We would like to thank Bigby1 for his commentary in this issue of the Archives on our article.2 Bigby's main concerns surround our strict case definition that may have underestimated the number of patients diagnosed with depression. We agree with Bigby that the underestimate likely occurred, but strict criteria were necessary to ascertain that we had real cases of depression. Defining cases by depression diagnostic codes alone is a questionable practice, given the unknown validity of these codes in the Régie de l'Assurance Maladie du Québec or any other administrative database. Similarly, using the presence of antidepressant prescriptions alone to define cases of depression would have led to the inclusion of noncases because antidepressants are used increasingly for indications other than depression. Thus, requiring that cases have either a diagnosis or hospitalization for depression and presence of an antidepressant prescription resolved these issues. In fact, we believe that the strict selection criteria represent one of the strengths of our study. Indeed, these criteria ensured the specificity of our case series, which resulted in a study of higher internal validity than the previous ones.3,4 Internal validity (unbiased estimates) is of more importance, in this case, than external validity. Certainly a biased estimate will never be generalizable.
Azoulay L, Bérard A. Does Isotretinoin Increase the Risk of Depression?—Reply. Arch Dermatol. 2008;144(9):1234-1235. doi:10.1001/archderm.144.9.1234