Mediterranean diet

    The Mediterranean diet ‘reduces brain shrinkage’? The study didn’t prove any such thing

    6 January 2017

    New research has claimed that a Mediterranean diet can protect the brain from shrinking in old age (however, see our analysis below).

    The study, published in the journal Neurology, found that pensioners in Scotland who tended to follow more of a Mediterranean diet appeared to have reduced brain shrinkage between the ages of 73 and 76.

    The diet of 401 healthy people was assessed on the basis of a questionnaire. Each subject was then given a score based on how closely they followed a Mediterranean-type diet, which in practice meant eating more fruit, vegetables and fish and limiting consumption of meat and dairy.

    The researchers carried out brain scans when the subjects were aged 73 and 76.

    It was claimed that those who followed the diet most closely retained ‘significantly’ greater brain volume over a period of three years. The researchers also concluded that diet accounted for about half the variation in brain volume observed in study participants and that, contrary to previous findings, fish consumption appeared to make no difference.

    The study’s lead author, Dr Michelle Luciano, said: ‘As we age, the brain shrinks and we lose brain cells, which can affect learning and memory. This study adds to the body of evidence that suggests the Mediterranean diet has a positive impact on brain health.

    ‘In our study, eating habits were measured before brain volume was, which suggests that the diet may be able to provide long-term protection to the brain. Larger studies are needed to confirm these results.’

    Instant analysis
    I doubt very much whether this study is worth getting excited about. Here’s why.

    (1) It doesn’t measure brain function, but just the volume of the brain from images.

    (2) Although it’s said that the subjects were followed over time, their brain volumes were measured at age 73 and 76 — a rather small time difference.

    (3) Each subject was categorised as belonging to a group that followed more closely the Mediterranean diet, or a group that followed it less closely. But diet was not monitored during the study. Rather, the category was determined on the basis of self-reported (notoriously unreliable) eating habits, on a single occasion before measurements began.

    (4) Because the subjects were not allocated randomly to diets, it’s impossible to tell whether any association between diet and brain volume is causal. This matters because if it is not causal then changing diet will have no effect.

    (5) It’s claimed that the diet was correlated with total brain volume, though not with either cortical thickness or with grey matter volume. But it’s entirely possible that the results are no more than statistical flukes. Table 2 contains 27 different tests of statistical significance, of which only three gave P values that were (a bit) less than 0.05. But you’d expect one in 20 results to appear to be ‘statistically significant’ by chance alone. It’s usual to apply a correction when many statistical tests are done and the authors say that if such a correction is done, none of the effects reach ‘statistical significance’, a crucial fact that is mentioned neither in the abstract, nor in the journal’s press release.

    (6) Even if the tests of significance are taken at face value, P values that are not much less than 0.05 do not allow one to conclude that an effect is real, but rather that the question might be worth another look (eg, see this or this)

    (7) None of these problems were mentioned in the press release issued by the journal, Neurology, in which the paper appeared. It is reprehensible that an academic journal should encourage uncritical reporting of studies in the mainstream media.
    Research score: 2/5