What the Research Says About Marine Oil Supplements and Arthritis Pain

We only cite studies published in peer-reviewed journals. We summarize findings without overstating conclusions.

This article summarizes a systematic review and meta-analysis published on January 6, 2017, in the open-access journal Nutrients (MDPI), titled “Marine Oil Supplements for Arthritis Pain: A Systematic Review and Meta-Analysis of Randomized Trials.” The study was conducted by a ten-member international research team based primarily at the Musculoskeletal Statistics Unit of the Parker Institute at Bispebjerg and Frederiksberg Hospital in Copenhagen, Denmark, with additional investigators from the Department of Nutrition, Exercise and Sports at the University of Copenhagen; the Division of Rheumatology at the University of California, Los Angeles; the University of Texas MD Anderson Cancer Center in Houston; and the Cochrane Musculoskeletal Group in Ottawa, Canada. The review is a meta-analysis – a study design that mathematically combines the results of many individual randomized trials to produce a single, more statistically powerful overall estimate. The authors searched six major databases, ultimately analyzing 42 eligible randomized controlled trials encompassing 2,751 patients with various arthritis diagnoses. Their primary goal was to determine whether oral marine oil supplements reduce pain in arthritis patients. Secondary goals included assessing effects on physical function, inflammation, tolerability, and safety. The full text is freely available at PubMed Central (PMC5295086).

The authors declare no conflicts of interest. Funding came indirectly from the Oak Foundation, a philanthropic organization; no commercial supplement industry funding was involved. One co-author, Maria E. Suarez-Almazor, held a Midcareer Investigator Award from the National Institute of Arthritis and Musculoskeletal and Skin Diseases (K24 AR053593). The authors explicitly state that neither the Oak Foundation nor any other funder had any role in study design, data collection, analysis, interpretation, or manuscript preparation.

Background: Why Arthritis Pain Drives Patients Toward Supplements

Arthritis is a broad term describing joint disorders characterized by pain, swelling, stiffness, and restricted movement. In the United States alone, survey data from 2010–2012 indicated that 22.7 percent of American adults had a medically diagnosed form of arthritis. The two most common varieties are rheumatoid arthritis and osteoarthritis, and despite their shared symptom of joint pain, they arise through very different mechanisms.

Osteoarthritis is a degenerative disease in which the cartilage cushioning the ends of bones gradually breaks down, often from a combination of mechanical wear, aging, and metabolic factors. It does involve inflammation, but the inflammatory component is generally considered less dominant than in rheumatoid arthritis. Rheumatoid arthritis, by contrast, is an autoimmune disease – one in which the body’s immune system mistakenly attacks the lining of its own joints. This produces a form of joint inflammation called synovitis (from synovium, the tissue lining the joint cavity), which is typically more intense and systemic than anything seen in osteoarthritis. Research has demonstrated that the degree of synovitis is directly associated with the severity of pain that patients experience.

For both conditions, pain is consistently ranked by patients as their most debilitating symptom. Nonsteroidal anti-inflammatory drugs (NSAIDs) – a category that includes common medications like ibuprofen and naproxen – are widely used to manage this pain across arthritis diagnoses. However, NSAIDs carry known risks, including serious gastrointestinal complications such as ulcers and bleeding, and occasional cardiovascular adverse events. These risks create a practical need for alternative or complementary pain management strategies, which is partly why surveys have found that approximately 90 percent of arthritis patients have used, or were currently using, some form of complementary therapy.

Marine oil supplements – primarily fish oil capsules, but also products derived from mussels, krill, seals, and cod liver – represent one of the most widely used of these complementary therapies. Their appeal rests on a plausible biological rationale: they are rich in two specific omega-3 fatty acids, eicosapentaenoic acid (EPA) and docosahexaenoic acid (DHA), which have established roles in modulating the body’s inflammatory processes. Despite their widespread use, the authors of this review noted that four previous meta-analyses on the subject had produced contradictory conclusions – three finding benefit, one finding none – and none of them had applied the comprehensive search strategies and rigorous evidence-quality grading that modern standards require. This review was designed to fill that gap.

The Biology: How EPA and DHA Are Thought to Reduce Joint Inflammation

To understand why marine oils were plausible candidates for arthritis treatment, it helps to understand how the body produces and regulates inflammation at a chemical level.

The body manufactures a large family of signaling molecules called eicosanoids – compounds that include prostaglandins, thromboxanes, and leukotrienes. These molecules play central roles in regulating inflammation, among many other physiological functions. The raw materials from which eicosanoids are built are fatty acids, and the particular fatty acid used determines the character of the resulting signaling molecule.

Arachidonic acid (AA) is an omega-6 fatty acid that is the primary precursor to pro-inflammatory eicosanoids. When the body’s immune system detects injury or foreign invaders, it draws on stores of arachidonic acid to build the inflammatory signals that produce the redness, heat, swelling, and pain characteristic of the inflammatory response. The modern Western diet is typically high in omega-6 fatty acids from vegetable oils, which means the body has abundant arachidonic acid available to drive this process.

EPA and DHA – the omega-3 fatty acids concentrated in marine oils – compete with arachidonic acid for the same enzymes and the same cellular incorporation sites. When EPA and DHA levels in the body rise through supplementation, they partially displace arachidonic acid. The result is a shift in the types of eicosanoids the body produces: the mediators built from EPA and DHA tend to be less potently pro-inflammatory than those built from arachidonic acid, and some of them – called resolvins and protectins – actively promote the resolution of inflammation rather than its continuation. This shift in the balance from inflammatory to pro-resolving lipid mediators represents the primary proposed mechanism by which marine oil supplements might reduce joint inflammation and, consequently, joint pain.

It is worth noting that plant-based sources of omega-3 fatty acids, such as flaxseed oil, provide a different omega-3 called alpha-linolenic acid (ALA). The human body can technically convert ALA into EPA and DHA, but research has shown that this conversion is inefficient – producing only marginal amounts of the active fatty acids. For this reason, the authors of this review explicitly excluded plant-based omega-3 sources and focused solely on marine-derived oils, where the active EPA and DHA are present directly.

Methodology: How the Review Was Conducted

Study Registration and Reporting Standards

Before conducting any analyses, the research team registered a detailed protocol on PROSPERO (the international prospective register of systematic reviews), under the identifier CRD42015016817. This pre-registration is an important quality control step: it means the researchers committed in advance to their methods, outcomes, and analytical approaches, which reduces the risk of selectively reporting only the results that look favorable. The review was also conducted according to the PRISMA statement (Preferred Reporting Items for Systematic Reviews and Meta-Analyses), the established international standard for transparent reporting of this type of research.

Database Search

The search was conducted on February 24, 2015, across six databases: MEDLINE (the primary biomedical literature database), the Cochrane Library, EMBASE (a European biomedical database with particularly strong coverage of drug research), ClinicalTrials.gov, the World Health Organization’s International Clinical Trial Registry Platform (ICTRP), and the Web of Science. The researchers also manually searched the reference lists of relevant publications to catch any studies the database searches might have missed. Two reviewers independently screened all retrieved records, with disagreements resolved by consensus or by consulting a third reviewer.

Inclusion Criteria

To be included in the review, a study had to be a randomized controlled trial comparing an oral marine oil supplement against any control treatment in patients diagnosed with any form of arthritis. There was no restriction on participant age or sex, no restriction on the dose or type of marine oil used, and no minimum publication date. The intervention had to last at least two weeks. The authors included any marine oil of marine origin – whole fish oil, cod liver oil, mussel extract, seal oil, and krill oil all qualified. Vegetable-derived omega-3 sources were explicitly excluded for the reasons described above.

Outcomes Measured

The primary outcome was pain, which the authors recognized as the most patient-relevant and consistently assessed outcome across all arthritis types. Secondary outcomes included physical function, markers of inflammation (with C-reactive protein, or CRP, the preferred measure), tolerability (measured as the proportion of patients who completed the full trial), withdrawals due to adverse events, and the occurrence of serious adverse events (SAEs).

Statistical Approach

Because the individual trials used different scales to measure pain, function, and inflammation, the researchers could not simply average their raw numbers. Instead, they converted each trial’s results into a standardized mean difference (SMD), a unit-free measure that expresses the magnitude of the treatment effect relative to the natural variation in the data. An SMD of 0 means no effect; negative SMDs for pain and inflammation indicate that the supplement group fared better (less pain, less inflammation); positive SMDs for function indicate improvement.

To account for the fact that the included trials were heterogeneous – conducted in different populations, at different doses, for different durations – the authors used a random-effects meta-analysis model based on restricted maximum likelihood (REML) estimation, which is the more conservative and appropriate statistical approach when studies are not expected to be measuring exactly the same effect in exactly the same way. Heterogeneity itself was quantified using the I² statistic, which expresses what percentage of the variation in results across trials reflects genuine differences between studies (rather than just random sampling error). An I² above roughly 50 percent is generally considered substantial.

Meta-regression analyses – a technique that tests whether specific trial characteristics, such as diagnosis type, oil type, or EPA/DHA dosage, explain some of the variation in results – were used to explore sources of heterogeneity. The overall quality of the body of evidence was rated using the GRADE framework (Grading of Recommendations Assessment, Development and Evaluation), which categorizes evidence as high, moderate, low, or very low quality based on factors including risk of bias, inconsistency, imprecision, and potential for publication bias.

Risk of Bias Assessment

Each included trial was assessed using the Cochrane Risk of Bias tool, which evaluates nine domains of potential bias including adequacy of randomization, allocation concealment, blinding of participants, blinding of personnel, blinding of outcome assessors, completeness of outcome data, selective outcome reporting, and other sources of bias. A separate tool developed by Dwan and colleagues was used to assess outcome-reporting bias – the tendency of researchers to measure an outcome but report it only when the result is statistically significant.

Results: What the Evidence Showed

Study Selection and Characteristics

The database search identified 3,389 records. After removing those that did not meet the inclusion criteria at the title and abstract level, 78 records describing 65 trials were considered for full-text review. Of these, 51 records describing 42 trials contained sufficient data for inclusion in the meta-analyses. An additional nine trials were identified through trial registry searches but had not been published and could not yield data.

The 42 included trials varied considerably in their design and populations. Treatment durations ranged from 2 weeks to 18 months. EPA doses ranged from 0.013 to 4.050 grams per day, and DHA doses ranged from 0.010 to 2.700 grams per day. Most trials used oil derived from whole fish, but some used cod liver oil, green-lipped mussel extract, seal oil, or krill oil. In terms of diagnosis, 32 trials focused exclusively on rheumatoid arthritis, 6 on osteoarthritis, and 4 on other or mixed arthritis types (including juvenile arthritis, psoriatic arthritis, and mixed RA/OA populations). Across all included trials, 2,751 patients participated, with a mean age of 53.8 years and a mean disease duration of 9.7 years.

Notably, no trial was judged to have low risk of bias across all nine domains assessed. Thirteen trials were judged to have a high risk of outcome-reporting bias for pain specifically. This widespread methodological weakness is an important context for interpreting all subsequent findings.

Primary Outcome: Pain

Thirty of the 42 trials provided complete data on pain and were included in the primary meta-analysis. The pooled result showed a statistically significant pain-reducing effect of marine oil supplementation, with an overall SMD of −0.24 (95% confidence interval: −0.42 to −0.07; p = 0.007). To put this effect size in concrete terms, the authors translated it into an estimated 8 percent improvement in pain scores on a 100-millimeter visual analog scale (VAS) – a horizontal line on which patients mark their pain from “no pain” at one end to “worst imaginable pain” at the other.

An SMD of −0.24 is classified as a small effect by conventional benchmarks. It is statistically significant – meaning it is unlikely to be a chance finding – but the practical size of the benefit requires context. An 8 percent improvement in pain for an individual patient is modest, though across a large population it may be meaningful. Critically, the confidence interval (−0.42 to −0.07) is entirely on the beneficial side of zero, confirming that the overall direction of the evidence favors marine oil over control. However, the prediction interval – a related statistic that estimates the range of effects one might plausibly expect in a future study – was very wide, from −1.05 to +0.57, indicating that some future trials may well find no benefit or even a slight harm.

Heterogeneity was substantial (I² = 63%), meaning the individual trials disagreed considerably with one another in their results. This heterogeneity was not unexpected given how varied the included trials were, but it means the overall estimate should be interpreted cautiously.

Pain Results by Arthritis Diagnosis

Meta-regression analyses revealed that the type of arthritis diagnosis was a statistically significant predictor of how much benefit was seen from marine oil (p = 0.024), and the results differed meaningfully across the three diagnosis groups.

In rheumatoid arthritis, 22 trials with 956 patients produced a statistically significant beneficial effect: SMD −0.21 (95% CI: −0.42 to −0.004). This narrowly reached statistical significance. When only RA trials were analyzed together, heterogeneity dropped substantially to I² = 32%, meaning the RA trials were considerably more consistent with one another than the full mixed pool. The GRADE rating for this finding was moderate quality – the highest rating achieved anywhere in this review. Moderate quality evidence means the research team had reasonable confidence in the direction of the effect but acknowledged that further research could alter the estimate.

In osteoarthritis, only 5 trials with 403 patients were available, and the result was not statistically significant: SMD −0.17 (95% CI: −0.57 to 0.24). The confidence interval crosses zero, meaning a null effect or a small negative effect cannot be ruled out. Heterogeneity was very high for OA (I² = 82%). The GRADE rating for this finding was low, indicating limited confidence in the estimate and substantial uncertainty. The authors noted that osteoarthritis itself is heterogeneous – it varies in the degree of inflammation present – which may partly explain why results were so inconsistent across the handful of OA trials available.

In patients with other or mixed arthritis diagnoses (juvenile arthritis, psoriatic arthritis, and mixed populations), 3 trials produced the largest pooled effect of any subgroup: SMD −0.63 (95% CI: −1.20 to −0.06). This was statistically significant, but the very small number of trials, very high heterogeneity (I² = 89%), and wide confidence interval mean this finding should be interpreted with extreme caution. The GRADE rating was low.

The Interesting Dose-Response Finding

One of the more counterintuitive findings from the meta-regression was the relationship between dose and effect. A statistically significant positive association was found between total EPA plus DHA dose and the SMD for pain (slope β = 0.13 per gram per day; 95% CI: 0.04 to 0.22; p = 0.006). Because a positive slope means the SMD was moving toward zero (less benefit) as dose increased, this translates to an inverse dose-response: higher doses were associated with less pain reduction, not more.

The authors acknowledged this was unexpected and offered several possible explanations. First, some of the highest-dose trials were conducted in Norwegian populations, who already consume relatively large amounts of fish and fish oil in their regular diet. Adding a high-dose supplement on top of an already omega-3-rich background diet may produce a smaller marginal effect than the same supplement in a population eating a more omega-3-depleted Western diet. Second, two of the highest-dose trials required participants to reduce their NSAID use during the trial, which may have masked the analgesic effect of the oil by simultaneously withdrawing a pain-relieving medication. Third, there is a concept of a saturation threshold in fatty acid biology – the body’s capacity to incorporate EPA and DHA into cell membranes is not unlimited, and beyond a certain point, additional supplementation may produce diminishing returns. The authors cautioned that none of these explanations could be confirmed from the available data, and that the inverse relationship should not be taken as firm evidence that lower doses are preferable.

Importantly, no significant duration-response relationship was found (p = 0.568), meaning the length of supplementation – across the range of durations studied – did not predict how large the pain-reducing effect would be.

EPA/DHA Ratio and Type of Marine Oil

The EPA-to-DHA ratio of the supplement emerged as a significant predictor of pain outcomes (p = 0.031). Marine oils with an EPA/DHA ratio greater than 1.5 – meaning substantially more EPA than DHA – showed statistically significant pain reduction on their own. This finding aligns with some mechanistic evidence suggesting that EPA may have a more potent direct effect on inflammatory pathways than DHA, though both are active compounds and the picture is not fully resolved.

A post-hoc analysis (meaning one conducted after the primary analyses were pre-planned, and therefore requiring more cautious interpretation) examined whether the specific source of marine oil mattered. Grouping oils into whole fish, mussel extract, other (cod liver oil, krill oil, seal oil, and mixtures), and unspecified, the analysis found that the type of oil was a significant predictor of pain outcomes (p = 0.012). Of the oil types, only mussel oil achieved statistical significance on its own as a beneficial treatment. This finding should be interpreted with caution, however, as only three trials used mussel extract, and a post-hoc finding in a small subgroup is hypothesis-generating rather than confirmatory.

Secondary Outcome: Inflammation

Twenty-five trials provided complete data on inflammation, and the pooled analysis showed a statistically significant benefit: SMD −0.28 (95% CI: −0.51 to −0.06; p = 0.013). The authors translated this into an estimated reduction in C-reactive protein (CRP) – a standard blood marker of inflammation – of approximately 1.7 mg/dL (95% CI: −3.0 to −0.4 mg/dL). CRP is measured in milligrams per deciliter of blood, and levels below 1.0 mg/dL are generally considered normal in healthy individuals, while arthritis patients often have levels of 2 to 10 mg/dL or higher during active disease. A reduction of 1.7 mg/dL is biologically plausible as meaningful, though substantial heterogeneity was again present (I² = 70%). Sensitivity analyses including trials with incomplete inflammation data and high outcome-reporting bias produced a somewhat smaller but still significant effect (SMD −0.16; 95% CI: −0.29 to −0.02; p = 0.021).

Secondary Outcome: Physical Function

Twenty-three trials reported on physical function. The pooled result showed no significant effect of marine oil on function: SMD −0.01 (95% CI: −0.19 to 0.18; p = 0.953), with moderate heterogeneity (I² = 60%). Including trials with incomplete data produced nearly identical null results (SMD −0.01; 95% CI: −0.13 to 0.10; p = 0.808). The GRADE rating for function was moderate quality, reflecting the higher consistency of the function data relative to pain, even though the effect itself was absent. Post-hoc meta-regression found that much of the heterogeneity in function outcomes could be attributed to diagnosis type and trial duration, though the overall conclusion – no meaningful improvement in physical function – remained stable.

Safety and Tolerability

The safety profile of marine oil supplements emerged as one of the more reassuring findings. Across the 28 trials with complete tolerability data, the proportion of patients who completed the trial was essentially identical between the marine oil and control groups, with a risk ratio (RR) of 1.00 (95% CI: 0.96 to 1.03; p = 0.814). For withdrawals due to adverse events, 21 trials were analyzed, and the risk ratio was 0.82 (95% CI: 0.57 to 1.17; p = 0.279), indicating no significant difference between groups. For serious adverse events, 24 trials were analyzed, producing a risk ratio of 0.75 (95% CI: 0.43 to 1.30; p = 0.308) – again, no significant difference. Heterogeneity was very low for all three safety outcomes (I² = 6.5%, 0%, and 0% respectively), meaning the trials were highly consistent on this point. This finding aligns with a prior systematic review of fish oil administration in older adults that similarly found no safety signal.

Publication Bias Assessment

The authors assessed the risk that small, unpublished negative studies were distorting the pooled results (so-called publication bias) using funnel plots and the Egger test. For pain, function, and inflammation, neither the funnel plots nor the Egger test showed evidence of asymmetry, suggesting that publication bias was not a major driver of the positive pain and inflammation results. Some asymmetry was apparent for the safety outcomes, but this was attributed to outcome-reporting bias rather than suppression of negative overall findings.

The Limitations: What the Study Cannot Tell Us

The authors were explicit and thorough in acknowledging the limitations of their conclusions, and these deserve careful attention by any reader attempting to apply the findings.

The most consequential limitation is the quality of the underlying trials. Not a single one of the 42 included trials achieved low risk of bias across all nine domains assessed. Thirteen trials had high risk of outcome-reporting bias for pain specifically – meaning their investigators measured pain but chose whether to report it partly based on how the numbers came out. This kind of selective reporting can distort meta-analyses even when publication bias is absent.

The OA evidence base was too thin to draw firm conclusions. Only five trials examined osteoarthritis patients, with high heterogeneity and a non-significant overall result. Given that OA is the more prevalent of the two major arthritis types, and given that it varies widely in its inflammatory character, the five available trials may simply not represent the full spectrum of OA patients.

The dose-response analysis was complicated by factors the study could not fully account for: participants’ background dietary intake of omega-3 fatty acids, actual compliance with the supplement regimen (which many trials did not rigorously monitor), and the concurrent use or withdrawal of NSAIDs. Any of these factors could obscure or distort the true dose-effect relationship.

The optimal type of marine oil, the optimal dose, and the optimal treatment duration could not be determined with confidence from the current evidence. Too few trials used non-fish-based marine oils (mussel, krill, seal, cod liver) to draw reliable subgroup conclusions, and the post-hoc nature of the oil-type analysis means the mussel oil finding needs prospective confirmation before it can be acted upon.

The review’s search was conducted in February 2015, so trials published in the subsequent two years before this review’s publication in January 2017 were not included. As the authors note, more high-quality research – particularly in OA and psoriatic arthritis – is needed before clinical guidance can move beyond the tentative conclusions this review supports.

Finally, while the safety data were reassuring, the review was not designed to characterize the specific types of adverse effects that did occur in individual trials, only to compare overall rates. Patients with specific medical conditions (such as those taking blood-thinning medications, for whom fish oil’s mild anticoagulant effect may be relevant) should discuss supplement use with their treating physician.

Summary of Key Takeaways

• This meta-analysis pooled data from 42 randomized trials and 2,751 arthritis patients across six major databases, making it the most comprehensive such analysis of marine oil supplementation in arthritis published to that point.
• Overall, marine oil supplements produced a small but statistically significant reduction in arthritis pain compared to control treatments: an SMD of −0.24 (95% CI: −0.42 to −0.07; p = 0.007), translating to an estimated 8 percent improvement on a pain scale.
• The most reliable finding applied to rheumatoid arthritis specifically: 22 RA trials showed a statistically significant pain benefit (SMD −0.21), with substantially lower heterogeneity (I² = 32%) and GRADE-rated moderate quality evidence – the highest quality rating in the review.
• For osteoarthritis, only 5 trials were available, the overall result was not statistically significant, heterogeneity was very high (I² = 82%), and the evidence was rated low quality. Marine oil cannot be confidently recommended for OA on the basis of current evidence.
• Marine oil supplementation was associated with a statistically significant reduction in the blood inflammation marker CRP (estimated reduction ~1.7 mg/dL), but showed no meaningful effect on physical function.
• Products with a higher ratio of EPA to DHA (greater than 1.5:1) appeared more beneficial for pain than products with a lower ratio, suggesting EPA may be the more active anti-inflammatory component.
• Counterintuitively, higher total doses of EPA plus DHA were associated with smaller pain reductions. The authors proposed several explanations, including dietary background effects and NSAID withdrawal in high-dose trials, but the finding remains incompletely understood.
• Marine oil supplements showed no significant difference from control in tolerability, withdrawal rates due to adverse events, or serious adverse events – a consistent finding across trials (I² near zero for all safety outcomes).
• The overall evidence base was rated low quality due to widespread methodological weaknesses in the underlying trials. Moderate quality evidence exists only for the RA pain finding. Substantially more well-designed trials – especially in OA – are needed before strong clinical recommendations can be made.

Senftleber, Ninna K., Sabrina M. Nielsen, Jens R. Andersen, Henning Bliddal, Simon Tarp, Lotte Lauritzen, Daniel E. Furst, Maria E. Suarez-Almazor, Anne Lyddiatt, and Robin Christensen. “Marine Oil Supplements for Arthritis Pain: A Systematic Review and Meta-Analysis of Randomized Trials.” Nutrients, vol. 9, no. 1, 2017, article 42. https://doi.org/10.3390/nu9010042. Full text available at: https://pmc.ncbi.nlm.nih.gov/articles/PMC5295086/.