“Following a strict Mediterranean diet offers substantial protection against heart disease, cancer, Parkinson's and Alzheimer's”, the Daily Express reported. It said the biggest ever study into the diet had shown that a diet rich in fruit, vegetables and fish can reduce the number of deaths from these diseases. It found that people who stick to the diet are 9% less likely to die young and show significant improvements in health.
The new study is a well-conducted systematic review, which statistically combined the results of 12 separate studies with a total of over 1.5 million subjects. This is probably the best type of evidence possible for assessing the long-term effects of this dietary pattern in healthy people. The results show that those who stick with this style of diet are more likely to live longer, and less likely to die from heart disease, stroke or cancer or to develop cancer, Parkinson’s disease or Alzheimer’s disease.
Where did the story come from?
Dr Francesco Sofi, a researcher in clinical nutrition, along with professorial colleagues from various organisations and departments associated with the University of Florence in Italy, carried out the research. The study was not externally funded and no competing interests were declared. The study was published in the peer-reviewed medical journal: The British Medical Journal.
What kind of scientific study was this?
This was a systematic review and meta-analysis. The authors aimed to find all the prospective cohort studies that have analysed the relationship between adherence to a Mediterranean diet, death, and rates of onset of selected long-term diseases in a primary prevention setting. That is, only studies conducted in healthy people, rather than those that tested how well the dietary pattern worked in people who were already known to have heart disease, for example.
The researchers used the databases PubMed, Embase, Web of Science, and the Cochrane Central Register of Controlled Trials to search for relevant literature up to 30 June 2008. Publications in all languages were eligible for inclusion, and the researchers also looked through the reference lists of the papers they found to identify further articles that might be relevant.
The initial search identified 62 articles, of which 12 were relevant. These 12 studies had a total of 1,574,299 subjects who had been followed up for an average of three to 18 years. The others were excluded after the abstract was read (20), or they were evaluated in more detail and were found to be case-control and cross-sectional studies (18) or duplicate studies. There were also exclusions where the diet or population evaluated was not relevant to the question.
The researchers used a standard form to extract the data from the original papers, and used standard statistical tests to pool the results and test for statistical significance. They also investigated if the studies were similar enough to each other to justify the statistical pooling of results (using tests of heterogeneity), and also checked for signs of publication bias (using tests to see if it appeared that studies reporting negative findings were missing from the data in a systematic way).
Six of the 12 studies were carried out in Mediterranean populations. The remaining studies were either conducted in US populations, northern Europeans, or a cohort of Europeans living in Australia. Although the total number of subjects from the 12 studies came to 1, 574, 299, none of the analyses looked at all subjects (because not all studies looked at all of the outcomes).
An adherence score was generated for each study. This estimated how much the population that was studied conformed to the traditional Mediterranean dietary pattern. A value of zero or one was assigned to each dietary component by using as cut offs the study participants’ average level of consumption. For example, if the people had a higher-than-average consumption of vegetables, fruits, legumes, cereals, fish, with a moderate intake of red wine during meals for their study they were assigned a value of one, whereas a value of zero was given to those whose consumption was below the average (median). By contrast, people who had a higher than average consumption of components not thought to be part of the typical Mediterranean diet (red and processed meats, dairy products) were assigned a value of zero, and the others had a value of one.
What were the results of the study?
Overall mortality (total deaths from any cause) was assessed by a meta-analysis of eight groups from nine studies totalling 514,816 subjects and including 33,576 deaths. It showed that each increase of two points in the score of adhering to the Mediterranean diet was significantly associated with a reduced risk of mortality (pooled relative risk 0.91, 95% confidence interval [CI] 0.89 to 0.94).
A greater adherence to the Mediterranean diet showed a benefit in reducing risk for various conditions. When the researchers looked at deaths from heart disease and stroke in three groups (from four studies) they found a reduction in relative risk of 9% (pooled relative risk 0.91, 95% CI 0.87 to 0.95). The five groups (from six studies) that were included in the pooled result for onset or death from cancer showed a 6% reduction in relative risk (pooled relative risk 0.94, 95% CI 0.92 to 0.96). Two groups (from three studies) looked at the onset of Parkinson’s disease and Alzheimer’s disease and showed a 13% reduction in relative risk of developing these conditions (pooled relative risk 0.87, 95% CI 0.80 to 0.96).
All these results approximate to a tenth, or 10% reduction in risk, which is statistically significant. The reduction ranged from 6% to 13%, and the 95% confidence intervals mean that the results are unlikely to have occurred by chance.
What interpretations did the researchers draw from these results?
The researchers conclude that greater adherence to a Mediterranean diet is associated with a significant improvement in health status. They say that the results seem to be “clinically relevant for public health, in particular for encouraging a Mediterranean-like dietary pattern for primary prevention of major chronic diseases.”
What does the NHS Knowledge Service make of this study?
This well-conducted meta-analysis provides strong evidence that a Mediterranean style diet can reduce the risk of major chronic diseases. There are statistical limitations to combining the results of observational (cohort) studies such as these with meta-analysis. However, the researchers have carefully assessed the differences between the studies and made attempts to eliminate or adjust for any bias that might have resulted from combining the studies. Other limitations mentioned by the researchers include:
- As the Mediterranean diet is not a uniform or standard pattern of eating, there is a variation in what the score means in each group. There are different ways of defining or grouping the legumes, nuts, and milk and dairy products for example.
- There is acknowledged controversy about the importance, and therefore categorisation, of different types of meats, and the definition of what is a moderate amount of alcohol intake. These are still matters of dispute among researchers and can differ among the selected studies.
- The studies that were included made different attempts to take into account any potential confounders (which can compromise the validity of any conclusions). This means that there may have been some confounding, especially for the non-Mediterranean cohorts, that was “left over” or residual after their analysis.
Overall, this study provides further evidence of the benefits of eating a Mediterranean-style diet and marks the start point for quantifying the extent of this benefit.
The researchers point out that it is important to estimate the effects of the dietary pattern as a whole rather than the individual components of the diet, as any analysis of single nutrients ignores the interactions between components and, more importantly, because people do not eat isolated nutrients.
Analysis by Bazian
Edited by NHS Website
Links to the science
BMJ 2008; 337:a1344
Cochrane Database Syst Rev 2007, Issue 4