How to read a research paper: Ecological validity, apples, and oranges.

Sir,

As far as I could make out, Gupta[1] agrees with my view[2] that, when reading a research paper, it is important to consider the ecological validity of the findings. Gupta[1] suggests that the group not exposed to SSRIs may have contained depressed patients who were not advised SSRIS, or who did not want these medications. This suggestion is indeed correct but, rather than arguing against the apples vs. oranges scenario, it actually supports the distinction. This is because depressed patients, who accept antidepressants despite pregnancy, are likely to be more severely ill and, hence, more likely to suffer from illness-related confounds than patients who are not advised antidepressants or who do not accept these drugs.

Gupta's[1] assertion that alcoholism and depression are often entangled further supports the apples vs. oranges argument in patients who received SSRIs because the co-morbidity would make such patients more different from those who did not receive SSRIs. His contention that depressed women with alcoholism may not have received SSRIs finds no support from the data presented in the original paper.[3]

In his last point, Gupta[1] states that I postulated that psychological and material support received by women in India is quantitatively inferior to that received by women in the West; what I had stated was actually exactly the opposite, as evident in the final paragraph of my paper,[2] before the discussion section. Finally, Gupta[1] observes that my statement “underlying psychiatric condition (depression) was likely responsible for both SSRI prescription and alcohol use” does not capture the full essence of the possible scenarios that would affect the interpretation of results; however, he does not support his point with an illustration of possible scenarios that differ from or expand upon what I offered.