We read with great interest the article published by Pakzad-Vaezi and associates. They concluded that atypical antipsychotic use has a protective effect on the rate of cataract surgery. We have a few comments regarding the potential limitations of the study.

First, it is well known that the decision to perform cataract surgery depends to a large extent on the willingness of the patient, his or her symptoms, and his or her occupational or vocational needs. The rate of cataract surgery in a population therefore is not representative of the incidence of cataract in the same population. Two cataracts may be anatomically identical; however, one may be symptomatic and the other not. Alternatively, two cataracts may produce similar symptoms, but one patient may elect to be operated on, because his cataract is bilateral or because of his occupational or vocational needs, whereas another patient may not require surgery because the visual acuity of his fellow eye is still good. Thus, the most relevant definition for cases to evaluate the association between antipsychotic and cataract formation should be a first diagnosis of cataract, and not a first cataract procedure.

In the Results, the authors should provide both unadjusted and adjusted rate ratios for all well-known risk factors for cataract formation to verify the results of previous epidemiologic studies. These rate ratios then can be used as positive controls of their analysis. Merely providing the adjusted rate ratio for oral steroids, as a positive control, is not sufficient. What about other risk factors, such as hypertension or diabetes? In particular, the authors should verify their previously published results, such as those relating to the increased risk of cataract for patients taking selective serotonin reuptake inhibitors. Indeed, it seems clear from the present study that the rate of patients taking selective serotonin reuptake inhibitors is not different between cases and controls (4.3% vs. 4.2%). The authors therefore should explain the discrepancy between these results and those of their previous report.

Furthermore, the authors did not include in their analysis 2 important confounding factors, which are smoking and diabetes, reducing the statistical relevance and the clinical impact of their analysis. Regarding diabetes, evaluating the level of control as measured by levels of hemoglobin A1c is more relevant than the rate of patients taking antidiabetic treatments.

In the Discussion, the authors argue that after 70 years of age, atypical antipsychotics were probably not prescribed for acute psychosis, and therefore the indication is likely not a confounding factor. However, 2 recent large cross-sectional studies showed that, among elderly patients, diagnosis of dementia was an independent risk factor for antipsychotic use. In this context, the indication for atypical antipsychotics is a possible major confounding factor in the evaluation of the risk of cataract surgery; therefore, it would have been of great interest to specify in this article the medical indication for the atypical antipsychotics use.

Considering the above remarks, we believe it is an overstatement to say on the basis of only 1 retrospective observational study that a protective association was established.