The authors appreciate the comments of Sanharawi and associates with respect to our recent paper. We agree that using the first date of cataract diagnosis for our case definition would be preferable when examining the etiologic relationship between atypical antipsychotics and cataracts. The primary concern in our regional practice patterns centered on the lack of standardized diagnostic criteria for ophthalmologists when evaluating lenticular opacity. In many instances, it may have been solely for billing reasons and bore no relationship with visual disability. As such, we chose to use first cataract procedure as our case definition to reflect the formation of a visually significant cataract for that patient in question, which accounts for factors mentioned, such as vocational needs.

Sanharawi and associates also suggest conducting separate, fully adjusted models on risk factors before deciding on whether a variable may be considered a control group in the study. We argue against this approach, which would treat a risk factor as the main study exposure. In epidemiologic studies, the exposure usually is defined based on an a priori hypothesis with careful consideration of the time to disease onset. Risk factors are defined more crudely at baseline, and thus a lack of association between a risk factor and disease in a particular study does not necessarily exclude the presence of an association. We chose oral corticosteroid users as a positive control because it is well established (both from our study and others) that they increase the risk of cataracts. Selective serotonin reuptake inhibitors did not increase the risk of cataracts in this study. We found a slight increase in the risk associated with selective serotonin reuptake inhibitors (adjusted rate ratio: 1.15; 95% confidence interval: 1.08-1.23) in a different cohort of patients with a history of cardiovascular disease. We agree that the risk of selective serotonin reuptake inhibitors and cataracts must be investigated in future studies.

With respect to not controlling for smoking, we believe this was addressed adequately in the Discussion. With respect to controlling for diabetes, antidiabetic medications were used to identify those with diabetes, the diagnosis of which is itself a cataract risk factor. Antidiabetic medications have been used previously as a variable related to cataract. Glucose control was not our variable of interest.

We agree that those patients being treated for end-stage dementia with antipsychotics may not have equal access to cataract surgery given issues with consent, cooperation, and occasionally anesthesia. As such, we appreciate the suggestion to correlate antipsychotic use with medications for dementia. We note that in our region and others, mild-to-moderate dementia is not a deterrent to cataract surgery referral, possibly with greater access given safety recommendations and possible cognitive improvement.

Our conclusions should be interpreted in the context of retrospective study limitations, particularly those discussed. As mentioned, the major contribution of our findings was to generate a hypothesis that requires further testing under more definitive circumstances, such as a future prospective study with comprehensive variable control. We believe this study is an important contribution to the scientific method, offering hypotheses both to clinicians and basic scientists alike.