Sunday, July 7, 2013

Why We Should Question Dietary Advice: Part 1

So diet and health are important, does that mean you should read your country’s dietary guidelines and blindly follow them?  I don’t think so.  I raised some objections to the draft for the new Australian Dietary Guidelines that I wrote a while ago (see here).  If I had to write it again I would add some things (like salt) and make a few changes.  The actual guidelines were released in April this year (see here).  Besides the submission, in this three part series I’ll give some reasons why we should question the standard dietary advice.

Recommendations from Observational Studies 

I haven’t read the USDA dietary guidelines, but I have read the Australian ones and I suspect they are very similar.  The Australian dietary guidelines mostly use observational studies to justify their recommendations, but there are some problems with this. 

The major weakness of many observational studies are confounding variables, particularly where humans are involved.  In diet related observational studies there are stacks of confounding variables.  To make things simple let divide the population into ‘health conscious’ and ‘not health conscious’.  The table below compares some of the differences. 

Health Conscious
Not Health Conscious
Diet
More fruits, vegetables, whole grains, legumes, nuts, fish and reduced fat dairy.  Low fat, SFA, TFA, salt and sugar
More junk food, meat, eggs, full fat dairy, refined grains.  High fat, SFA, TFA, salt and sugar
Lifestyle
More exercise, more and better sleep, better educated, higher income, more health check-ups
More smoking, heavy drinking, less education, lower income, fewer health check-ups

I could keep going on, but these are the most obvious differences.  The point is that if you do yet another ‘meat causes cancer’ observational study, the problem (besides the clichĂ© correlation doesn’t equal causation) is that there are many confounding variables.  How can you be sure that you are measuring the relationship between just meat and cancer.  Even if you adjust for all the other factors on the list, you can’t adjust for the factors you don’t measure.

There have been occasions where something that looks good in observational studies actually performs badly in RCTs, such as hormone replacement therapy (HRT) (in the WHI [1]) and grain fibre (in DART [2]).  This is because people were told that HRT and grain fibre were good for them, so then the health conscious people get HRT and eat grain fibre.  As a result they look favourable in observational studies even through they have been found to be harmful in RCTs, because they positive effect of all the confounding variables they didn't adjust for in the observational studies were greater than the negative effect of HRT, grain fibre, etc.  This is called the healthy user effect or healthy adherer effect and has been seen elsewhere, such as: vitamin supplementation, statins for other diseases and flu vaccines [3]

In my submission to the draft dietary guidelines last year I disagreed with the DAA when they said 

“Not a lot has changed in the new draft Guidelines, which shows we’ve generally been on the right track with the food and nutrition advice we’ve been giving Australians.” 

All this shows is that: 

  • You recommend X
  • Health conscious people eat X
  • Future observational studies show X is associated with better health
  • You therefore conclude X is good and you recommend X
 
There are other issues in observational studies [4]:
 
  • Food frequency questionnaires (FFQs) are used to assess food intake.  But FFQs are done very infrequently (generally less than once a year) and require people to accurately estimate their food consumption from years ago, even though they can't accurately report what they ate 24 hours ago
  • People underreport calories and 'bad' foods, while over-reporting 'good' foods and exercise.  This is particularly true with face-to-face interviews and people who are overweight do this even more (called 'self-report bias')

Researchers who conduct observational studies are also guilty of making recommendations based on observational studies.  A recent study found that in 56% of the observational studies examined, the authors made a clinical recommendation, whereas only 14% of the authors mentioned a need for RCTs [5] 

* Another weakness of observational studies, some more so than others, is confusing the most likely cause an effect.  For example, let’s say you do a study and find that a higher proportion of people with depression are taking antidepressants and then conclude that antidepressants cause depression.  This can be an issue in some medical related studies, but usually not much of an issue in diet studies, maybe except for stuff like calcium/dairy intake and osteoporosis 

** I like using observational studies where confounding variables are less likely to be a factor.  For example, ones that compare a genetic polymorphism with a disease to provide supporting evidence for a particular mechanism.  Such as the GG genotype for MPO, which results in a higher expression of MPO and has an HR of 5.5 for CV events (also note the high HR as opposed to the ~1.1 that you often see in ‘meat causes cancer’ studies).  See Immune Related Mechanisms (of Atherosclerosis) 

Ties to the Food Industry 

The organisations giving out dietary advice may receive corporate sponsorship from the food industry.  An example of this is the AND, which receives sponsorship from big businesses who produce junk/processed food, such as The Cocacola Company, Kelloggs, Mars Food, ConAgra and General Mills, as well as others such as the National Cattleman’s Beef Association and the National Dairy Council.  It wouldn’t be a big deal if the AND just said ‘thanks for money, goodbye', but as you probably suspect that’s not what’s happening.  What is happening is this: 

  • “Companies on AND’s list of approved continuing education providers include Coca-Cola, Kraft Foods, NestlĂ©, and PepsiCo.”
  • “Among the messages taught in Coca-Cola-sponsored continuing education courses are: sugar is not harmful to children; aspartame is completely safe, including for children over one year; and the Institute of Medicine is too restrictive in its school nutrition standards.” 
  • “At AND’s 2012 annual meeting, 18 organizations – less than five percent of all exhibitors – captured 25 percent of the total exhibitor space. Only two out of the 18 represented whole, non-processed foods.” 
  • “Roughly 23 percent of annual meeting speakers had industry ties, although most of these conflicts were not disclosed in the program session description.”
  • “To date, AND has not supported controversial nutrition policies that might upset corporate sponsors, such as limits on soft drink sizes, soda taxes, or GMO labels” 
  • Also something that is quite telling “That AND saw no conflict of interest in using industry money to study the potential bias of using industry money is troubling to say the least.”
  • “Historically, the Academy has never been much of a leader on nutrition policy and at times, quite the opposite. For example, in 1999, the Academy actually opposed mandated labeling of “trans fats” on food packaging” 

The AND and others will tell us that they need to work with the food industry to reduce obesity/T2D/CVD/etc, but of course that’s rubbish.  The food industry is the problem and they aren’t likely to ever be part of the solution*.  (One could argue that if the AND is in it for its members then: promote junk food as part of a ‘balanced diet’ >> more obesity >> more consults for dieticians) 

Now it would be wrong to use this information to completely ignore what the AND has to say, nor should this be a negative reflection on dieticians (the majority of whom find sponsorship from junk food corporations to be unacceptable), that would be ad hominem.  However, I think it’s appropriate to question the AND’s motivations and to look elsewhere for your primary source of nutrition advice 

See And Now a Word From Our Sponsors (It’s worthwhile to read the whole thing) 

It’s likely this problem exists, perhaps not on the same magnitude, with other organisations and in other countries.  Most Australian’s have seen the Heart Foundation’s tick of approval on junk food like most breakfast cereals and cereal bars, the University of Sydney’s nutrition department* has McDonalds and Kelloggs, among others, as corporate governors [6] and of course there’s the USDA handing out dietary advice. 

* In this article the author discusses ‘how junk food can end obesity’ with a model of obesity that is as reductionist as sugar, salt and fat.  So we blame junk food for being too high in sugar, salt or fat, then the manufactures change a few things, but how well has sugar alcohols/aspartame, MSG and olestra/guar gum/xanthin gum worked for us.  Or instead we blame junk food for being low in nutrients, then the manufactures add a few things, but free iron is great for bacterial overgrowth, folic acid promotes cancer growth and multivitamins may increase mortality. 

** You may remember the University of Sydney’s nutrition department for 'The Carnivore Connection Hypothesis' (see my take here and here) and 'The Australian Paradox' (that apparently Australian’s are eating less sugar now than they did a few decades ago.  See this website) 

Further Reading:
(1) Healthy User and Related Biases in Observational Studies of Preventive Interventions: A Primer for Physicians
(2) Modern Diet-Health Epidemiology: a Self-Fulfilling Prophecy? Part I
(3) Modern Diet-Health Epidemiology: a Self-Fulfilling Prophecy? Part II
(4) The Challenges and Failures of Nutrition Studies
(5) And Now a Word From Out Sponsors
(6) Heart Foundation says sugar isn't relevant

No comments:

Post a Comment