 个点。由于这个推导队列由 100 只眼睛组成，假设差异正常，不超过 5 只眼睛应该在 95% 的一致性限制之外，而在本研究中，共有 7 个点在限制之外并且似乎没有正如人们在正态分布中所期望的那样，在下边缘和上边缘之间平均分配。