They said that while a diet with large amounts of fruit and vegetables, little red meat and a "glass or two" of red wine a day was a recipe for longer life, a seemingly healthy diet of low dairy and high amounts of fish and seafood did not prolong life.

This large European cohort study found that an increased adherence to a Mediterranean diet decreased risk of death from any cause by about 14%. However, the analysis was complex and it is not possible to say that any individual component of the diet decreases risk.

In addition, although "moderate" alcohol consumption was found to decrease risk compared to low or high amounts, the scoring system was broad. As such, this research needs to be interpreted with care and it is not possible to promote daily moderate alcohol consumption based on this research alone.

Professor Antonia Trichopoulou and colleagues from the University of Athens carried out this research. The study was funded by the Europe Against Cancer Program of the European Commission, the Greek Ministries of Health and Education and a grant to the Hellenic Health Foundation by the Stavros Niarchos Foundation.

This was a cohort study designed to investigate the relative importance of individual components of the Mediterranean diet and how increased adherence to this diet can affect overall mortality.

The researchers used members of the Greek segment of the European Prospective Investigation into Cancer and Nutrition (EPIC). This is a large study, taking place across 10 European countries, which investigates the nutrition and energy provided by different food items and how this relates to cancer and chronic disease. The study involved 23,349 healthy men and women (aged 20 to 86) who had no history of cancer, coronary heart disease or diabetes when they were recruited to EPIC (between 1994 and 1997). Their survival status was documented up to June 2008.

At enrolment, a validated food frequency questionnaire was used to assess the participants' diet over the previous year. This study focused on nine food groups: vegetables, legumes, fruits and nuts, dairy products, cereals, meat and meat products, fish and seafood, alcohol, and the ratio of monounsaturated to saturated fats. A food composition database was used to assess the nutritional content of the foods and standard portion sizes were used to estimate the quantities consumed.

The participants' adherence to the Mediterranean diet was assessed on a 10-unit scale (zero to nine).  For each of the nine food groups listed above, participants were given a score of either zero or one depending on their consumption of the particular foods (which were categorised as being either beneficial or not beneficial). A score of zero was given to people whose consumption of foods thought to be beneficial was below the median (average) and a score of one given to people whose consumption was equal to or above the median. A score of one was given to people whose consumption of a food not considered beneficial was below the median, and a score of zero if above the median. For alcohol, one to six units of alcohol a day for men and half a unit to three units a day for women was given a score of one (i.e., this was thought to be beneficial). Any other alcohol consumption was scored as zero.

Therefore, the total Mediterranean diet was scored from zero (minimal conformity to the traditional Mediterranean diet) to nine (maximal conformity).

In addition to diet, a lifestyle questionnaire assessed the participants' physical activity (each activity was assigned a metabolic equivalent of task, or MET, value), smoking status, BMI, specific diseases (cancer, diabetes and coronary artery disease) and educational level, all of which were taken into account in dietary analyses.

The average length of follow-up was 8.5 years, after which time the researchers assessed the participants' survival rates and looked at the effects of Mediterranean diet score and those of individual dietary components.

Of the 23,349 participants, 54% (12,694) had a Mediterranean diet score of zero to four and 10,655 participants had a score of five or more. There were 652 deaths in the zero to four scoring group and 423 in the group that scored five or more. Higher adherence to a Mediterranean diet reduced the risk of death from any cause by about 14% (adjusted mortality ratio per two-unit increase in score was 0.864, 95% confidence interval 0.802 to 0.932).

The researchers then subtracted each individual food group from this analysis to see what effect the individual food type had on the association between Mediterranean diet score and mortality risk. Using this, they calculated the "reduction in apparent effect" of the two-unit increase in diet score when this food item was excluded. This showed that moderate alcohol consumption contributed most to decreased mortality risk (reduced the effect of two-point increase by 23.5%), followed by low consumption of meat and meat products (16.6%), high vegetable consumption (16.2%), high fruit and nut consumption (11.2%), consuming a high ratio of monounsaturated to saturated fats (10.6%) and high legume consumption (9.7%).

However, when the researchers examined risk of death by consumption of any of the food groups individually, they found that moderate consumption of alcohol (compared to low or high consumption), above median consumption of vegetables, fruit and nuts and legumes, and high monounsaturated to saturated fat ratio decreased the risk of death (with only the effect of alcohol being statistically significant). Above median consumption of meat, dairy, fish and seafood increased the risk of death (although none of these effects were significant).

The researchers concluded that the Mediterranean diet lowers risk of death, and the principal components of the diet that cause this decreased risk are moderate alcohol consumption, low meat consumption and high consumption of vegetables, fruits and nuts, olive oil and legumes. Minimal effects were found for cereals, dairy and fish and seafood.

This large cohort study indicates that adherence to the Mediterranean diet lowered the risk of death from any cause. However, there are several points that need to be taken into account when considering which food groups contributed to the benefit:

• This study involved a complex analysis. The researchers found that a two-point increase in Mediterranean diet score decreased the risk of death by 14% and that removing different food groups from the analysis had varying effects on the size of this reduced risk. However, when each individual food group was assessed for the effect its consumption had on the risk of death, only alcohol was significant.

• Dietary questionnaires have various limitations due to their reliance on estimation of diet over the past year (which is unlikely to remain consistent over time) and variation in individuals’ estimation of amounts, portion sizes and energy content of food. The diet is also likely to have changed during the eight years before the outcome was assessed.

• The division into consumption above or below median amounts of "beneficial" or "non-beneficial" foods is very broad. Assigning and basing adherence to the Mediterranean diet on this is unlikely to ensure complete accuracy.

• Although certain possible confounders were adjusted for (taken into account), many medical factors that may have an influence on mortality risk were not assessed. Although people with diabetes, cancer and coronary heart disease were excluded before the study began, this was by self-report only. Additionally, these and other medical illnesses may have developed during follow-up.

• The outcome of "death from any cause" provides no information on quality of life and whether survivors are living in health or illness.

• Fish and seafood consumption was not found to be beneficial for mortality risk, but the typical Greek diet contains lower amounts of these foods compared to other food items. This could weaken the strength of any observations.

This study explores the contribution of different foods to the known benefit of following a Mediterranean diet. However, the nature of the statistical analysis and way in which the foods were scored means that it is not possible to say for certain how much of each component it would be optimal to consume, for example how much alcohol it is best to drink or how much red meat is bad.

The study was published in the peer-reviewed British Medical Journal, June 2009.