Response on the JAMA publication: Association Between Omega-3 Fatty Acid Supplementation and Risk of Cardiovascular Disease Events: On September 12, 2012 Rizos et al. published a meta-analysis titled: “Association Between Omega-3 Fatty Acid Supplementation and Risk of Major Cardiovascular Disease Events” on fish oil and concluded no benefit. Spokespeople from the AHA have been on TV saying that omega-3 supplements clearly don’t ‘work’. here are some of my thoughts…
Positve view of the studyThey included all relevant studies; they did not exclude (like the previous meta-analysis (Kwak) did) the non-placebo controlled trials (GISSI-Prevenzione and JELIS)
First negative view about the study
They showed in Figure 3 that there was a significant benefit of omega-3 on cardiac death, and trends towards benefit in total mortality, sudden death and MI (plus trends towards increased stroke). But in the text, they said that there was NO significant effect on cardiac death – this is because they set the p-value for significance at 0.006, a much higher hurdle (than the usual 0.05) for concluding benefit. In my view, this is completely inappropriate and excessively conservative, especially for a very safe intervention. In other words, if you are testing a new drug that has potential benefits AND side/adverse effects, then you want to be very conservative in concluding “benefit” (i.e., you want to require a very small p-value) since – if you’re wrong and the drug really isn’t helpful (false positive) - your ‘endorsement’ of the drug will lead to increased use and thus the potential for increased adverse effects. However, for very low risk interventions (n-3 fatty acids), you don’t worry about adverse effects… you want people to use the treatment even if there is only a trend towards benefit. A favorable benefit-risk ratio. (I’d even suggest that in this setting, a p-value for ‘significant effect’ should be 0.1 instead of the traditional 0.05). In addition, nobody I know of ever adjusts for multiple testing (sets a lower p-value than 0.05 as the target for significance) in a meta-analysis. Therefore, I believe that the authors were far too conservative in this analysis, which led to their “no benefit” conclusion.
Second negative view about the sudy
They should have been much more nuanced in their conclusions. They said, “Our findings do not justify the use of omega-3 in structured [?] intervention in everyday clinical practice or guidelines supporting dietary omega-3 PUFA administration.” They should have said, “In patients of average age 63, with existing cardiovascular disease and under optimal medical care (which, by the way, is very UNcommon), the administration of about 1 g of EPA+DHA for 4 years will not affect major clinical outcomes.” Their study does NOT show that treating with a higher dose for a longer period of time, or treating patients earlier in the disease process or those who are not receiving “optimal medical therapy” will NOT be beneficial.
There may (or may not!) be a slight silver lining to all of this: Here is what Tom Barringer and I ended a chapter on n3 and CVD with in an upcoming book on “Omega-3 Deficiency”:
It should be stressed that future research will be significantly hampered if clinicians and patients are dogmatic in their belief that the value of n-3 fatty acids in CVD is already well-established. If such unfounded certainty is widespread, it will become very difficult to find patients (and investigators and IRBs) willing to participate in or approve the placebo-controlled clinical trials that are so desperately needed to properly evaluate the value of these nutrients in the treatment and prevention of CVD.”
Clearly with Rizos’ paper, we now won’t have any problem convincing the world that the question of omega-3s and CHD risk is still open.
In summary, they were too conservative in their analysis and they were not thoughtful in drawing their conclusions. It’s quite likely true that 1 g of EPA+DHA won’t affect outcomes over a few years in older people started later in life who are well-treated pharmacologically – but that’s a far cry from USA Today’s Headline “Fish oil pills with omega-3 don’t help against disease”
William Harris PhD
William Harris holds a PhD in Nutritional Biochemistry from the University of Minnesota, and did 4 years of post-doctoral research at the Oregon Health Sciences University. He was Director of the Lipid Research Laboratories at the University of Kansas Medical Center (KUMC) and at the Mid America Heart Institute, both in Kansas City, MO, for 22 years, and was on the faculty at KUMC and at the University of Missouri-Kansas City School of Medicine. Between 2006 and 2011 was the Director of the Cardiovascular Health Research Center at Sanford Research/USD (Sioux Falls, SD).
Read the complete article here.