Lack of statistical power and refractive outcomes.

In their excellent study, Sahin et al. investigated the effect of diabetes mellitus on various corneal biomechanical parameters, as measured by the Ocular Response Analyzer (ORA; Reichert, Inc., Depew, NY). The rationale of their study is very interesting. To my surprise, the authors showed a decrease in corneal hysteresis (CH) rather than the expected increase. Several factors suggest that diabetes mellitus would actually enhance corneal biomechanics by an increase in the crosslinking rate: First, an earlier retrospective study showed a lower incidence of keratoconus in diabetic patients, suggesting that corneal biomechanics are enhanced in diabetic corneas. Second, the nonenzymatic glycosylation of proteins (Maillard reaction) that is prominent in diabetes mellitus, results in the formation of advanced glycosylation end products (AGEs). AGEs induce cross-links between connective tissue collagen and increase tissue rigidity, especially in the presence of glucose. Similar to diabetes, tobacco smoking represents a source of AGEs, and moreover, by-products of cigarette smoke, such as nitrogen oxides, nitrite, and formaldehyde, induce cross-links between collagen fibers. A recent epidemiologic study showed that the incidence of keratoconus in smokers is considerably lower than in the nonsmoking population, and we have recently performed a prospective comparative case series to investigate the effect of chronic tobacco smoking on corneal biomechanics using the ORA. Our results showed that chronic smoking increases corneal rigidity in a statistically significant manner. The study by Sahin et al. shows the opposite and was performed in Turkey. From 1990 to 1999, Turkey had the second highest growth rate in cigarette consumption in the world, and in 1999, Turkey accounted for 2.2% of the total world cigarette consumption. Therefore, accounting for the smoking status of the participants in this study would be essential for the outcome and might have significantly altered the results. The authors could not be aware of the influence chronic tobacco smoking might have on their results, because at the time of publication of their study our paper, now published, was in press. I therefore suggest that Sahin et al. determine the smoking status of their patients and perform the statistical analysis in light of their findings. Farhad Hafezi