Sir,

We thank Drs Shah et al for their interest in our paper on obstructive sleep apnoea (OSA) and optic neuropathy.1 Here, we attempt to elaborate on a few points mentioned by their letter to the editor.

In our study, we analysed each eye independently because normal tension glaucoma and various forms of optic neuropathies like nonarteritic ischaemic optic neuropathy, which may be related to OSA, often present unilaterally and asymmetrically, especially in the early stage.2, 3, 4 Anyway, as suggested by Dr Shah et al, we re-analysed the data by collapsing the data into individuals and analysed using Wilcoxon's test (a nonparametric test). The differences of visual field indexes between OSA and control arms were still found to be statistically significant (P=0.021 for the difference between mean deviation in the two arms, and P=0.001 for the difference between pattern standard deviation in the two arms, respectively).

Different statistical tools help researchers to analyse data in different aspects and present them at different angles. Shah et al suggested that we should present the difference of visual function indices in means with confidence interval (CI). However, in our series, only a small proportion of OSA patients developed optic neuropathies. Comparing their mean may not yield conclusive information. Therefore, as an alternative, we chose to present the visual field index in a graphical method showing the distribution of data, which gave a clear overall picture demonstrating that a small peak of suboptimal visual field indices was found in the OSA group. Concerning the values of CI, we did present the data in the footnotes of all tables and figures.1

Finally, we would like to thank Dr Shah et al for mentioning the importance of study design methodology and terminology. We share their view that these are important aspects in any research.