We thank Dr Çalışkan and his co-workers for their beneficial comments.1

Distribution of all retrobulbar blood flow variables was normal except in superior ophthalmic vein (SOV) maximal blood, which had a P value of 0.041 in the Kolmogorov−Smirnov test. However, using Mann−Whitney U-test did not change the result (P=0.008).

Although the equality of variances is an assumption of ANCOVA, it should hold for the residual of this regression model (after adjustment for the confounder variables). However, it has been shown that ANCOVA would obtain the correct type I error even with unequal sample size. Indeed, the estimator of the effect at the observed mean is not different between equal and unequal variance assumptions.2 Also, we used type III sum of squares, which is more robust in obtaining the variance of treatment effect.

In our subjects, all vessels were successfully detected except one SOV in smokers. These vessels have normal anteroposterior flow. Some of the previous studies did not mention a reverse flow in orbital vessels in thyroid eye disease (TED).3, 4 However, undetectable SOV in a patient may be attributed to more reduced blood flow. Furthermore, high percentage of reverse flow in the SOV of patients with TED was observed in dysthyroid optic neuropathy.5 Thus, in our study, lack of reverse retrobulbar blood flow might probably be the result of having no patients with dysthyroid optic neuropathy or other most severe forms of TED.