Name: Nutrition Mental Health
Creator: Realised
Published: 2026-06-01

Summary

What you eat measurably moves mood — but the honest version is narrower than the headlines: improving a genuinely poor diet and adding EPA-rich omega-3 are real adjuncts to depression treatment (not replacements), correcting an actual nutrient deficiency (B12, folate, iron, vitamin D, zinc) reliably lifts the mood and fatigue that the deficiency was causing, and swallowing those same nutrients when you are already replete does close to nothing for your mind.

Why Moderate

Dietary pattern as adjunct (A) — Tier 2 (Moderate):
• because: independent replication across populations (SMILES, HELFIMED, AMMEND) with a consistent dose-response signal and coherent mechanism — more than "one study + plausibility."
• NOT Tier 1 because: cannot be blinded, the flagship effect size is implausibly large (expectation-inflated), the flagship was underpowered with disclosure issues, and reverse-causation/confounding haunt the supporting observational base.
• NOT Tier 3 because: it has replicated intervention evidence, not just associations.

Omega-3 EPA adjunct (B) — Tier 2 (Moderate):
• because: it is the single best-evidenced nutrient adjunct in Firth's meta-review of meta-analyses (13 RCTs, no publication bias in that pool), with a specific, replicated EPA-vs-DHA and dose pattern.
• NOT Tier 1 because: Cochrane grades the broader literature low/very-low quality, some trials carry industry ties, and prevention in healthy people is null (VITAL-DEP).
• NOT Tier 3 because: the adjunctive-treatment signal survives publication-bias correction in the keystone synthesis.

Deficiency nutrients (C) — Tier 2 when correcting a measured deficiency (B12, iron, vitamin D, folate/methylfolate); Tier 3 for zinc and for supplementing the replete:
• because: deficiency-correction rests on Tier-1 biochemistry plus consistent symptom improvement on repletion (clearest for B12, iron-related fatigue/anxiety, methylfolate augmentation).
• NOT Tier 1 because: deficiency-restricted RCTs are scarce, mixed-population trials dilute the signal, and observational associations are partly confounded (folate/homocysteine attenuates after adjustment).
• Population-wide supplementation of replete people for mood is Tier 3–4 — null or near-null in the best-powered trials (VITAL-DEP), and largely an industry claim.

(The underlying biochemistry — one-carbon metabolism, monoamine cofactors, EPA in membrane/inflammation pathways — is Tier-1 physiology; the tiering above concerns the clinical/outcome claims, not the chemistry.)

Practical takeaway

This is recovery-register, not optimisation: you are removing dietary drag and correcting deficits, not "boosting" your brain.

First, and most important — the safety triage. If mood is moderate-to-severe, persistent, or there is any thought of self-harm, this is an adjunct alongside professional treatment, never a substitute. Diet and supplements are the floor you build under treatment, not a replacement for it.

Step 1 — Fix the pattern before reaching for pills (highest yield, lowest risk).
• Shift toward the dietary pattern that was actually tested: more vegetables, fruit, whole grains, legumes, nuts/seeds, olive oil, oily fish; less ultra-processed food, refined sugar, and sugary/fatty takeaway. This is the SMILES/HELFIMED/AMMEND intervention in plain terms (see diet_foundations_for_baseline).
• What "working" looks like: the dose-response thread from the trials means the people who changed their diet most improved most — so track adherence honestly, and expect mood/energy shifts over weeks (≈4–12), not days.
• Self-compassion clause: if your diet is poor because you are depressed (low energy, no motivation to cook), that is the reverse-causation reality — start with the smallest sustainable change, not a perfectionist overhaul.

Step 2 — Test before you supplement the "deficiency four." The single most useful action for a tired/low person is bloodwork, not a supplement haul. Ask a clinician about: ferritin (iron stores), vitamin B12, vitamin D (25-OH-D), and folate — and thyroid while you're at it (thyroid_dysfunction), because hypothyroidism mimics depression. Then correct what is actually low. (See micronutrient_deficiency_screening, fatigue_cross_pillar_diagnostic.)
• Highest-yield groups to test: menstruating/pregnant women (iron), vegans + older adults + metformin users (B12), high-latitude/winter/dark-skinned/indoor populations (vitamin D).

Step 3 — EPA omega-3, if depressed and on or seeking treatment.
• Form and dose that matches the evidence: an EPA-predominant product (≥60% EPA), targeting ~1–2 g of EPA per day (read the label for the EPA number, not the total fish-oil number — most products quote a big "fish oil" figure with little EPA). More than 2 g/day EPA gives no extra benefit.
• Best evidence is as an add-on to an antidepressant, not solo.
• Response window: weeks. Track mood alongside, and treat it as augmentation, not magic.

Step 4 — Methylfolate (clinician-guided augmentation). L-methylfolate (the trials used up to 15 mg/day as SSRI augmentation) is a reasonable clinician-directed adjunct, especially with known impaired folate metabolism. Plain folic acid is not the evidence-based form here.

What to track: a simple weekly mood/energy rating, dietary adherence, and — if you corrected a deficiency — a recheck of the blood level once repletion should have occurred (e.g. ferritin after ~3 months of iron), so you are treating a number, not a hunch.

What "working" does NOT look like: a same-week mood transformation from a pill; "fixing" moderate-severe depression with diet alone; or needing to stay on a fistful of supplements indefinitely when you were never deficient.

Evidence detail

Why This Entry Exists

A Realised user is low, flat, tired, or anxious and has read — correctly — that "food affects mood." They arrive with one of two over-corrections, and this entry has to head off both.

Over-correction one: "diet is fluff, just take the meds / hit the gym." This dismisses a genuine, replicated finding. People with depression eat measurably worse diets than people without; correcting a poor diet as an adjunct has now beaten an active control in more than one RCT; and a sub-group of low-mood, low-energy people are quietly running on a correctable nutrient deficit (low iron, low B12, low vitamin D) that no amount of therapy or SSRI will fix because the problem isn't psychological. Missing that is a real clinical failure — particularly for menstruating women (iron), vegans/older adults (B12), and anyone living at high latitude through winter (vitamin D).

Over-correction two: "I read that diet cures depression — I'll fix this with food and supplements instead of treatment." This is the dangerous one. The flagship trial (SMILES) produced an effect size so large (Cohen's d ≈ 1.2) that it is four to five times the effect of standard antidepressants — a number that should make any honest reader suspicious, not excited, because it almost certainly reflects an unblindable intervention plus expectation, not a food effect that size. Nutritional psychiatry is a real and useful adjunct. It is not a monotherapy, it is not a substitute for treatment in moderate-to-severe or suicidal depression, and the supplement aisle is mostly selling deficiency-correction to people who aren't deficient.

What bad advice does this protect against? The supplement-industry pitch that everyone should be on a fistful of "mood" pills. The wellness pitch that a clean diet replaces psychiatric care. The opposite, equally wrong, medical reflex that diet and micronutrients are irrelevant to mental health and never worth checking. And the specific trap of a depressed person blaming themselves for a poor diet that is, in significant part, a symptom of the depression — the reverse-causation problem that sits underneath this entire field.

Evidence

Read this as three evidence bases of different strength, kept apart on purpose. Conflating them is the field's signature error.
A. Dietary pattern as adjunctive treatment of depression — Tier 2 (Moderate), with a big honest asterisk

The flagship — SMILES (independent/government + foundation funded):
Jacka FN, O'Neil A, Opie R, et al. (2017). "A randomised controlled trial of dietary improvement for adults with major depression (the 'SMILES' trial)." BMC Medicine 15:23. Funded by Australian NHMRC and several foundations; not industry. N = 67 randomised (56 completed) adults with moderate-to-severe depression, most also on antidepressants and/or psychotherapy. 12 weeks. Arms: dietitian-led ModiMed (Mediterranean-style) dietary support vs. a befriending/social-support control matched for visit schedule.
• 32% of the diet group reached remission (MADRS) vs. 8% of the social-support control.
• The reported between-group effect size was Cohen's d ≈ 1.16 — and this is the problem, not the triumph. That is four-to-five times the effect of antidepressants over placebo. A real, replicable food effect almost certainly is not that big; an effect that large in an unblindable behavioural trial is the fingerprint of expectation and demand characteristics riding on top of a smaller true effect.
• Degree of dietary change correlated with degree of symptom improvement (a dose-response signal that argues something real is in there).

The documented weaknesses of SMILES (the Counter-Check, applied to the field's own flagship):
• Underpowered. N=56 completers is roughly one-third of the sample the authors' own power calculation said was needed.
• Unblindable. You cannot blind a person to whether they are being coached to eat better. The trial used blinded outcome raters, but the participants knew, and the published data themselves contain signs blinding was imperfect — the textbook setup for expectation bias (Molendijk et al., 2018, correspondence in BMC Medicine).
• Undisclosed recruitment. The original paper did not state that a website was used for recruitment in the final year; critics (Molendijk 2018) argued undisclosed recruitment practices could partly explain the remarkably large effect. (Funding-bias note: the critique here is NOT industry-driven; it is methodological hygiene from academic peers.)

Replication — and this is what rescues it from dismissal:
• HELFIMED (Parletta et al., 2019, Nutritional Neuroscience; independent): a larger (n≈152), group-based Mediterranean-diet-plus-fish-oil intervention produced similar improvements, and again the degree of dietary change tracked the degree of mood improvement.
• AMMEND (Bayes J, et al., 2022, American Journal of Clinical Nutrition; independent/university): a 12-week RCT of a Mediterranean diet in young men (18–25) with moderate-to-severe depression found a significant reduction in depressive symptoms vs. control. This matters because it is a different population (young men, a group usually under-recruited) reaching a concordant result.

So the pattern claim does not rest on one flashy underpowered trial; it has independent replication across populations, with a consistent dose-response thread. What it does not have is a clean, blinded, adequately-powered trial — which is why it lands at honest Tier 2, not Tier 1, and why the magnitude (the d≈1.2) should be treated as inflated even if the direction is real.
B. Omega-3 (EPA) for depression — Tier 2 (Moderate) as an adjunct; the EPA-vs-DHA split is the load-bearing detail

The keystone — a meta-review of meta-analyses (independent/government funded):
Firth J, Teasdale SB, Allott K, et al. (2019). "The efficacy and safety of nutrient supplements in the treatment of mental disorders: a meta-review of meta-analyses of randomized controlled trials." World Psychiatry 18(3):308–324. 33 meta-analyses, 10,951 people. Funded by a UK NIHR Clinical Lectureship / Dept of Health — not industry.
• **Strongest signal of any nutrient supplement: omega-3 (specifically EPA) as an adjunct to antidepressants in major depression.** Across 13 RCTs in ~1,233 people with MDD (mean ~1,422 mg/day EPA), omega-3 reduced depressive symptoms beyond antidepressants alone, with no evidence of publication bias in that pooled set.
• The same review's blunt bottom line: no nutrient supplement has evidence to support use as a stand-alone, first-line, or monotherapy treatment for any psychiatric disorder. Adjunct only.
• High-dose methylfolate (see C) was the other adjunct with positive signal; N-acetylcysteine had emerging signal (see nac_n_acetyl_cysteine).

The EPA-vs-DHA split (this is where most supplement-buyers go wrong):
• A consistent finding across meta-analyses (Sublette 2011; Mocking 2016; Liao 2019): EPA, not DHA, carries the antidepressant signal. Pure-EPA or EPA-predominant (≥60% EPA) formulations show benefit; pure-DHA or DHA-predominant formulations do not. Most cheap fish-oil capsules are DHA-heavy or balanced — i.e., the wrong ratio.
• Dose window: benefit clusters at roughly 1–2 g/day of EPA; ≥2 g/day does not look more effective (no extra benefit from megadosing). Effect is larger when added to an antidepressant than used alone.

The honest skeptical counterweight (don't skip this):
• The Cochrane review (Appleton et al., 2021, omega-3 for depression; independent) graded the overall body of evidence low-to-very-low quality under GRADE — predominantly small trials at high risk of bias, and the apparent benefit was "largely driven by the most imprecise studies." Cochrane's read of the same literature is markedly more cautious than Firth's.
• Some individual trials carry industry ties (non-financial support from fish-oil makers e.g. Nordic Naturals, Pronova/BASF, Mars Edge in scattered studies) — not the dominant funding pattern, but a reason to weight the EPA-specific, publication-bias-corrected pools (Firth's) over enthusiastic single trials.
• Large prevention trials in non-depressed people are null: VITAL-DEP (Okereke et al., 2021, JAMA; NIH-funded) found marine omega-3 did not prevent depression or improve mood scores in ~18,000 older adults without depression. Treating existing depression ≠ preventing it in healthy people.

Net: EPA is the single best-evidenced adjunct nutrient for diagnosed depression, the EPA/dose detail is real and actionable, and a sober reader holds Firth (positive, bias-corrected, adjunctive) and Cochrane (cautious, low-quality literature) at the same time. Hence honest Tier 2 — not Tier 1.
C. The "fix-the-deficiency" nutrients — B12/folate, iron, vitamin D, zinc — Tier 2–3, and the tier depends entirely on whether the person is actually deficient

The unifying truth: these correct a deficit; they are not mood drugs. Strong where there is a measured deficiency to correct; weak-to-null when sprinkled on the already-replete.

Folate / methylfolate:
• High-dose L-methylfolate (the active form) shows positive adjunctive signal in MDD (Firth 2019; Papakostas 2012 RCTs at 15 mg/day as SSRI augmentation). Plain folic acid was ineffective in the meta-review — the form matters, especially in people with MTHFR-pattern impaired folate metabolism.
• Caveat: the folate–depression association in observational data is partly explained away by physical comorbidity/disability (Rotterdam Study, Tiemeier 2002) — i.e., some of it is confounding, not folate.

Vitamin B12:
• B12 deficiency genuinely causes low mood, fatigue, and cognitive fog; it is independently associated with depression even after adjusting for comorbidity (Rotterdam Study) — the cleanest "this one may be causal" signal among the B-vitamins. Screen and correct in at-risk groups (vegans, older adults, metformin users, malabsorption). Supplementing replete people for mood has no good evidence.

Vitamin D — the textbook deficiency-vs-population distinction:
• Meta-analyses show vitamin D supplementation reduces depressive symptom scores, with the largest effect in people who are depressed and/or deficient and dose-response up to high intakes.
• BUT the large, well-powered **VITAL-DEP trial (Okereke 2020, JAMA; NIH-funded, ~18,000 older adults, 2000 IU/day) found no protection against depression** in a general, largely-replete older population. Older-adult-specific meta-analyses are also frequently null.
• Reconciliation: vitamin D plausibly helps correct a deficiency-linked low mood; it does not work as a population-wide mood prophylactic. Test, then treat the low number — don't supplement blind for mood. (See vitamin_d3_high_dose_supplementation, light_exposure_vitamin_d.)

Iron:
• Iron is a cofactor for serotonin/dopamine/noradrenaline synthesis; deficiency (even without anaemia — low ferritin) is a well-documented, under-checked cause of fatigue, low mood, anxiety, and poor concentration, disproportionately in menstruating and pregnant women.
• RCT picture is split by symptom: supplementation in iron-deficient (non-anaemic) people reliably improves fatigue, anxiety, and cognitive measures; effects on depression scores specifically are less consistent, though pre-post studies show improvement. The honest claim: check ferritin in tired/low women before assuming the problem is purely psychological — but do not iron-load people who aren't deficient (iron overload is harmful; see Risks).

Zinc:
• Depressed people have lower serum zinc on average (Swardfager 2013 meta-analysis: ~ −1.85 µmol/L vs. controls), and lower zinc tracks treatment-resistance. Adjunctive zinc has a modest positive signal in some RCTs. This is the thinnest of the four — real associational signal, limited and small-trial intervention data. Tier 3.
D. The mechanism-adjacent supporting layer (why "diet → mood" is biologically coherent at all)

Gut-microbiome/fibre, blood-sugar stability, and inflammation pathways give the dietary-pattern finding a plausible mechanism (see Mechanism, and diet_gut_microbiome, dietary_fiber_diversity_and_microbiome_health, blood_sugar_regulation). Plausible mechanism + replicated adjunct trials is exactly the profile that earns Tier 2, not the over-reach to Tier 1.

Mechanism

Four reinforcing pathways make "food moves mood" biologically coherent — which is what keeps this out of the fluff bin even where the trials are thin.

1. Substrate for neurotransmitters and membranes. Monoamine synthesis (serotonin, dopamine, noradrenaline) is enzymatic chemistry that requires cofactors — iron, B6, folate, B12, zinc. EPA/DHA are structural components of neuronal membranes and modulate the resolution of inflammation. A genuine deficit in any of these is a genuine bottleneck in brain chemistry; this is Tier-1 biochemistry and explains cleanly why correcting a deficiency works while topping up the replete does not — you cannot relieve a bottleneck that isn't there.

2. One-carbon / methylation (the folate-B12-homocysteine axis). Folate and B12 drive methylation reactions that produce monoamine neurotransmitters; impaired methylation raises homocysteine, which is repeatedly associated with depression. The active form (methylfolate) bypasses the MTHFR conversion step, which is why methylfolate works in the trials where plain folic acid does not.

3. Inflammation and the gut. Depression carries a low-grade inflammatory signature in a meaningful sub-group. A whole-food, high-fibre, lower-ultra-processed dietary pattern feeds a more diverse microbiome, improves gut-barrier integrity, lowers inflammatory tone, and supports short-chain-fatty-acid and neurotransmitter-precursor production via the gut-brain axis. EPA's anti-inflammatory (pro-resolving lipid mediator) action plugs into the same pathway — a plausible reason EPA, not DHA, carries the mood signal.

4. Glycaemic stability. Blood-sugar swings from ultra-processed, high-refined-carbohydrate eating produce reactive lows that read subjectively as irritability, anxiety, fatigue, and crashes — symptoms easily mistaken for, or compounding, a mood disorder. A steadier dietary pattern flattens that (see blood_sugar_regulation).

The honest mechanistic limit: every pathway above is necessary but not sufficient to prove a food intervention works at the magnitude headlines claim. The biochemistry guarantees that genuine deficiencies matter; it does not guarantee that a moderately-better diet produces a d=1.2 antidepressant effect. The mechanism justifies the direction; the trials (with their flaws) set the honest size.

Risks And Contraindications

• The headline risk is delaying real treatment. Using diet/supplements instead of professional care for moderate-to-severe or suicidal depression is the dangerous failure mode this entry exists to prevent. Adjunct, not replacement. Crisis or self-harm thoughts → urgent professional help, not a meal plan.
• Iron is the one to be careful with. Do not supplement iron without confirmed deficiency. Iron overload (and conditions like haemochromatosis) is harmful; excess iron is pro-oxidant and can worsen, not help, well-being. Test ferritin first; treat to target; recheck. Iron also causes GI upset/constipation and can be dangerous in overdose (especially a poisoning risk for children in the house).
• Omega-3: generally well-tolerated; high doses (esp. >3 g/day total) have a mild antiplatelet/bleeding-tendency effect — caution with anticoagulants/antiplatelets and before surgery. Fishy reflux is common; check sourcing for oxidation/heavy-metal quality.
• Vitamin D: safe in normal repletion doses; very high chronic dosing risks hypercalcaemia. Supplement to correct a low level, not blindly (see vitamin_d3_high_dose_supplementation for the dosing/cofactor detail).
• B12/folate: very safe (water-soluble). One real caveat: high folate/folic acid can mask the haematological signs of B12 deficiency while nerve damage progresses — which is exactly why B12 should be checked/corrected alongside folate, not folate alone.
• Zinc: chronic high-dose zinc induces copper deficiency (which itself causes neurological problems). Modest, time-limited, and ideally guided — not an open-ended high dose.
• Supplement quality/interactions generally: the supplement market is poorly regulated; products vary in actual content and purity, and "natural" is not "harmless." Disclose all supplements to prescribers — several interact with medications.
• Eating-disorder caution: an intense focus on "clean" or "perfect" eating for mental health can tip into restriction/orthorexia in vulnerable people. The goal is a sustainable, generous, whole-food pattern, not rigidity (see eating_disorder_body_image_diagnostic).

Controversy

Nature: A young field (nutritional psychiatry) with a real core finding wrapped in genuine methodological problems — plus a supplement industry that over-sells the periphery. Not a fringe-vs-establishment war; a question of how big and how clean the effect really is.

Position A — Nutritional-psychiatry proponents (Jacka and the field she founded; mostly academic/independent, some advocacy/dietetics-professional interest).
• Claim: diet is a modifiable risk factor for depression; improving diet is an effective, low-cost, side-benefit-rich adjunct (and possibly preventive) treatment.
• Best evidence: SMILES + HELFIMED + AMMEND (replicated, dose-responsive); Firth 2019 (EPA + methylfolate adjuncts); strong, coherent biochemistry.
• Limitation: the flagship effect size is implausibly large; the trials cannot be blinded; some advocacy enthusiasm runs ahead of the data.

Position B — Methodological skeptics (Molendijk et al.; the Cochrane omega-3 reviewers; epidemiologists; mostly independent/academic).
• Claim: the diet–depression link is heavily confounded and reverse-caused; the trial effects are inflated by unblindable expectation; supplement evidence is low-quality and publication-biased.
• Best evidence: the SMILES underpowering + undisclosed-recruitment + unblinding critiques; Cochrane's low/very-low GRADE on omega-3; VITAL-DEP's null prevention result; the observation that prior depression predicts later better diet (clean reverse-causation evidence).
• Limitation: can over-discount a finding that has now replicated across independent groups and populations with a consistent dose-response signal; "low-quality evidence" partly reflects how hard (and unfundable) clean diet RCTs are, not a tested negative.

The reverse-causation problem (the deepest issue, sitting under both positions): poor diet can be a cause of depression or a consequence of it (low energy, anhedonia → worse eating), and the relationship is almost certainly bidirectional. Diet, exercise, smoking, and socioeconomic status all co-travel with mood and with each other, so residual confounding is hard to eliminate even in well-adjusted studies. This is why intervention trials (which break reverse causation by randomising the diet) matter more than the mountain of observational correlations — and why even those trials' unblindability keeps the ceiling at Tier 2.

The funding/bias dimension (runs in two opposite directions — name both):
• On dietary pattern: whole food is unpatentable and unmarketable, so there is little industry money to fund the large, clean trials that would settle it — the classic "underfunded, not disproven" signature. The pivotal critiques are academic peers, not industry shills, which makes them credible rather than dismissible.
• On supplements: here the bias inverts — a large, lightly-regulated supplement industry profits from selling "mood" nutrients to the already-replete, and some individual omega-3 trials carry manufacturer ties. This is why the entry trusts the bias-corrected, EPA-specific, adjunctive pools (Firth) and the deficiency-correction framing, and is skeptical of "everyone should supplement for mood."

Realised Position: Food genuinely moves mood, and the mechanism-grounded, replicated core deserves confident coaching — but the honest version is specific and bounded. Improving a poor diet and adding EPA are real adjuncts in clinical depression (Tier 2, with the SMILES magnitude treated as inflated). Correcting a measured deficiency reliably lifts the mood/energy that deficiency was causing (test, then treat). Supplementing the already-replete for mood is mostly the supplement industry's revenue, not your recovery. And none of it replaces treatment in moderate-to-severe depression. See What Would Change Our Mind.

Cross-Pillar Connections

• Mental (depression_lifestyle_interventions, exercise_mental_health): diet is one lever in the lifestyle-depression toolkit; exercise has stronger, better-blinded evidence and should usually be co-prescribed. magnesium_supplementation_depression is the sibling single-nutrient entry; nac_n_acetyl_cysteine is the other emerging adjunct from Firth 2019.
• Diet (diet_foundations_for_baseline, dietary_fiber_diversity_and_microbiome_health, diet_gut_microbiome, blood_sugar_regulation): the dietary-pattern intervention is the Diet-pillar foundation; the gut-microbiome and glycaemic-stability entries carry the mechanism detail this entry references.
• Deficiency screening (micronutrient_deficiency_screening, omega3_repletion, vitamin_d3_high_dose_supplementation, light_exposure_vitamin_d): the "test before you supplement" step routes here; omega3_repletion owns the EPA/DHA repletion detail this entry points to.
• Differential diagnosis (thyroid_dysfunction, fatigue_cross_pillar_diagnostic): low mood + fatigue should trigger a check for hypothyroidism and the deficiency four before being treated as purely psychological — these entries own that triage.

What would change our mind

Falsifiability: explicit upgrade/downgrade criteria from source

We would UPGRADE the dietary-pattern claim (A) toward Tier 1 if:
• An adequately-powered RCT with a credible attention/expectation-matched active control (the SMILES control was already active-ish, which is good) reproduced a moderate effect — confirming the direction while shrinking the implausible d≈1.2 toward a believable size.
• Diet-intervention effects held up in a registered, pre-specified replication that closed the unblinding/recruitment gaps the SMILES critics flagged.

We would UPGRADE omega-3 (EPA) toward Tier 1 if:
• Large, independent (non-industry), low-risk-of-bias RCTs of EPA-predominant supplementation reproduced the adjunctive antidepressant effect, narrowing the gap between Firth's positive read and Cochrane's cautious one.

We would UPGRADE a deficiency nutrient toward Tier 1 if:
• RCTs restricted to confirmed-deficient participants (rather than mixed populations) showed consistent, replicated mood improvement on correction — the cleanest test, rarely done.

We would DOWNGRADE if:
• Better-blinded diet trials showed the QoL/mood benefit collapses toward the control once expectation is properly matched.
• The replication record reversed (a well-powered diet RCT clearly null), or omega-3's signal proved to be publication-bias artefact once the grey literature is included.
• Mendelian-randomisation / genetic-instrument work showed the diet–depression link is overwhelmingly reverse-caused/confounded rather than causal.

Industry bias note

Structural incentives the evidence base may reflect

This topic has bias pressure in both directions, and naming both is the value-add.
• Dietary pattern — underfunded, not disproven. Whole food is unpatentable; no one earns subscription revenue from "eat more vegetables." Large, clean diet RCTs are expensive and hard to fund and harder to blind, so the evidence is thinner than the plausibility — the classic suppression-by-neglect signature. Crucially, the most-cited critiques of SMILES are academic methodologists, not industry — so they are credible quality control, not a hit job, and the entry takes them seriously.
• Supplements — over-sold, especially to the replete. The supplement industry (lightly regulated, large) profits from selling "mood" nutrients to people who are not deficient, and a minority of omega-3 trials carry manufacturer ties. The honest defence is to (a) trust the publication-bias-corrected, EPA-specific, adjunctive pools over enthusiastic single trials, and (b) anchor the deficiency nutrients to measured low levels, which converts a marketing claim into a clinical one.
• The resolution Realised uses: weight independent/government/foundation-funded syntheses (Firth 2019 NIHR; Cochrane; VITAL-DEP NIH; the SMILES NHMRC funding) over industry-tied single trials, and frame nutrients as deficiency-correction and adjunct, never as monotherapy or as a population-wide mood pill.

Nutrition Mental Health