Fish Oil Research Details 
(created 10/2011 from previous material) 

Summary:  Here are details from my  table of all the studies. Don't bother with this unless you're really, really interested. Instead, you'd be better off to look at the big review below which pooled all these studies for a single, striking result regarding dose of EPA relative to DHA. 

There's not much question that fish oil is better than a placebo. Until recently there was a big question about dose, but the big review below addresses that. 

Big Review

Bottom line: at least  60% of the omega-3's in each pill should be EPA. Less than that, it won't work.

In 2011, a research team at the New York State Psychiatric Institute did some simple, amazing math.Sublette   They just divided all the fish oil research studies into 2 groups. In the first group, EPA was less than 60% of the total EPA plus DHA in each pill. In the second group, the dose of EPA was 60% or more of that total.

They found that fish oil was not better than a placebo in any of the studies in the first group (in fact, in some studies, it was worse than a placebo). By contrast, fish oil was better than a placebo in all of the studies in the second group except one. 

You can see this result graphically below, even if you are not familiar with this type of math.  See how all but one "better than placebo" studies are in the black group? The two big diamonds with no bar through them (one each at the bottom of the black and white groups) represent the overall average effect

In this kind of graph, effect size means "how much better was the treatment than a placebo?"  You can see that the black group is much farther right, on average, than the white group. Far enough to make a difference in real people's lives? According to these data, yes. In general, a small "effect size" is 0.2 or so. A medium effect size is 0.5.  Getting up to 1.0 means the difference between the treatment and a placebo treatment is quite large.  

So, see the black diamond, the average: the effect size there is 0.53.  Another similar analysis reached a similar conclusionMartins , with an effect size of 0.46. But in the latter study, the biggest and best designed research trials were actually less likely to find such a difference between fish oil and placebo, so the story is not over yet.  

Details of the Frangou study from 2006

The most recent study in bipolar disorder is the 2006 report by Frangou and colleagues. This appeared in the British Journal of Psychiatry, a well-respected journal. The study design was good.  The sample is somewhat small, which may explain why they did not see an EPA-placebo difference unless they lumped both EPA groups together.  But when lumped, there appears to be a clear effect. The effect is not large, however. Note also that their patients were also taking mood stabilizers, so this is fish oil as an add-on, not by itself as in the original study by Stoll that started this whole thing. Nevertheless, it suggests that a manageable dose of EPA might be sufficient (as opposed to larger 4-gram dose used in the Stanley Bipolar Foundation study reported by Post and colleagues, which was not effective). Interestingly, in this 2006 study, the response appears to show up by the 4th week of treatment (they didn't look sooner, trying to minimize the placebo effect of their attentions).  

A patient whose long-term course is reported in Bipolar Network News demonstrated a stunning improvement which lasted over 1 year on omega-3's.  Look at her life-chart picture on page 7 ( v.8, Issue 2, scroll to page 7) for the dramatic change in course, but notice how it took about a year to really take hold (or so we might guess, from this report, although the author points out that there could have been other variables to account for this change also).  Notice also that the change was to stop depressive episodes, in this woman who had not had a manic phase in several years by that time on the medications which were continued as the omega-3's were added.  

Finally, notice that the dose was 6 grams.  In the accompanying article, which you can read via the link to that issue, some data are cited suggesting that a 6 gram dose of E-EPA may be too high and that a 1-4 gram dose may be preferable.  However, if you want to hear a real fan letter for omega-3's, from a fellow who's taking more than 4 grams a day, read this testimonial.  A local colleague takes just 2 grams of fish oil a day (500 mg of omega-3's total).  She is certain it helped dramatically with PMS and mild mood swings (family history of bipolar disorder).  She found 1 gram pills for $8.00/300 at Costco, so her cost is about $1.60 per month. 

Update 2007: a recent study indicates that people with severe Major Depression had less omega-three in a region of the brain associated with mood problems (frontal cortex).McNamara  Interestingly, only DHA, not EPA, was decreased in this region.   Not clear what this means. One study suggest the ratio between EPA and DHA matters -- that's next, below.