We read with interest the article ‘Effect of smoking on retrobulbar blood flow in thyroid eye disease (TED)’.1 The results are interesting but we have a few concerns and comments.

First, the authors used the parametric tests to compare continuous variables between the groups. Parametric tests assume that continuous variables have a normal distribution. Violation of these assumptions changes interpretation of the results and the conclusion of the study. The authors did not state about the distribution of variables. When we look at the variables carefully, the standard deviation (SD) of the mean velocity values of the superior ophthalmic vein (SOV) is slightly high. For example, the SD of the mean maximal velocity of SOV in the non-smoker group is 2.14, close to half of the mean value (5.27). As a rule, a variable with a SD that is larger than one half of the mean value is accepted as non-normally distributed.2 Second, the sample sizes between groups (n=16 in the smoker group and n=35 in the non-smoker group) were unequal; therefore, it is likely that the variances will be unequal.3 The assumptions for ANCOVA include that the outcome variable is normally distributed and the variances are similar between groups (homogeneity of variance). We suggest that the authors mention about the distributions of variables and gives the ranges of the outcome variables.

Lastly, it is known that reversed blood flow or absence of SOV flow may occur in patients with TED. The reversed SOV flow rate was reported to be about 13–15% in the series.4 We wonder whether the authors excluded these cases or these were not encountered in their study.