In Reply: Dr Feng raises the question of whether including incident dementia in the linear mixed models, rather than baseline MCI, would have been the better approach to differentiate normal age-associated cognitive change from decline associated with a dementia prodrome. It would be incorrect in evaluating effect of treatment in a randomized clinical trial to adjust for another outcome (dementia) that is intrinsically related to the outcome of interest in the analysis (cognitive decline). Furthermore, it is unclear what would be the clinical relevance with respect to informing choice of treatment if, for instance, G biloba had been shown to have an effect, but only among those who would go on to develop dementia in the next 7 years. Knowledge of a future clinical event gained in a trial and applied retrospectively to the study group at baseline cannot be generalizable to real-world populations and settings.
Snitz BE, Arnold AM, DeKosky ST. Ginkgo biloba and Cognitive Decline—Reply. JAMA. 2010;303(15):1477-1478. doi:10.1001/jama.2010.435