Saturday, May 9, 2015

To Err on Which Side of Caution

Diet Heart Dogma and the Totality of the Evidence

When someone publishes a negative meta-analysis of the diet heart trials (e.g. Ramsden and Harcombe) the overwhelming response appeals to the ‘totality of the evidence’, which in this case also includes observational studies (only the favourable ones though) and how SFA and PUFA affect blood lipids.  Essentially this becomes little more than an appeal to authority as the respondent will often say that the AHA/etc has looked at the ‘totality of the evidence’ and recommends replacing SFA with PUFA

·         A new study is published
·         Do the results support your position?
o   If yes: claim the study was well designed and that it supports your position.  Whatever you do, don’t discuss the limitations of the study
o   If no: criticise the study and appeal to the ‘totality of the evidence’

This doesn’t just happen when ‘someone else’ publishes something negative, but also when pro-diet heart hypothesis researches publish a negative paper.  A good example of this comes from the meta-analysis by Skeaff & Miller:

“The evidence from metabolic ward studies clearly shows that diets low in SFA reduce total cholesterol and should therefore reduce the risk of CHD. However, the meta-analysis of results from cohort studies – albeit from a limited number of studies – showed no association between SFA intake and CHD, demonstrating their unreliability. The observational evidence for an association between dietary PUFA and CHD risk is inconsistent and is unreliable.” – Skeaff & Miller (2009)

In addition, the Minnesota Coronary Survey (MCS) was conducted during 1970 to 1973 [1].  The results were presented in 1975 at one of the American Heart Association's Annual Scientific Sessions and the abstracts were published in a supplement of the AHA’s journal ‘Circulation’ in 1975 (the first three studies in my blog post, which don’t exist on the internet, but I was able to access through a university library).  However, the full study was not published until 1989, 14 years later!  In GCBC, Gary Taubes mentions that: “When I asked Frantz in late 2003 why the study went unpublished for sixteen years, he said, “We were just disappointed in the way it came out””

This begs the question: why bother conducting clinical trials or doing a meta-analysis in the first place if you will simply ignore/rationalise unfavourable results.

Nutrition/dietetics is probably among the most dogmatic fields in science and within nutrition/dietetics, the diet heart hypothesis could easily be the worst

So in response to my results that replacing SFA with PUFA doesn’t reduce CHD or mortality, I’m fully expecting that: (1) some people will criticise the trials and suggest that we should instead rely on blood lipids and the (favourable) Jakobsen or Farvid meta-analysis (and not the Skeaff, Mente or Chowdhury meta-analyses, which found no benefit); and (2) some people will talk about moving away from reductionist recommendations while still recommending we reduce SFA, or use certain ‘eating patterns’ to justify themselves (blog)

In the Absence of Perfect Evidence

I agree that the trials aren’t perfect.  That’s why I excluded 4 of them for being ‘inadequately controlled’, described the other 5 as being ‘adequately controlled’ rather than ‘well controlled’ and removed SDHS just in case its negative result was due to a difference in TFA intake.  Most of the participants were middle-aged men with CHD (probably white also), with very few women, people above 70 and without pre-existing CHD.

Ideally a large and long, well designed, randomised controlled trial would be conducted in both men and women to provide a better understanding of whether a valid* and fair** replacement of SFA with PUFA would reduce CHD events, CHD mortality and total mortality (and also how this affects non-CHD events such as gallstones).  However, such a trial is not likely to happen in the near future and even if it began tomorrow there would be period of several years before the results would be in.

So in any event, we still have to decisions about what to eat today.  Should we err on the side of caution of replacing SFA with PUFA, or perhaps we should be cautious about reducing SFA and/or increasing PUFA and should we instead turn our attention elsewhere?

In the next few posts I’m going to argue that we shouldn’t rely on risk factors on observational studies for dietary recommendations; discuss the evidence on whether PUFAs are harmful; and mention reasons not to reduce saturated fat and to instead focus on the bigger picture

* ‘Valid’ = comparing items of food used for similar purposes.  For example: should I put add butter or margarine to my bread/potatoes/etc; and should I cook with butter/coconut oil or vegetable oils? 

** ‘Fair’ = replacing refined/processed sources of SFA (butter, coconut oil, palm oil) with refined/processed sources of PUFA (vegetable oils); or replacing whole food sources of SFA (full fat dairy, fatty meat, coconut) with whole food sources of PUFA (peanuts, nuts and seeds).  No difference in TFA intake, baked goods, fruits and vegetables, etc.  (Another example of a reasonably ‘fair’ comparison is the ‘filled’ meats and milk that were sometimes used in the trials, whereby the fat in meats and milk were taken out and replaced with vegetable oils.  Although it’s debatable as to whether this is a ‘valid’ comparison)

No comments:

Post a Comment