Reviews/Dry Eye Disease Treatment/Omega-3 does not improve dry eye disease symptoms
Omega 3 Supplementation · DED-OM3-02 · v2
Borderline: within 4 points of the Uncertain/Doubtful boundary, so a re-grade could land either side.
Whether taking omega-3 capsules by mouth eases dry-eye symptoms is genuinely unsettled in this set of studies. The largest and most carefully run treatment trial [1], a recent seven-site trial [11], and a study of what happens when people stop the supplement [5] all found no meaningful symptom improvement over placebo, and a very large prevention trial [7] found it did not stop dry eye from developing. Against this, many smaller trials did report clear symptom gains [3, 4, 6, 12, 13, 14]. These positive trials used different oils, different patients, and were mostly small and run at single centres, so they do not cleanly outweigh the big negative trials. Because strong evidence pulls one way and a larger number of weaker studies pulls the other, the flat statement that omega-3 does not help cannot be confirmed or ruled out.
The best-powered treatment trial [1] (N=535) found the OSDI change no different from olive-oil placebo (-1.9, 95% CI -5.0 to 1.1, P=0.21), and the multi-site Eom trial [11] was likewise null (12-week difference -3.9, 95% CI -10.90 to 3.13, P=.28), as was the withdrawal extension [5] (-0.6, 95% CI -10.7 to 9.5, p=0.91) and the large prevention trial [7] (HR 0.97, 95% CI 0.81-1.16). Yet several smaller RCTs report significant symptom benefit: [4] (65% vs 33% improved, P=0.005), [12] (26% vs 4% OSDI improvement, P=0.004), [13] (70% vs 37% became asymptomatic), [3] in the krill arm (-18.6 vs -10.5, P=0.02), [6] in visual-display-terminal users, and [14] after cataract surgery (P=0.026). The positive trials differ widely in product (krill, fish oil, flax blends, omega-3/omega-6 combinations, plant oils) and population. The signal therefore points in opposite directions depending on trial size and design. The contested point: The division is between the two highest-quality, best-powered treatment trials [1, 11], which find no symptom benefit, and a more numerous body of smaller, heterogeneous RCTs [3, 4, 6, 12, 13, 14] that do. The panel split on whether the weight of the large null trials or the volume of smaller positive trials should govern an absolute negation.
All 14 papers behind this claim with their DOIs and what each does to it, the panel’s arbitration, the evidence gap map and the open research questions. Free, and sent to your inbox.
We’ll also send you other claims from Dry Eye Disease Treatment, one at a time, and tell you when a release or a pivotal trial changes a grade. Unsubscribe by replying.
One reviewer's only job is to attack the finding. The challenges below were weighed before the grade was set.
The claim is an absolute negation, but at least six included RCTs ([3] krill arm, [4], [6], [12], [13], [14]) report statistically significant symptom/OSDI improvement, so 'does not improve' overreaches past a real positive signal.
The positive trials are mostly small, single-centre and heterogeneous in product (krill, flax-blend, plant oils, omega-3/omega-6 combinations) and population (post-cataract, VDT users), limiting their weight against the pooled-population DREAM null.
Where it remains uncertain. A larger number of smaller RCTs [3, 4, 6, 12, 13, 14] report significant symptom improvement, but they vary in product and population and individually carry less weight than [1].
Aleatory
Several positive trials are small and single-centre ([12] N=64, [13] N=36, [14] N=48), so their significant results carry wide uncertainty and could reflect chance.
Epistemic
Whether olive oil [1, 3] and grape-seed oil [11] are biologically inert is not established in the corpus, so the null results cannot be cleanly interpreted.
Inferential
Trials use krill [3], fish oil, flax blends [13], plant oils [8] and omega-3/omega-6 combinations [2]; pooling them into one verdict on 'oral omega-3' assumes equivalence the corpus does not show.
The load-bearing papers from a corpus of 14, each classified by what it does to the claim. Every DOI is taken from the corpus the claim was graded on.
Best-powered treatment RCT (N=535); null primary symptom outcome supports the claim.
“The mean change in the OSDI score was not significantly different between the active supplement group and the placebo group (-13.9 points and -12.5 points, respectively; mean difference in change after imputation of missing data, -1.9 points; 95% confidence interval [CI], -5.0 to 1.1; P=0.21).”
Panel auditInternal validity: high · scope of inference: Patients with moderate-to-severe dry eye disease taking 3000 mg fish-derived EPA/DHA for 12 months.Key limitations: olive oil comparator may not be inert · moderate-to-severe DED only
Withdrawal extension; no symptom penalty for stopping, consistent with no treatment effect.
“the mean change in OSDI score between month 12 and 24 was similar between treatment groups (mean difference in change -0.6 points, 95% confidence interval [CI], (-10.7, 9.5), p = 0.91)”
Panel auditInternal validity: medium · scope of inference: Patients who had already taken omega-3 for 12 months, studied for a further 12 months of continuation vs withdrawal.Key limitations: small sample, wide CI · withdrawal design not initiation · prior-user population
Large prevention trial null on incident severe symptoms; indirect support for no symptom effect.
“there was no difference between groups for the secondary end point of diagnosed DED plus incident severe DED symptoms (1044 participants [8.9%] with end points in the treated group vs 1074 [9.1%] with end points in the placebo group; hazard ratio, 0.97; 95% CI, 0.89-1.06).”
Panel auditInternal validity: high · scope of inference: US adults without prior DED diagnosis and not experiencing severe symptoms at baseline, i.e. a prevention rather than treatment question.Key limitations: prevention/incidence endpoint, not symptom treatment · 1 g/day dose lower than treatment trials · population free of DED at entry
Placebo-controlled symptom benefit in the krill arm; contradicts claim for one formulation.
“OSDI score was significantly reduced at day 90 relative to baseline in the krill oil group only, compared with placebo (-18.6±2.4 vs. -10.5±3.3; P = 0.02).”
Panel auditInternal validity: medium · scope of inference: Mild-to-moderate DED patients taking moderate-dose krill or fish oil for 3 months.Key limitations: small per-arm n · symptom benefit only in krill arm, not fish oil · single site
4 of 14 papers shown. The full review lists all of them, with the panel’s arbitration →
Claims are graded at each quarterly release, and re-graded between releases only when a pivotal trial lands. Every change is a new version of the claim.
High-quality landmark trials support the null while a substantial body of smaller, varied RCTs show symptom benefit, placing the claim at the contested Uncertain boundary.
Release 2026-q4 re-graded every claim from its corpus under protocol v0.19. An earlier release's grade is shown for the record; it came from an earlier protocol and is not a movement.
Assumptions that carry weight but are not themselves established.
The panel records the gap that would most move the claim, and the study that would close it.
Large symptom trial with a confirmed inert placebo. The null verdict rests partly on lipid comparators whose inertness is unestablished [1, 3, 11], leaving the direction of bias unknown. The panel rates this the highest-importance open gap for the claim, and names the study design that would close it: Adequately powered double-masked RCT of oral omega-3 versus a demonstrably inert, non-lipid placebo with OSDI as the primary endpoint.
Open question: Does oral omega-3 improve dry-eye symptoms in the most symptomatic patients even if it does not in the general DED population? Why it is still open: [2] shows larger improvement only in the OSDI>=52 subgroup and [1] is null overall, so a severity-dependent effect is neither confirmed nor excluded.
Priors is a sceptical first reader, not peer review, and not medical advice.
Release 2026-q4 · graded 21 Oct 2026 · protocol v0.19