2.3. Sample Size Calculation

The statistical programme G*Power 3.1 was used to calculate the sample needed to detect possible differences between independent groups (study patients and control subjects) in relation to the thinning of the different retinal layers. Assuming one case per control, and in accordance with previous studies [41], with the expectation of a mean difference or effect size of 0.38 between cases and controls, 87 CM patients and 87 control subjects would be needed to obtain a confidence level of 95%, an alpha risk (∝) = 0.05, a statistical power of 80%, and a beta risk (β) = 0.20.