Three! journal articles were published on whole grains in the past couple of weeks; not sure how that happened. Zong et al had this article published in Circulation: “Whole Grain Intake and Mortality From All Causes, Cardiovascular Disease, and Cancer. A Meta-Analysis of Prospective Cohort Studies.” Aune et al had this article published in the BMJ: “Whole grain consumption and risk of cardiovascular disease, cancer, and all cause and cause specific mortality: systematic review and dose-response meta-analysis of prospective studies.” Chen et al had this one published in the AJCN “Whole-grain intake and total, cardiovascular, and cancer mortality: a systematic review and meta-analysis of prospective studies.”
The US Circulation/Harvard team study was the one that dominated the media headlines, not only in the US, but also in the UK, where we were misinformed: “Three slices of wholemeal bread a day slash risk of dying from heart disease by 25%”.
All three studies were meta-analyses of prospective cohort studies. Just to make sense of this – a meta-analysis is generally seen as the highest level of evidence possible. It is a statistical technique to pool together many similar studies, working on the principle that looking at several studies together is more powerful than looking at just one. Prospective cohort studies are also called population studies, or epidemiological studies. These types of studies follow populations over a period of time. At the start of the study they record as many things as possible about the participants’ lifestyle (smoking/alcohol/diet/exercise etc) and they record as many things as possible about the participants’ characteristics (age/gender/education etc) and then they see what happens to the people over the following years. The goal with prospective cohort studies is simply to spot patterns (associations) – do people who drink alcohol get liver disease? Do people who do yoga get fewer diagnoses of stress?
The standard issues
Every time a study hits the headline – e.g. “wearing red socks will reduce your risk of dying from boredom by 25%” – there are two standard issues: association is not causation and relative risk is not absolute risk…
1) Association is not causation.
Observational studies can only establish associations. They cannot say that A causes B. They can only say that A and B are associated. For example, observational studies are good ways of establishing that smoking and getting lung cancer are associated. The next question to ask is – does this have a plausible mechanism? The answer is yes – we have evidence of the ways in which substances in cigarettes damage lungs. We could then test the hypothesis “Smoking causes lung cancer” by conducting a randomised controlled trial where the intervention is smoking vs. not smoking and nothing else changes.
The two studies from last week have shown an association between consuming whole grains and mortality and that’s it. Is there a plausible mechanism? The BMJ article tries to suggest a few (are people who eat whole grains slimmer? Are whole grains anti-inflammatory? Read Dr William Davis’s Wheat Belly book and you’ll conclude the opposite.) There is nothing intrinsically healthy about whole grains, so there is no obvious plausible mechanism. I will suggest below an explanation for the observed association – the marker vs. maker argument.
2) Relative risk is not absolute risk.
I don’t blame the media for this one – I blame the press releases from the journals. These should know better than to put “25% reduced risk” in a press release – knowing that this is misleading relative risk hysteria and not scientific absolute risk information.
European heart data tell us that 33 in every 100,000 men died from coronary heart disease in 2009 and 8 in every 100,000 women died from CHD in 2009. IF eating whole grains were causal and IF eating whole grains could reduce this by 25%, then – taking the women – approximately 7 in 100,000 women in the top whole grain consumption category would be likely to die from CHD and 9 in 100,000 women in the never/hardly ever eat whole grains category would be likely to die from CHD (the difference between 7.1 and 8.9 being just over 25%, while maintaining 8 as the average/mean).
7 vs. 9 in 100,000. Hardly hold the front page now is it?!
The other key points
1) Dietary advice to eat whole grains is not evidence based.
For something to be evidence based, it needs to be based in evidence, If something is based in evidence, the evidence comes first. Advice to consume whole grains dates back to the 1980 Dietary Guidelines for Americans (if not earlier). If you can see the full BMJ article you will notice that only one study that appears in all the meta-analysis results even comes from the last century. This one study, Liu et al, dates back to 1999. All other studies used as evidence are from the year 2000 onwards (a high proportion are from the past couple of years).
Dietary Guidelines have been under serious attack from real food proponents for the past few years. I can understand wanting to try to find retrospective evidence for guidelines in this climate, but it will never make the guidelines evidence based. The evidence didn’t come first and it never will.
2) The comparator group is Jekyll & Hyde.
These studies claim to have found an association between whole grain consumption and reduced mortality (i.e. living longer). The groups that they compared were those in the highest intake of whole grain consumption (more than 3oz per day) vs. those in the lowest intake group. People in the lowest intake group were those who (self) reported “rarely or never” eating whole grains.
The 2010 Dietary Guidelines for Americans tell us: “Less than 5 percent of Americans consume the minimum recommended amount of whole grains, which for many is about 3 ounce-equivalents per day. On average, Americans eat less than 1 ounce-equivalent of whole grains per day.” (p.36)[Ref 1]
Hence – both studies have used a very small section of the population as the comparator group (<5%). There are two polarised groups of people in the “never/rare consumers of whole grains”: i) people who avoid all grains and ii) people who eat refined grains instead of whole grains. I would expect whole grain eaters to be healthier than refined grain eaters. The comparison that has not been done is the whole grain eaters vs. the no grains-at-all eaters (the latter, virtually guaranteed, also avoid sugar).
3) Whole grain consumption is a marker, not maker, of a healthy lifestyle.
I would expect people who consume whole grains regularly (the <5%) to: not smoke; not drink; be affluent; do yoga; be slim; shop at Whole Foods/Waitrose; eat at restaurants, not takeaways; have children called Olivia and Tarquin and so on. The whole grain consumption is a marker of good health, not the maker of good health.
The BMJ study noted this as one of the limitations of their research: “People with a high intake of whole grains might have different lifestyles, diets, or socioeconomic status than those with a low intake, thus confounding by other lifestyle factors is a potential source of bias.”
That’s journal speak for “Whole grain consumption is a marker, not maker, of a healthy lifestyle.”
The headlines imply that people just need to up their intake of whole grains and they will “slash their risk of dying from heart disease by 25%!” This could not be further from the truth. It’s not causal, the absolute difference is tiny and it’s a whole lifestyle being depicted in these studies – not a whole grain.
To prove me wrong, the authors of these studies need to give 3oz of whole grains daily to the smoking, drinking, obese, sedentary, aimless, fourth generation unemployed, living-on-benefits, deprived populations in the Welsh valleys and change nothing else. Do you think that will “slash their risk of dying from heart disease by 25%”?!
Ref 1: Department of Health and Human Services (HHS). Dietary Guidelines for Americans. In: Department of Health and Human Services (HHS), ed., 2010.