Omega-3 Supplements: New data, deep concerns

Omega-3 sounds like the newest Transformer, or an expensive watch, though the term actually refers to a group of polyunsaturated fats that are essential for body and brain.

Omega-3 supplements are a multi-billion dollar global industry, but most people get enough of these fatty acids from their food. Why then would anyone would buy the supplements? If you already get enough omega-3s, why take more?

Partly, it's the data. Not good data necessarily, or data that scientists agree on, or data that's understood by salespeople in GNC stores. Rather, just enough data to distinguish fish oil from snake oil.

For instance, the FDA now allows manufacturers to claim that over-the-counter omega-3 supplements "may" lower the risk of high blood pressure and heart disease, so long as they add a disclaimer (e.g., a description of the evidence as "inconsistent and inconclusive". Most manufacturers rely on the familiar warning that the product isn't "intended" to treat, cure, or prevent any disease.)

There is evidence – most of it inconclusive at best – for other health benefits. Occasionally there's bad news, like a study published in BMJ Medicine this May linking regular use of fish oil supplements to higher risk of stroke and abnormal heart rhythms.

The omega-3 supplement industry thrives on a combination of suggestive data, slick marketing, consumer preferences for "natural" products, and a history of goodwill that Paul Greenberg, author of The Omega Principle, traces back to cod liver oil.

I'll save a discussion of the data for another newsletter, because I want to get right to my  topic: A new meta-analysis from University of Pennsylvania researchers suggesting that omega-3 supplements can reduce aggression.

My original plan was to share this paper with you, offer a critique, and comment on how wonderful it is that modern statistics, barely a century old, can help us lead healthier lives. That's what I do in most of my newsletters. However, digging into this paper uncovered such troubling concerns I took the unusual step of emailing the journal's editor-in-chief. The editor shares some of my concerns and we are currently discussing further action. So, parts of this newsletter will have an exposé-ish tone rather than just suggesting flaws. At the end I'll comment briefly on whether omega-3 supplements are worth taking anyway.

A new meta-analysis

The new meta-analysis, authored by Dr. Adrian Raine and Ms. Lia Brodrick, appears in the current issue of Aggression and Violent Behavior.

A meta-analysis is a review of studies that uses statistical methods to combine results and draw conclusions. In this case, Raine and Brodrick reviewed 29 studies on the relationship between omega-3 supplement consumption and aggressive behavior. These studies appeared between 1996 and 2024 (one was unpublished) and were carried out in 19 different labs. Across studies, data was available for a total of 3,198 people.

All 29 of the studies were randomized controlled trials (RCTs), meaning that participants had been randomly assigned to either take omega-3 supplements or a placebo, and each participant's aggression was measured both before and after taking the pills.

Raine and Brodrick excluded studies on feelings (e.g., anger), attitudes (e.g., hostility), mood, self-harm, etc. They emphasized that their focus was on "explicit measures" of aggressive behavior.

Main finding

Overall, omega-3 supplementation was associated with reduced aggression. Effects were small but significant.

(If you're a stats person, the Hedge's g values – i.e., Cohen's ds with a correction for small sample size – were .204 for the 29 studies, and .162 for the 35 samples derived from those studies.)

Because sample sizes and effects were small, we have to take these findings with a grain of salt (or perhaps a drop of fish oil), but still, it seems impressive that the supplements may have actually reduced aggressive behavior.

How could they do that? Raine and Brodick point out that aggression has a neurobiological basis, and that omega-3s have a critical role in brain structure and function. Thus, they conclude

"it is reasonable to believe that omega-3 supplementation could play a causal role in reducing aggression by upregulating brain mechanisms that may be dysfunctional...given the assumption that there is, in part, a neurobiological basis to aggression."

This is a fancy way of saying nothing. In fact, it's worse than saying nothing: It borders on something called the transitive fallacy.

In essence, Raine and Brodrick are saying that because omega-3s influences brain functioning, and brain functioning underlies aggression, omega-3s may influence aggression.

The fallacy here is obvious: Lots of things influence brain functioning. Broccoli, for instance. But it doesn't follow that eating more broccoli would make someone less aggressive. Whether we're talking about broccoli or omega-3s or anything else, links to aggression don't make sense unless we have some idea about the mechanism that's responsible. In plain English, "upregulating brain mechanisms that may be dysfunctional" just means: Making your brain work better in some unexplained way.

Ordinarily there's no reason to complain about speculative comments in a Discussion section. Raine and Brodrick showed that omega-3 supplements seem to reduce aggression; they weren't attempting to explain why. I only mention this because the lack of a plausible explanation tracks with serious concerns about what they claimed to find.

Red flags

I had a favorable impression of this paper when I first read it, although two quirky details stood out. In retrospect, they seem like red flags.

First, citations were not provided for the articles reviewed. Raine and Brodrick included a figure containing the last name of each lead author and year of publication (e.g., "Hamazaki 2002") but no other information about the sources was made available (except for a few that were included in their main reference list). Meta-analyses routinely include citations for the articles reviewed.

Second, in the Discussion section, the authors offer the following public health recommendation:

"based on these findings, our considered opinion is that there is now sufficient evidence to begin to implement omega-3 supplementation to reduce aggression in children and adults at a modest level - irrespective of whether the setting is the community, the clinic, or the criminal justice system."

That's breathtaking. "Community" appears to mean everyone. (Or perhaps everyone who exhibits aggression.) Raine and Brodrick's opinion seems to be that most if not all of us should be taking omega-3 supplements. Better living through chemicals...

When I first noticed these quirks, they didn't strike me as invalidating the findings. Rather, my preliminary take went something like this: The meta-analysis suggests that omega-3 supplements reduce aggression, but we can't be sure yet, because the effects were small, the samples were small, and the authors seem strikingly biased in favor of supplements.

Then I dug deeper.

Actually, all I wanted to do was to better understand what Raine and Brodrick meant by "aggressive behavior". (The lead author has not responded to an emailed query.) I decided to skim each of the 29 original studies and see how aggression was measured (and to get a sense of the quality of the research, as Raine and Brodrick didn't do much evaluative work there). What I found were three problems that go beyond the usual scholarly grumbling about methods and interpretation. This brings me to the exposé-ish part of the newsletter.

More red flags

1. Conflict of interest.

There's the letter of the law and the spirit of the law. I doubt that Raine and Brodrick literally broke any conflict of interest rules, but I found what seem like genuine conflicts of interest that might lead you to question their findings.

In the meta-analysis itself, Raine and Brodrick declare no competing interests. This may be true, but only in a narrow, legalistic sense. Consider:

a) Of the 29 studies reviewed, Raine was lead author on 6 of them. One of these studies is unpublished. (Of the 35 samples examined in the 29 studies, 11 samples came from Raines' own papers.)

b) In prior studies, including some included in the meta-analysis, Raine acknowledges funding by Smartfish, a manufacturer of omega-3 and other supplements based in Oslo, Norway.

c) More than two-thirds of the 29 studies reviewed were led by authors funded by manufacturers of omega-3 supplements. (In addition, one researcher, lead author on four papers and co-author on another one, was funded by the Japan Fisheries Association, which presumably has a keen interest in favorable omega-3s data.)

The fact that the supplement or fishing industry funded most of the research doesn't prove that the findings are untrustworthy. Earlier this week, I sought guidance from Dr. Chris Adkins, a professor of basic pharmaceutical science at High Point University. Because his comments were fair and illuminating – and will lead me to a more serious concern – I want to quote Dr. Adkins at length:

"It is fairly common for researchers to receive funding directly from drug or supplement manufacturers... However, in the pharmaceutical industry, experienced (big) drug companies typically go to great lengths to insulate themselves (and the resources they provide) from those who carry out the research related to the safety and efficacy of their products...

[C]linical trials should be pre-registered at clinicaltrials.gov before the study is carried out to discourage companies from strategic maneuvering and dodging the reporting of undesirable results. Most reputable journals now require any authors planning to publish clinical trial results to ensure their trial is registered, who funded the research, and, of course, whether any of the authors or contributors have any conflicts of interest.

 [Company funding] does not guarantee that there was unethical or improper influence or decision making related to the study and/or its publication...When there are conflicts of interests present AND some questionable decisions about the study or how it is reported…then that definitely should raise caution."

(The pre-registration that Dr. Adkins mentioned refers to the practice of documenting the design, methods, and analyses to be used in a study before data collection begins. ClinicalTrials.gov, the oldest and largest registry, has been operating for just over two decades.)

Few the 29 studies appear to have been pre-registered. (Tellingly, Dr. Nienke de Bles, lead author of one of the pre-registered studies, pointed out to me in an email that her study, and one other pre-registered study, led by David Gast, are among the few that failed to show reductions in aggressive behavior following omega-3 supplementation.) Some of the studies were conducted before pre-registration was available, or at least widely practiced. What we can say at least is that nowadays, most of the studies would not meet the most stringent criteria for clinical trials. Those that do tended to be the ones showing no benefits of omega-3s for reducing aggressive behavior.

Bottom line: The lead author of a meta-analysis, who's funded by a company that manufactures omega-3 supplements, reviews his own research as well as studies led by others who receive funding from omega-3 manufacturers (or a national fishery organization), and then draws a favorable conclusion about omega-3 supplements. This looks like conflict of interest but, as Dr. Adkins might say, it doesn't guarantee that anything went wrong. But let's revisit his last comment to me:

"When there are conflicts of interests present AND some questionable decisions about the study or how it is reported…then that definitely should raise caution."

There were indeed some "questionable" decisions about the study. The fact that citations were not provided for individual articles now takes on greater significance, as does the inclusion of one of Raine's own unpublished studies. It seems clearer now why Raine and Brodrick end up recommending that everyone take omega-3 supplements.

What I've written so far might be called circumstantial evidence for conflict of interest. I haven't definitively proven it or shown that the data were affected. Here's increasingly direct evidence:

2. Mistaken exclusions.

I can't claim to have conducted an exhaustive literature review, but I will say this: It only took about 10 minutes to find three studies that appear to meet all of Raine and Brodrick's inclusionary criteria (RCT design, omega-3 supplementation, measurement of aggressive behavior, data provided on effect sizes, publication in English) but weren't included. All three studies fail to link omega-3 supplements to changes in aggression.

For example, a 2010 RCT study, led by Ap Zaalberg, found no significant effects of omega-3s on aggressive behavior among young adults. (One could quibble that in this study, participants received other nutrients besides omega-3s, but the same is true of other studies that Raine and Brodrick did include. So, either all of these studies should be included, or all should be excluded. Raine and Brodrick included those that supported their central claim, while excluding at least one that contradicted it.)

To be fair, the exclusions may simply be oversights rather than intentional. Some of the studies that Raine and Brodrick review don't actually concern aggression. These studies includes dozens of outcome variables, and aggression just happens to be one of several subscales on just one measure. Perhaps it's easy to overlook studies like that.

Either way, exclusion of conflicting data is a serious problem – especially for  meta-analysis like Raine and Brodrick's, where the effects are tiny. Adding a few contradictory findings to the mix would surely make those tiny effects disappear.

3. Mistaken inclusions.

Recall that Raine and Brodrick insisted that their focus was on aggressive behavior. They claimed to exclude studies that measure omega-3 impacts on thoughts, feelings of anger, mood states, etc. However, several of the studies they reviewed do focus exclusively on these variables. For instance, a 2009 study led by Niki Antypa showed a slight association between omega-3 supplementation and scores on the aggression subscale of the Leiden Index of Depression Sensitivity-Revised. The entire LEIDS-R is publicly available, and the aggression questions clearly pertain to thoughts and mood rather than behavior (e.g., "In a sad mood, I am bothered more by aggressive thoughts").

Editorial response

I'm not one of those people who contacts editors about bad methodology. In this case, my concerns were ethical.

Here's the most negative way of framing what I discovered: A researcher who's funded by an omega-3 manufacturer reviews studies on the impact of omega-3 supplements. The researcher and his co-author are less than transparent about the sources reviewed, because, as it turns out, they have not chosen those sources fairly. Rather, they've selectively excluded and included sources so that the final set of 29, taken together, suggest that omega-3 supplements slightly reduce aggression. The authors then recommend that we all take these supplements.

I emailed a brief (and more neutral) summary of my concerns to Dr. Izabela Zych, a professor at the Universidad de Córdoba and Editor-in-Chief of Aggression and Violent Behavior. Dr. Zych responded quickly and cordially.

1. Regarding the lack of citations, Dr. Zych indicated that the journal's production team had made a mistake in not spotting their absence from the reference list.

This makes sense. Raine and Brodrick only provided author names and dates in a figure. The production team probably just checked whether studies cited in the text were included among the references.

Still, I don't think this absolves the researchers from a lack of transparency. Why wouldn't they include the full citations?

2. Regarding conflict of interest, Dr. Zych noted that "[a]fter analyzing this issue, in my view, competing interests should be reported per project/study... I would not expect past competing interests to influence all future projects."

In other words, Raine may have received funding from Smartfish for studies of his own that he included in this meta-analysis, but this doesn't represent a conflict of interest, because the meta-analysis wasn't funded by Smartfish.

I agree with Dr. Zych with respect to editorial practice. This is the letter of the law, and it's defensible.

As for the spirit of the law, consumers need to know that Raine and most of the research he reviewed was funded by supplement companies and/or a national fishery organization. (In fact, it may be that all of the studies have industry ties. I noticed some "fishy" connections that I didn't have time to track down.)

3. Regarding exclusion of articles, Dr. Zych agrees with me on this point, and as of today we are discussing how to proceed. (This is gracious of her, considering that she's not obligated to listen to anything I say.) We're still discussing the inclusion issue.

Conclusions

I see no clear evidence that omega-3 supplements reduce aggression.

One detail I haven't mentioned is some of the studies shouldn't have been included in the meta-analysis owing to purely interpretive dead-ends. When aggression declines after people take a bunch of nutritional supplements, including omega-3s, we can't be sure what influence, if any, the omega-3s had.

This is not a flaw of the original studies but rather the meta-analysis. In such cases, the original researchers didn't claim to be studying omega-3 effects. Since they recorded no outcomes specific to omega-3 consumption, their studies shouldn't have been included.

Take those studies out of the meta-analysis and the tiny effects would disappear. Add the excluded studies I mentioned and the effects would disappear. Exclude the mistakenly included studies I mentioned and the effects would disappear. Take the financial incentives away from the researchers, and, who knows, they might not find those effects to begin with.

Should we be taking omega-3 supplements anyway, for other reasons?

Based on the data, I wouldn't recommend them.

If you're taking omega-3s and you feel good, I wouldn't want to dissuade you. Appropriate levels (see here) are considered safe, even if we see occasional warning signs, like the recent study linking use to a higher risk of stroke and irregular heart rhythms.

If you're taking them because you think you don't get enough in your diet, please check with your health care provider. You're probably fine without them. Even vegetarians and vegans, who don't consume two important types of omega-3s (EPAs and DHAs) because they're only found in fish, still tend to get enough of both, because our bodies convert ALA, a type of omega-3 found in plants, to EPAs and DHAs.

If you're taking omega-3s because you think they'll benefit your health, please know that the evidence for health benefits is inconclusive at best. Check out this NIH summary, as well as Dr. Andrea Love's recent Immunologic post on supplements of all varieties.

If you're taking omega-3s because you think they'll make you less aggressive…maybe try some anger management techniques instead.

Thanks for reading!