We read with much interest the original investigation by Richards et al.1 They indicate that daily SSRI use in adults 50 years and older was associated with a 2-fold increased risk of clinical fragility fracture after adjustment for potential covariates. These findings are very interesting, and, as the authors state, may have important public health consequences. However, some comments on this article can be made.
Our first concern pertains to the statistical analysis. It is unclear from the article how many fractures were included in the study. Hence, we cannot judge the validity of the statistical model including numerous covariates. Also, it is not mentioned how the authors dealt with dropout and other types of fracture. Furthermore, a Cox-proportional hazards model can be used when the time to event is known, whereas patients in this study only had to notify whether they had had a fracture in the previous year. Second, the absence of a duration effect relationship is puzzling, especially in the light of an effect through bone strength. In our opinion, a major limitation of the study is the lack of precise exposure data on a daily basis for the entire follow-up period, which would have allowed proper analysis of duration effects as well as attenuation effects after discontinuation.