Skip to main content

Study Card C: The Mediterranean diet (PREDIMED)

For the explainer. Read this, then plan what you’ll say.

The study in one paragraph

In Spain, 7,447 adults aged 55 to 80 who were likely to develop heart disease (for example, they had diabetes, smoked, or had high blood pressure) but did not have it yet were assigned to one of three diets: a Mediterranean diet with extra-virgin olive oil (the study gave them olive oil), a Mediterranean diet with nuts (the study gave them nuts), or a control diet with advice to cut down on fat. Researchers counted heart attacks, strokes, and deaths from heart disease over a median of 4.8 years.

The results were first published in 2013. Then problems were found in how people were assigned to diets at some sites. The 2013 paper was retracted and a corrected analysis was published in 2018.

Source: Estruch, R., et al. (2018). Primary prevention of cardiovascular disease with a Mediterranean diet supplemented with extra-virgin olive oil or nuts. New England Journal of Medicine, 378, e34. doi.org/10.1056/NEJMoa1800389

Key facts

Key facts about Study C
Question Answer
Who was studied? Adults in Spain, aged 55 to 80, who were likely to develop heart disease
Explanatory variable Diet (three groups)
Response variable Heart attack, stroke, or death from heart disease (yes or no)
How were they found? Recruited through health centers
How many? 7,447
Design Randomized experiment (mostly; see below)
What went wrong? At some sites, people were not assigned one at a time. For example, people in the same household were put on the same diet, and at one site whole clinics were assigned the same diet. About 1,588 participants were affected.

The numbers to explain

As counts of people (over about 5 years):

Events per 1,000 people in each group
Group Out of every 1,000 people, about this many had a heart attack, stroke, or heart-disease death Sample proportion
Control diet 44 0.044
Mediterranean + olive oil 38 0.038
Mediterranean + nuts 34 0.034

Comparing two groups. The researchers compared how often these events happened in the olive-oil group with how often they happened in the control group. They report the comparison as a ratio:

rate of events in the olive-oil group ÷ rate of events in the control group = 0.69

95% confidence interval: 0.53 to 0.91

(Nuts compared with control: 0.72, 95% confidence interval 0.54 to 0.95.)

How to read it, using what you know:

  • A ratio of 1 means “no difference” between the groups. (For a slope, the “no relationship” value was 0. For a ratio, it is 1.)
  • 0.69 means the rate of events in the olive-oil group was about 31% lower than in the control group.
  • The whole 95% confidence interval is below 1. So the result is statistically significant: we reject the null hypothesis that the two diets have the same rate.
  • The ratio holds other variables constant and accounts for how long each person was followed, like the slopes in multiple regression. That is why it is not simply 38 ÷ 44. Use the counts to help your relative picture the result, and the ratio as the researchers’ best estimate.
  • Is it practically significant? About 6 fewer events per 1,000 people over five years. That’s for you and Grandma to discuss.

“Out of every 1,000 people on the regular low-fat advice, about 44 had a heart attack or stroke over five years. With the Mediterranean diet and olive oil, it was about 38. The researchers’ best estimate is about 30% lower, and the whole range of answers that fit the data is below ‘no difference.’”

But statistics can’t repair a broken design. The analysis of an experiment relies on random assignment to make the groups comparable. Where the assignment didn’t follow the plan, the researchers had to reconsider their original analysis. They re-analyzed the data with methods that don’t rely only on random assignment, and also checked the results without the 1,588 affected people. The results were similar.

In the paper, this ratio is called a hazard ratio. We have not used that term in class, and you don’t need it at the table.

Plan before you talk

Try to answer each question yourself, then open the box to check.

1. Your relative says “they keep changing their minds.” What’s a respectful, honest response?

Check your thinking

Agree first: nutrition advice really has changed a lot, and it’s reasonable to be tired of it. Then explain why it changes: many nutrition headlines come from single observational studies, which can’t separate diet from everything else people do. This one is an experiment, which is stronger. And this study is itself an example of science changing its mind in public: researchers found a problem, retracted the paper, and fixed it. That’s the system working, not failing.

2. Who can we generalize to?

Check your thinking

Older adults (55–80) in Spain who were likely to develop heart disease. It’s less clear for younger people, healthier people, or people in places where the “usual” diet is very different from Spain’s.

3. What can’t this study tell us?

Check your thinking

It compares a Mediterranean diet to advice to eat less fat, not to any diet. It can’t tell us which single food did the work. And the assignment problems mean the result is a little less certain than a clean experiment would be.

Jargon swaps

Jargon swaps for Study C
Instead of… Try…
random assignment “for most people, a coin flip decided which diet they got”
retracted “the journal took the paper back because of a mistake”
control group “the people who just got the usual advice”
statistically significant “would be unusual if the diet made no difference”
confidence interval “the range of answers that fit the data”