In April 2006 the British Medical Journal published the results of a Cochrane review of the risks and benefits of omega 3 fats for mortality, cardiovascular disease, and cancer, by Lee Hooper and colleagues (BMJ. 2006;332:752-760).  In view of the fact that this review might have a bearing on ISSFAL's 2004 recommendation on polyunsaturated fatty acid intake, the Executive Committee of ISSFAL asked a panel of experts for their view on the Hooper  publication, and in particular on whether ISSFAL needed to review its recommendation in the light of the conclusions drawn by Hooper et al.

The panel comprised Clemens von Schacky, from the University of Munich,  in Germany; William S Harris, Sanford School of  Medicine, University of South Dakota , South Dakota , USA; Dariush Mozaffarian, Harvard Medical School ,Boston ,USA; and Penny M. Kris-Etherton,  The Pennslyvania State University, University Park, Pennsylvania,USA.

The response from the panel of experts is published below:

        A Cochrane meta-analysis has been published in the British Medical Journal, which came to the conclusion that “long chain and shorter chain omega 3 fats do not have a clear effect on total mortality, combined cardiovascular events, or cancer” 1. With respect to the former two endpoints, we disagree. 

       The null conclusion of the Cochrane report rests entirely upon inclusion of one trial, DART 2 2. This was a randomized dietary trial with clinical endpoints testing the effects on total mortality of either giving advice to eat fish or providing fish oil capsules to men with angina. Surprisingly, while total mortality was not statistically different in the two groups, there was less sudden death in the control group than in the intervention group. Upon exclusion of DART 2 from the meta-analysis, the overall decrease in relative risk with omega-3 consumption became similar to that reported in a previous meta-analysis by Bucher et al: 0.83 (95% confidence interval, 0.75 to 0.91) 3 .  

        According to a number of criteria, the results of DART 2 have the characteristics of an outlier, as reflected in the Hooper paper by a positive test for heterogeneity which disappeared after removing DART 2 from the analysis.  DART 2 was an open-label, dietary advice trial designed similarly to DART 1 4 , only in men being treated for stable angina.  Although reasonably well-designed, it was seriously under-funded and thus not properly conducted or reported. For example, only a rudimentary set of baseline parameters is presented for all participants, while the rest of the data refer to small subgroups at a subset of time points. Compliance by analysis of blood fatty acid levels was checked in only 2% of the cohort and only at 6 months. Neither long-term compliance with the advice nor how concomitant medications and health behaviours may have changed are known. The authors offered several possible explanations for their admittedly aberrant findings, but none were particularly satisfying to them or to us. Taken together, the results were generated with inadequate methodology, and therefore, trying to make sense of the trial’s results is doomed to fail. For Hooper et al. to allow this one, highly suspect study to sway their conclusions reflects, in our view, poor judgement.    

        Even as it stands, the Cochrane analysis suggests a 16 – 17 % reduction in total mortality by ingestion of omega-3 fatty acids, an effect size that is even larger than that of statins 5 . That the confidence intervals overlap 1.0 indicates that further studies were needed to improve precision of the estimate, not that the estimate is wrong. A number of large omega-3 and CHD studies (both epidemiological and interventional) are currently underway (e.g., 6 ), and these will provide a much clearer picture of the extent to which cardiovascular morbidity and mortality may be reduced by omega-3 fatty acids.  

        Other concerns about the Cochrane analysis include:  

  Omission of biomarker-based studies. Biomarkers, like serum, whole blood or red cell eicosapentaenoic acid and docosahexaenoic acid (e.g. the omega-3 index) or like adipose tissue fatty acids, better reflect dietary intakes of these FA than a diet questionnaire. It is well-known that fish vary widely in omega-3 fatty acid content across species, living conditions and a host of other factors. In the Cochrane analysis, however, biomarker-based epidemiological studies were (we believe inappropriately) excluded.

   Omission of relevant cohort studies. Only 3 cohort studies were included, while a number of other cohort studies were not because they “assessed only fish intake,” which the authors said, “is not clearly related to omega-3 intake”. Omega-3 intake is clearly (but not     perfectly) related to fish intake, so to exclude these studies seriously weakens the study’s conclusions.