The interesting article by Bousquet and associates investigated the risk factors for central serous chorioretinopathy (CSCR) by selecting shift work and sleep disturbances as independent variables. By use of forward stepwise multivariate logistic regression analysis, odds ratios (95% confidence intervals [CIs]) of shift work, steroid use, and recent psychological stress for CSCR were 5 (1.2-20.4), 5.5 (1.1-26.2), and 15.3 (4.1-54.5), respectively. In contrast, the significance of the Insomnia Severity Index for CSCR disappeared by multivariate analysis. I have some concerns on their study.

First, the authors recruited 40 patients with active CSCR and 40 controls, and mentioned that they conducted a prospective case-control study. From the study design, I suspect that they conducted a retrospective study. In addition, the authors selected a forward stepwise multivariate logistic regression analysis. If matching pair was kept during the statistical analysis, conditional logistic regression analysis is preferable. I ask that they please specify the procedure.

Second, the number of the sample was quite limited, which had been recognized by the authors. The 95% CIs of the significant independent variables were wide, and I am afraid of a lack of statistical power in their study. To confirm the risk factors such as shift work, steroid use, and recent psychological stress for CSCR, summing up the number of samples seems indispensable.

Third, Liu and associates conducted a systematic review with meta-analysis to measure risk factors for CSCR and concluded that hypertension, Helicobacter pylori infection, steroid usage, sleeping disturbance, autoimmune disease, psychopharmacologic medication use, and type A behavior were risk factors relating to the occurrence of CSCR. Psychological factors were selected as significant contributors for CSCR occurrence, and Lahousen and associates also recognized that CSCR patients reported more psychosomatic problems, unfavorable stress coping strategies, and critical life events. Shift work is suspected to be related to physical and metal workload, and causal association between shift work and CSCR should be specified by a prospective study or an interventional study.

Finally, Bousquet and associates did not recognize a significant association between working hours and CSCR in their study. There is an inverse dose-response relationship between working hours and sleep duration, and further study is needed with special reference to long working hours.