Food and mood: what the diet trials actually show

Changing what you eat helps low mood a little. In trials covering 45,826 people, those asked to change their diet scored slightly better on depression questionnaires than those who were not, and no better on anxiety.¹ In adults who were already depressed, a Mediterranean-style diet did more.²

What was actually tested

Three reviews sit behind almost every claim you will read on this subject, and they asked different questions.

The broadest one collected every trial that changed what people ate and then measured symptoms of depression or anxiety. It found sixteen, with results for 45,826 people.¹ One detail shapes everything that follows: in fifteen of those sixteen, the participants were not clinically depressed.¹ People with little low mood to lose have little room to improve.

A second review kept only Mediterranean-diet trials in adults who did have depression or depressive symptoms. It found five, covering 1,507 participants.²

The third is the largest. It set 21 trials beside 92 long-running surveys that had tracked more than 700,000 people, and it separated trials treating existing depression from trials trying to prevent it.³

All three tested whole eating patterns rather than single foods. That is worth holding onto when the question turns to which foods help.

How big the effect was

The numbers below share one unit, so it is worth saying what it is. Different trials use different symptom questionnaires, so results get converted into standard deviations: the ordinary spread of scores within a group. A shift of about a fifth of one is called small, half of one moderate, four-fifths large.

Across the broad set of trials, depression improved by 0.162.¹ Small. That is a real average shift across thousands of people, and not a change any one person would notice in themselves. Anxiety came out at 0.085, which is nothing.¹ The depression result did survive scrutiny: 0.171 among the better-run trials, and it held whether the comparison group got nothing, 0.114, or an active alternative, 0.224.¹

Among adults who were already depressed, the figure grows to 0.53 on the Mediterranean diet, which is moderate.² Two things sit next to it. The five trials disagreed with each other sharply, and the range a new trial would plausibly land in runs from -1.86 to 0.81, taking in no benefit and taking in harm.² The reviewers rated the certainty of that evidence low.²

The largest analysis tells the same story from another angle. In trials treating existing depression the effect reached 0.80, which is large, but the authors flagged that internal validity was low: the way the trials were built left room for the result to be an artifact.³ In participants who were not patients it shrank to 0.17.³ And in the prevention trials, where people who were not depressed changed their diets to avoid becoming so, the finding was a firm absence of any effect.³

Where the bigger claims come from

The claim that a diet cuts your risk of depression does not come from trials. It comes from surveys that record what people eat and then follow them for years.

One review of that literature found its strongest signal for the Mediterranean diet. Across four long-running studies, people eating closest to that pattern had about two-thirds the risk of developing depression compared with those eating furthest from it, a relative risk of 0.67.⁴ Anti-inflammatory eating patterns gave 0.76, and general healthy-eating scores 0.65.⁴

The newer and larger analysis found the same direction at a fraction of the size. Mediterranean adherence went with an odds ratio of 0.91 for self-reported symptoms, with no sign that sticking to it harder helped more.³ For most eating patterns and food groups, fish included, the data supported no effect at all.³

Two further findings undercut the headlines. The associations appeared when depression was self-reported and disappeared when it was diagnosed.³ And the authors rated the overall strength of this evidence very low, largely because of reverse causation: low mood changes what people eat, and a survey cannot tell which came first.³

Fiber, where the two kinds of evidence split

Fiber shows that split at its cleanest, because somebody went and tested it.

In surveys covering 181,405 people, eating more fiber went with slightly fewer symptoms. The shifts were 0.11 for depression and 0.25 for anxiety in the snapshot data, and 0.07 for depression in the studies that followed people over time.⁵

Then the trials. Ten of them, 740 participants, every one using a fiber supplement rather than food, and only one enrolling people with a diagnosis.⁵ Against placebo there was no difference for depression, -0.47 with a range running from -1.26 to 0.31, and none for anxiety, -0.30.⁵

Those ranges are wide, so the trials were too small to rule a benefit out. They also failed to find one. Taking the nutrient the surveys pointed at and putting it in a capsule did not reproduce the association.

Fish went the same way. A review of 95 earlier reviews found a hint of something for fish as food, where eating more went with a lower risk of depression, a relative risk of 0.78.⁶ For the fats taken as supplements, the direction looked favorable but the credibility was weak, there and across a long list of other conditions.⁶ The reviewers' own summary was that the associations, and the evidence behind them, were generally weak.⁶

Who the small effect applied to

Read together, the trials describe a narrow group. Young and middle-aged adults with depression, eating a Mediterranean-style diet, improved by a moderate amount on evidence rated low certainty.² People who were not depressed moved by 0.17, and gained no protection against becoming depressed.³ Trials with women in them improved more than trials without.¹ Anxiety did not respond.¹

None of these abstracts says that diet works instead of other treatment. Participants carried on with whatever care they were already receiving, and the results describe symptom scores shifting on top of that.

What remains unknown

The reviewers are candid about the gaps, and the gaps are wide.

The broad review says nobody yet knows which parts of a dietary change do the work, or how best to deliver one.¹ The Mediterranean review says large, long, well-run trials are still needed before anyone makes firm recommendations.² The largest analysis traces its own very low confidence to reverse causation and to weaknesses in how the studies were built and what they measured.³ It does flag one finding worth a proper test: fish and omega-3 intake in postpartum depression, an odds ratio of 0.84, where more did seem to mean better.³

On fiber, exactly one trial enrolled people with a clinical diagnosis.⁵ The group most likely to benefit is the group least studied. And whether any of this holds for diagnosed depression rather than self-reported symptoms is still open.³