Food and diet

Does fish in Mediterranean diet combat memory loss?

The health benefits of a Mediterranean-style diet have hit the headlines, with The Daily Telegraph reporting we should, 'Eat oily fish to prevent memory loss,' while the Mail Online focuses on how oily fish can keep the memory 'sharp'.

These somewhat overenthusiastic headlines are based on a large study that looked at how eating habits similar to those seen in Mediterranean countries could affect your chances of developing cognitive impairment.

Researchers found that older adults from the US who followed a Mediterranean-style diet were 13% less likely to have impaired cognitive abilities. This was still the case even after adjustment for other health and lifestyle factors that could also have an influence. However, this association was not seen in people with diabetes.

The Mediterranean diet has been linked to a lower risk of developing several diseases, including cardiovascular conditions such as heart disease and stroke, and forms of dementia (such as Alzheimer's disease).

Drawing conclusions from research into associations between diet and health is difficult, as it is hard to measure the influence of dietary habits properly.

In the case of this study, the diet measurements may be quite specific to the US, so the findings may not apply to British people's diets in the same way.

Overall, however, this large well-conducted study suggests that sticking to a largely Mediterranean-style diet may have benefits for the cognitive functioning of people who do not have diabetes.

Where did the story come from?

The study was carried out by researchers from the University of Athens, the University of Alabama at Birmingham, and other research institutes in the US and the Czech Republic. It was funded by the US National Institutes of Health and was published in the peer-reviewed medical journal Neurology.

The media headlines regarding the importance of eating oily fish instead of red meat do not fully capture the research results. While oily fish is part of the Mediterranean diet, there are many other components that could also have beneficial effects. This study assessed all the components together rather than focusing on oily fish alone, as the media suggests – in fact, the word 'fish' does not appear once in the Neurology article.

Additionally, the 19% reduction in risk quoted by both The Daily Telegraph and the Daily Mail is incorrectly attributed to "people who adhere to a Mediterranean-style diet". This figure actually only applies to non-diabetic people. The risk reduction for the entire study sample was a more moderate 13% reduction in odds. However, both newspapers covered the main methods of the study well.

What kind of research was this?

This was a prospective cohort study that assessed the relationship between how much people adhered to a Mediterranean diet and their odds of developing cognitive impairment over time.

The Mediterranean diet involves eating lots of fruit, vegetables and olive oil, and eating few saturated fats, meat and dairy products. According to the study's authors, the diet has been associated with longer lifespan and a reduced risk of heart conditions, some cancers and Alzheimer's disease. It is the frequent subject of research into the influence of lifestyle on health and longevity.

As a prospective cohort study, this research can tell us how people's diet is linked to new cases of cognitive impairment over time. It also confirms that diet preceded any change in cognitive function, a key factor in assessing causality.

What did the research involve?

As part of the Reasons for Geographic and Racial Difference in Stroke (REGARDS) study, researchers recruited more than 30,000 individuals over the age of 45 from various regions of the United States. At the beginning of the study (baseline), individuals completed a version of the food frequency questionnaire (FFQ) that assessed their dietary habits. It was adapted specifically around foods commonly eaten in the US. Their cognitive functioning was also assessed using the six-item screener (SIS).

Participants were excluded if:

  • they had a history of stroke
  • there was missing data from the diet questionnaire
  • they completed fewer than two cognitive assessments during the study
  • the baseline tests revealed impaired cognitive status

The participants were asked to complete the FFQ several times over the first year to validate the results of the baseline diet assessment. The questionnaire was scored on a 10-point scale (0 to 9), with higher scores indicating higher adherence to a Mediterranean diet.

The SIS cognitive assessment was carried out at baseline and on a yearly basis to detect changes in cognitive status and new cases of cognitive impairment. For the statistical analyses, the researchers defined incident (new) cognitive impairment as a change from intact cognitive functions (an SIS score of 5 to 6) to impaired cognitive status during follow-up assessment (an SIS score of 4 or less).

Using data from the FFQ scores, the researchers separated participants into two categories according to their adherence to a typical Mediterranean diet. Scores of 0 to 4 were taken to indicate low adherence, while scores of 5 to 9 indicated high adherence.

The researchers then calculated the odds of new onset cognitive impairment in the group with high diet adherence, and compared this with the odds of new cognitive impairment in the low adherence group.

They adjusted the analyses to control for factors shown to be associated with new-onset cognitive impairment, including:

  • demographic factors, such as age, race and sex
  • socioeconomic factors, such as region of residence, household income and education
  • health status, such as history of heart disease, diabetes, atrial fibrillation, blood pressure, high cholesterol, use of blood pressure medications, symptoms of depression, and perceived general health
  • other risk factors, such as body mass index (BMI), waist circumference, smoking status, alcohol use and physical activity level

They also assessed how having diabetes influenced the relationship between diet and cognitive impairment. For this, they carried out two separate analyses similar to those described above: one for individuals with diabetes, and the other for diabetes-free participants only.

What were the basic results?

Main analysis

The primary analysis included 17,478 (58%) of the participants originally enrolled in the study, with an average age of 64.4 years (range 45 to 98). Approximately 31% of the participants were black, 43% were male, 17% had diabetes and 56% came from a part of the southeastern US known as the 'stroke belt', which has higher stroke mortality rates than the rest of the country. During an average follow-up period of four years, 1,248 (7%) of these participants were identified as having incident cognitive impairment.

Participants included in the analysis differed from those who were excluded in several important ways. They were more likely to:

  • be white
  • have graduated from university
  • to have an income above $75,000 (approximately £49,000)

The fully adjusted analysis included 14,701 participants. It found that individuals reporting high adherence to the Mediterranean diet had 13% lower odds of suffering incident cognitive impairment during the study, compared with participants with low diet adherence (odds ratio [OR] 0.87, 95% confidence interval [CI] 0.76 to 1.00). This finding was of 'borderline significance', meaning that it was not significant. However, it adds weight to the need to do further research on larger groups of people to better clarify whether the relationship is truly significant or not.

Analysis of the role of diabetes

The separate analysis looking at the role of diabetes included 14,758 participants. In this analysis, the researchers found that among people who didn't have diabetes, there was a 19% reduction in the odds of incident cognitive impairment in those with high Mediterranean diet adherence compared with those with low adherence (adjusted OR 0.81, 95% CI 0.70 to 0.94). This appears to be the risk reduction figure quoted by The Daily Telegraph and the Daily Mail.

This analysis also revealed that the association between a Mediterranean diet and incident cognitive impairment was not significant among people who had diabetes (adjusted OR 1.27, 95% CI 0.95 to 1.71).

How did the researchers interpret the results?

The researchers concluded that, "higher adherence to the MeD [Mediterranean diet] was associated with a lower likelihood of incident cognitive impairment", even when taking into account potential confounding variables.


This research suggests that older people who follow a Mediterranean-style diet are less likely to develop cognitive impairment than their peers, who tend to be less adherent to the Mediterranean diet. However, this relationship appears to only be the case for people who do not have diabetes.

This study has a number of strengths, including its:

  • prospective design, which allows us to be confident that dietary patterns existed prior to any change in cognitive impairment and were not subject to recall bias
  • large sample size, which allows us to be quite confident that the study was powered to detect an effect
  • validation of baseline dietary habit measures, which decreases the chances that the high versus low diet adherence scores used throughout the analysis were subject to misclassification bias due to a one-off (baseline) measurement
  • adjustment for numerous health and lifestyle factors that could be influencing the relationship between diet pattern and cognitive function

While this study seems to be good news for people who consistently eat a Mediterranean-style diet, there are several limitations (some noted by the study authors) that should be considered. These include:

  • A low participation rate (58%) and differences in key risk factors between those included and excluded from the analysis. It is not clear what the association between diet and cognitive status would look like if more of the enrolled individuals were included in the analysis (especially those with lower incomes and educational attainment).
  • While there are benefits to the large prospective cohort design, and the researchers attempted to adjust for many confounders, it is unable to account for all potential confounding factors (such as genetics). A randomised controlled trial would be needed for this.
  • Incident cognitive impairment was assessed using a measure less sensitive to certain types of cognitive impairment and to small changes in cognitive status, and is unable to differentiate between important clinical outcomes (i.e. between mild cognitive impairment and dementia).
  • There are concerns about how generalisable the food frequency questionnaire is. It was designed specifically to include food choices common in the United States. This means that the Mediterranean diet adherence measurements used in this study are (ironically) specific to the US, and may not be available or applicable to other countries with different dietary habits

Overall, this study adds to the evidence that a diet high in fruit, vegetables and olive oil, and low in saturated fat, meat and dairy, may have several health benefits. More large, population-based cohort studies investigating the impact of high Mediterranean diet adherence on diagnosed dementia would strengthen this study's findings, and address some of the limitations of the approach taken in this study.

NHS Attribution