Our study has some potential limitations that should be mentioned. Because it was conducted in middle-aged
subjects and older adults with a high risk of CVD, our findings cannot be extrapolated to the general population.
Moreover, PREDIMED is a clinical trial and the sub-analysis has been conducted as an observational cohort, so
although the statistical analyses have been adjusted for each intervention group, a potential residual effect of
dietary intervention on the final results cannot be discounted. Also, although other major potential confounders
have been adjusted in the current approach, we cannot completely rule out the possibility of residual
confounding from measured and unmeasured factors. The lack of specific measurements of body composition
and protein metabolism, such as DEXA and urinary nitrogen, could limit our findings. Finally, it should also be
taken into account that the E% evaluated by quintiles of protein in the PREDIMED cohort fluctuates from 13.9 to
19.5, a narrower range than has been found by other studies based on hyperproteic diets. In contrast, our study
is the first to have evaluated the efficacy and safety of protein consumption at the same time. It has a large
sample of subjects, followed up subjects for a medium-long period, and accurately ascertained CV events and
death. We also used the cumulative average method, which corrects for fluctuating values of protein
consumption, and changes in body weight or fat distribution during the follow-up. Finally,we used macronutrient
energy substitution models to analyse the association between dietary protein intake and several outcomes when
this intake replaced carbohydrates or fats.