Effect of the Mediterranean diet on cognition and brain morphology and function: a systematic review of randomized...

Observational studies of the Mediterranean diet suggest cognitive benefits, potentially reducing dementia risk.

We performed the first published review to our knowledge of randomized controlled trials (RCTs) investigating Mediterranean diet effects on cognition or brain morphology and function, with an additional focus on intervention diet quality and its relation to "traditional" Mediterranean dietary patterns.

We searched 9 databases from inception (final update December 2017) for RCTs testing a Mediterranean compared with alternate diet for cognitive or brain morphology and function outcomes.

Analyses were based on 66 cognitive tests and 1 brain function outcome from 5 included studies (n = 1888 participants). The prescribed Mediterranean diets varied considerably between studies, particularly with regards to quantitative food advice. Only 8/66 (12.1%) of individual cognitive outcomes at trial level significantly favored a Mediterranean diet for cognitive performance, with effect sizes (ESs) ranging from small (0.32) to large (1.66), whereas 2 outcomes favored controls. Data limitations precluded a meta-analysis. Of 8 domain composite cognitive scores from 2 studies, the 3 (Memory, Frontal, and Global function) from PREDIMED (PREvención con DIeta MEDiterránea) were significant, with ESs ranging from 0.39 to 1.29. A posttest comparison at a second PREDIMED site found that the Mediterranean diet modulates the effect of several genotypes associated with dementia risk for some cognitive outcomes, with mixed results. Finally, the risk of low-plasma brain-derived neurotrophic factor was reduced by 78% (OR = 0.22; 95% CI: 0.05, 0.90) in those who consumed a Mediterranean diet compared to control diet at 3 y in this trial. There was no benefit of the Mediterranean diet for incident cognitive impairment or dementia.

Five RCTs of the Mediterranean diet and cognition have been published to date. The data are mostly nonsignificant, with small ESs. However, the significant improvements in cognitive domain composites in the most robustly designed study warrant additional research.