Lessons from the PREDIMED Retraction
In June 2018 the prestigious New England Journal of Medicine (NEJM) retracted a paper it originally published April 4, 2013 on the PREDIMED study. The study was conducted in Spain on about 7400 older men and women at fairly high risk of cardiovascular disease (CVD) but who had not yet suffered a major CVD event (e.g. heart attack or stroke). The subjects were supposedly divided randomly into 3 groups. The control group was advised to follow a “low-fat” diet and the other two groups received dietary counseling and group counseling on how to follow a Mediterranean-style diet. Subjects in one of the two experimental “Mediterranean Diet” groups were reportedly given a free liter of olive oil each week and the other “Mediterranean Diet” group was given free mixed nuts (50% walnuts and 25% almonds and filberts) equal to 210g per week (or a little over an ounce of free nuts per day). After an average of 4.8 years of follow-up, the study was stopped prematurely because the results showed significantly fewer composite CVD events in the two “Mediterranean Diet” groups compared to the control group. The PREDIMED study results had shown: “A primary end-point event [heart attack, stroke, or death from CVD] occurred in 288 participants. The multivariable-adjusted hazard ratios were 0.70 (95% confidence interval [CI], 0.54 to 0.92) and 0.72 (95% CI, 0.54 to 0.96) for the group assigned to a Mediterranean diet with extra-virgin olive oil (96 events) and the group assigned to a Mediterranean diet with nuts (83 events), respectively, versus the control group (109 events).” They found the two Mediterranean diet groups experienced about a 30% reduction in total CVD events compared to the “low-fat” group (which received no free food nor dietary counseling for the first 3 years). This rather impressive reduction in CVD events over a few years of follow-up led Dr. Ramón Estruch et. al. to halt the study prematurely and conclude: “Among persons at high cardiovascular risk, a Mediterranean diet supplemented with extra-virgin olive oil or nuts reduced the incidence of major cardiovascular events.” The reason the original paper was retracted was due to statistical anomalies, the failure to appropriately randomize subjects in some of the 11 study centers, and other concerns. There certainly is a growing problem of poor-quality research being published and it is not limited to only diet and nutrition studies. Dr. Milton Packer described the peer review — or lack of peer review — problems and has also presented a plausible way to expose poor quality and questionable studies after publication. Lesson #1 Given the now questionable data and analysis in the original PREDIMED study there is no question that the NEJM did the right thing by retracting the original 2013 PREDIMED study. However, the NEJM then published in their June 21, 2018 issue a slightly modified PREDIMED study which basically came to the same conclusion as the retracted study from 2013: “In conclusion, in this primary prevention study involving persons at high risk for cardiovascular events, those assigned to an energy-unrestricted Mediterranean diet, supplemented with extra-virgin olive oil or nuts, had a lower rate of major cardiovascular events than those assigned to a reduced-fat diet. Our findings support a beneficial effect of the Mediterranean diet for the primary prevention of cardiovascular disease.” That conclusion appears to be too sweeping and also warrants being retracted, in this reviewer’s opinion. Why? The data only showed the two “Mediterranean diet” groups who received far more dietary counseling and advice and were given either free extra virgin olive oil (EVOO) or free mixed nuts (NUTS) did not experience a statistically significant reduction in heart attacks or fatal heart attacks compared to those who were assigned to the “low-fat” diet. According to the original PREDIMED study and the revised PREDIMED study, the only significant reduction in CVD events was for strokes. Calling the control group’s diet “low-fat” (LF) or even “reduced fat” was questionable since PREDIMED data showed dietary fat in this group declined from 39% of calories to 37% of energy based on notoriously inaccurate food frequency questionnaires. Since when is a diet with 37% fat calories “low-fat”? Of course, it is not just the percent fat but the type of fat that impacts CVD risk and the diet protocol did not prioritize reducing saturated fat, trans fat, and cholesterol rather than essential fat. While those on the two Mediterranean diets were told to consume fatty fish, unsalted nuts & seeds, and sofrito (made from crushed tomatoes, onions, garlic, and green peppers sautéed in olive oil), the LF diet group was told to avoid these foods as well as other fatty foods including olive and other vegetable oils. The LF group was also advised to eat more breads, rice, and cereals (all likely high in salt and frequently refined). It is likely that the LF diet group consumed more salt, less unsaturated fat and especially less omega-3s than the two “Mediterranean Diet” groups. Might an LF “control” diet higher in salt and lower in omega-3s tend to increase the risk of stroke and help explain why the two “Mediterranean diet” groups (EVOO & NUTS) suffered fewer strokes than the “control” group? Lesson #2 Most of the subjects in the PREDIMED study were overweight and yet losing weight was not a goal for any of the 3 diet groups who were told to eat ad libitum. Not surprisingly, there was no significant mean weight lost in any of the groups. Therefore, it is certain that the extra calories consumed as free EVOO or free NUTS led to a reduction in calories from other foods. The small reduction in dietary fat in the LF group likewise must have led to increased calories from other foods with more breads and cereals being the most likely source as the participants were encouraged to consume more of these foods. The PREDIMED study and media reports about it led many people to believe more olive oil or nuts would “magically” lower CVD risk. However, the PREDIMED data only showed that the EVOO and NUTS groups only saw fewer strokes than the “LF Diet” group. For all we know, that benefit might not have been the result of more EVOO or NUTS but rather reduced consumption of salt-rich breads and cereals. Or it may have been the result of the polyphenols found in those nuts and olive oil but found in minimal amounts in refined grains and cereals and oils. Of course, if these phytochemicals help reduce CVD events and strokes, then there are many other dietary sources of these polyphenols than EVOO and NUTS. Indeed, polyphenols occur in much higher amounts per calorie in cocoa powder, tea, berries, and many of the spices found in an “Italian Blend” seasoning such as oregano, basil, sage, rosemary, and thyme than in EVOO or NUTS. Whole grains and many vegetables including those found in Spanish sofrito are also good sources of polyphenols and likely other phytochemicals that may protect us from CVD, cancer, and many other ills.Lesson #3 While bias and/or clinical mistakes often lead to misclassified CVD events, there is little doubt about a subject dying. Indeed, reducing total mortality seems even more important than reducing some composite CVD event rate. So how many people ended up dying in the EVOO-, NUTS-, and LF-diet groups? After about 5 years, the total number of deaths in these 3 diet groups was 118, 116, and 109, respectively. Given that these subjects entered the study at high risk of CVD mortality, how is it that total mortality was certainly no greater in the LF-diet group than the two Mediterranean diet groups? One cannot conclude from the PREDIMED study that adding extra virgin olive oil and/or more nuts to a diet high in salt, saturated fat, cholesterol, and refined carbohydrates will magically reduce one’s risk of dying from CVD and other causes. While it is likely that replacing calories from butter, cheese, and processed & red meats with olive oil and/or nuts would reduce CVD events and total mortality, it is even more likely that replacing those high saturated fat, cholesterol, and salt-rich foods in the diet with more beans, whole grains, whole fruits, and vegetables would cut the risk of CVD and total mortality by a much greater amount than observed in this PREDIMED study. In addition, the latter substitution would have been far more likely to lead to a reduction in ad libitum energy intake and the loss of excess body fat stores as well. Telling people to consume more nuts and olive oil (as the subjects in the EVOO and NUTS groups were) is unlikely to lead to reduced energy intake as they are among the most calorie-dense of all dietary options. What if the “LF Diet” group was given the same amount of energy as unsalted whole grains and beans, and whole fruits and vegetables and told to focus on reducing foods rich in saturated fat, trans fat, cholesterol, and salt, and cut way back on refined grains, sugars, and highly-processed foods rather than the questionable dietary advice and non-food “gifts” they received? However, both the now-retracted and the “rebooted” PREDIMED studies provide no useful information on these more important and widely-promoted dietary changes.By James J. Kenney, PhD, FACNReferences:
- https://www.nejm.org/doi/full/10.1056/NEJMoa1200303.
- https://www.foodpolitics.com/2018/06/whats-up-with-the-retracted-mediterranean-diet-study/
- https://www.vox.com/science-and-health/2018/6/20/17464906/mediterranean-diet-science-health-predimed
- http://blogs.plos.org/absolutely-maybe/2018/06/14/what-does-the-predimed-trial-retraction-re-boot-mean-for-the-mediterranean-diet/
- https://www.medpagetoday.com/blogs/revolutionandrevelation/73459
- https://www.medpagetoday.com/blogs/revolutionandrevelation/73587
- https://www.nejm.org/doi/full/10.1056/NEJMoa1800389
- https://www.medpagetoday.com/cardiology/arrhythmias/73605?xid=nl_mpt_%20SRNeurology_2018-06-23&eun=g430481d0r&utm_source=Sailthru&utm_medium=email&utm_campaign=CardioScope_062318&utm_term=Breaking%20News%20-%20Cardiology
