Meta-analysis of dietary interventions cannot be treated like drug trials
Comment on Risks and benefits of omega 3 fats for mortality,
cardiovascular disease, and cancer: systematic review Lee Hooper et al
BMJ, doi:10.1136/bmj.38755.366331.2F (published 24 March 2006)
All diagrams purporting to support the conclusion of Hooper et al (BMJ
24th March 2006) are based on relative risk - favours high or favours low
omega 3 - which like a drug trial assumes the variable is only the omega 3
dose. This falsifies the review..
Efficacy of a fatty acid depends on which fatty acid(s), position in
the glyceride, in the specific family or indeed the nature of a product
(e.g phosphoglycerides, triglyceride, enteric) , dietary prehistory of
individuals and importantly, on other fats in the diet which compete for
Moreover, it is DART II that skews the data. It was a stop start
affair and used MAXEPA capsules. MAXEPA is an EPA rich oil derived from
but not a fish oil!
The alternative was advice, which is notorious for non compliance.
To put it bluntly, advice to eat fish if taken could be translated in
different ways. Cardiff has good fish and chip shops or did they eat
farmed salmon in butter and mayonnaise sauce and mashed potatoes cooked
with margarine, to soak up the fats? Or did they eat raw salmon (Norwegian
Gravidlax perhaps) , lightly grilled tuna or swordfish with a mixed salad
and plain boiled potatoes or rice? Did they use olive oil or sunflower
oil? These menus would have quite different effects. Careful validation
with full reporting would be required to reach any degree of scientific
In an attempt at validation DART II reports plasma levels of EPA in
only 39 test compared to 29 control after 6 months fish advice = 4.58
(3.54) compared to no fish advice 3.03 (1.34) mg/dl. The standard
deviation of the test is larger than the mean of the alleged control. Are
these really different (no statistics are given)?.
Of critical importance, they do not report docosahexaenoic acid
(DHA). Is that because there was no difference or maybe less? DHA is the
principle cell membrane omega 3 fatty acid of the endothelium, smooth
muscle cells, heart, immune, neural and other cells. The receptors are in
the membrane and that is where you would want to see a difference if your
measure was effective. There is relatively little EPA in the receptor
sites. What is in the plasma can be very different to what is around the
membrane receptors. Moreover control levels of EPA seem far higher than
others report so maybe they are not measuring EPA at all. This is simply
not a credible study and even its authors had concerns.
There is an abundant and robust science showing that other fatty
acids in the diet (e.g saturated fats, linoleic, arachidonic acid) will
affect differentially the utilisation of omega 3 fatty acids e.g. alpha-
linolenic, stearodonic, eicosapentaenoic, docosapentaenoic or
docosahexaenoic acids. Little attempt has been made by Hooper et al to
control for these confounders. To illustrate the point, there is a recent
review of long standing data (1) which concludes "that eicosapentaenoic
acid and docosahexaenoic acid have differing haemodynamic and anti-
atherogenic properties. The effects of the two fatty acids may also differ
depending on the target population".
There is a lack of understanding in this review possibly because it
is not the lead author’s field. However, that is no excuse. A meta-
analysis which makes no attempt to control for these basic and varied
metabolic and dietary confounding factors is simply valueless and likely
to be misleading.
1. Mori TA, Woodman RJ. The independent effects of eicosapentaenoic
acid and docosahexaenoic acid on cardiovascular risk factors in humans.
Curr Opin Clin Nutr Metab Care. 2006 Mar;9(2):95-104
The health of the Nation
Competing interests: No competing interests