We read with great interest the article by Nicolo and associates, which addresses an important issue in the treatment of chronic central serous chorioretinopathy (CSC). They have used the mean as a measure of central tendency for logarithm of the minimal angle of resolution (logMAR) best-corrected visual acuity (BCVA). But the sample size is small and standard deviation is large. In our opinion, median and interquartile range are better for measurement of central tendency and spread, as they are robust against outliers. Paired and unpaired t test were used by the authors. When the sample size is small (31eyes received half-fluence photodynamic therapy [PDT] and 29 eyes received half-dose PDT), nonparametric tests are much more powerful than their parametric counterparts. Wilcoxon rank-sum test and Mann-Whitney U test would have been more appropriate statistical tests.
The extraocular risk factors that are associated with CSC, such as type A personality, systemic glucocorticoid administration, untreated hypertension, pregnancy, and alcohol intake, were neither evaluated nor matched in the study groups. To prove the study hypothesis showing superiority of a particular procedure over the other, these confounding factors should have been considered to measure the efficacy, especially when the disease recurred in 9 eyes in the half-fluence group and 5 eyes in the half-dose group.
The authors state that no atrophic changes of the retinal pigment epithelium (RPE) in the area of PDT application were seen upon analysis of fundus photographs and fluorescein angiograms. Fundus autofluorescence would have been a better tool for evaluation of RPE function.