Fish oil: what the prescriber needs to know
© BioMed Central Ltd 2005
Published: 21 December 2005
Skip to main content
© BioMed Central Ltd 2005
Published: 21 December 2005
There is a general belief among doctors, in part grounded in experience, that patients with arthritis need nonsteroidal anti-inflammatory drugs (NSAIDs). Implicit in this view is that these patients require the symptomatic relief provided by inhibiting synthesis of nociceptive prostaglandin E2, a downstream product of the enzyme cyclo-oxygenase (COX), which is inhibited by NSAIDs. However, the concept of 'safe' NSAIDs has collapsed following a multiplicity of observations establishing increased risk for cardiovascular events associated with NSAID use, especially but not uniquely with the new COX-2-selective NSAIDs. This mandates greater parsimony in the use of these agents. Fish oils contain a natural inhibitor of COX, reduce reliance on NSAIDs, and reduce cardiovascular risk through multiple mechanisms. Fish oil thus warrants consideration as a component of therapy for arthritis, especially rheumatoid arthritis, in which its symptomatic benefits are well established. A major barrier to the therapeutic use of fish oil in inflammatory diseases is ignorance of its mechanism, range of beneficial effects, safety profile, availability of suitable products, effective dose, latency of effects and instructions for administration. This review provides an evidence-based resource for doctors and patients who may choose to prescribe or take fish oil.
Essential dietary constituents are those that cannot be synthesized endogenously. Vitamins are familiar examples of essential micronutrients. The dietary essential fatty acids are polyunsaturated fatty acids (PUFAs) that contain the n6 with or without the n3 double bond, neither of which can be synthesized endogenously. The n6 (or ω6) PUFAs contain the n6 double bond, and the n3 (or ω3) PUFAs have both n6 and n3 double bonds. (The n or ω notation refers to the position of the double bond relative to the methyl terminus of the fatty acid molecule.) In contrast to vitamins, n6 and n3 fatty acids are macronutrients, and diets in industrialized Western countries are generally abundant in n6 PUFAs and poor in n3 PUFAs. This is potentially important because the ratios of these fatty acids in the tissues are determined largely by their ratios in the diet [1, 2].
Dietary sources of fatty acids
Foods and ingredients
Fatty acids contained in the foods
Fish and/or fish oil
Long chain n3 PUFAs such as EPA (C20:5n-3) and DHA (C22:6n-3)
EPA and DHA are the beneficial n3 PUFAs
Flaxseed and canola oil
The shorter chain n3 PUFA ALA
ALA is converted to EPA or DHA after ingestion, but not very efficiently. However, it can still provide a useful dietary source of EPA and DHA precursor. Whether it has a direct beneficial effect is unknown
Olive and canola oil
The MUFA OA (C18:1n-9)
OA has a neutral effect on n-3 PUFA metabolism and incorporation into tissues; therefore, it provides a useful 'background' dietary fat for maximizing n3 tissue content from dietary n3 PUFAs
Sunflower, peanut, soybean and cottonseed oil
The n6 PUFA LA (C18:2n-6)
Intake in modern Western diets is generally high and far in excess of what is required to prevent deficiency. Dietary LA can decrease conversion of dietary ALA to tissue EPA and can decrease tissue levels of EPA and DHA. LA is a precursor of AA (C20:4n-6), which is a metabolic antagonist of EPA
Because Western diets are typically low in LC n3 PUFAs, substantial increases in tissue LC n3 can be achieved by taking a fish oil supplement without further dietary modification . However, choice of spreads that are rich in n3 PUFAs or rich in MUFAs and low in n6 PUFAs allows higher tissue n3 levels to be reached with a given dose of fish oil [3, 4]. To achieve anti-inflammatory doses of LC n3 PUFAs by eating fish, a more substantial intake is required than would be practical for most people. The conversion of C18 n3 PUFAs to C20 and C22 n3 PUFAs occurs relatively inefficiently in humans, and so vegetable sources of dietary n3 PUFAs alone fail to achieve the tissue levels seen with fish oil .
The pain of arthritis is mediated in part by prostaglandin (PG)E2 – a nociceptive factor that is synthesized at sites of inflammation through the inducible isoform of cyclo-oxygenase (COX), namely COX-2. The COX isozymes, whether COX-1 or COX-2, are inhibited by nonsteroidal anti-inflammatory drugs (NSAIDs).
The effect on PGE2 may explain in part the symptomatic benefit of fish oil seen in rheumatoid arthritis (RA) [7, 8] (see the section on clinical evidence for the anti-inflammatory effects of fish oil, below) and the reduced discretionary use of NSAIDs seen in RA patients taking anti-inflammatory doses of fish oil [9–11].
The development of selective COX-2 inhibitors for use as NSAIDs with reduced or no upper gastrointestinal adverse effects was predicated on the observation that PGE2 at inflammatory foci was COX-2 derived, whereas gastro-protective PGE2 was COX-1 derived. This scheme fails to acknowledge that PGH2 and not PGE2 is the immediate product of AA metabolism by COX, and that PGH2 is the common precursor for many eicosanoids, including vasoactive and platelet active TXA2 and prostacyclin (PGI2; Fig. 2). For reasons explained by the enzymology of the individual synthases, antithrombotic PGI2 is mainly COX-2 derived whereas prothrombotic TXA2 is mainly COX-1 derived . Thus, selective COX-2 inhibitors suppress prostacyclin synthesis but not thromboxane synthesis [13, 14], which is a potential mechanism for the excess of adverse cardiovascular events seen with use of the coxibs . This outcome is not observed with use of dietary EPA as fish oil.
Other inflammatory mediators whose production is inhibited by fish oil are the cytokines tumour necrosis factor (TNF)-α and interleukin-1β, which are involved not only in production of inflammatory signs and symptoms but also in cartilage degradation (Fig. 3) [18–21]. In contrast to its inhibition by fish oil, TNF-α synthesis by monocytes is increased by NSAIDs .
Fish oil has been shown to reduce symptoms in RA in a dose-dependent manner [8, 23], relapse rates in Crohn's disease  and progression to renal failure in immunoglobulin A nephropathy . Fish oil improves control in systemic lupus erythematosus  and has a preventive effect when given prophylactically to mice genetically predisposed to lupus .
Investigations across a variety of inflammatory diseases have used doses of fish oil that provide daily intakes of LC n3 PUFAs that range from less than 1 g to more than 6 g . Collectively, these studies indicate that the anti-inflammatory dose of fish oil requires delivery of 2.7 g or more of LC n3 PUFAs daily, and that higher doses are also safe and effective. A daily intake of 2.7 g EPA plus docosahexaenoic acid (DHA) is provided by a daily dose of nine or more standard fish oil capsules, which typically contain 30% LC n3 PUFAs w/w. People who self medicate with fish oil generally take one or two capsules daily. This is insufficient for an anti-inflammatory effect but it may provide cardiovascular benefit.
The symptomatic benefit of fish oil in RA can be delayed 2–3 months . Earlier improvement with higher doses suggests a possible loading effect. It is important that potential users understand that this delay exists.
Fatty acid information for food choices
Cooking oils, salad dressings and spreads
Canola oil products
ALA (18-carbon n3 PUFA)
Olive oil products
Sunflower, cottonseed, peanut, soybean oil products
LA (18-carbon n-6 PUFA)
Preprepared food such as frozen chips/fries
Nutrient information on the packet will allow a choice of foods prepared in canola or olive oils
Because fish oil is rich in long chain n3 fats, fatty fish (e.g. sardines, herrings) have higher n3 content than lean fish. However, all marine fish contain long chain n3 fats. Canned fish have n3 fat content also, but note that canned tuna has less fat (and therefore less n3 fat) than fresh tuna
EPA (20-carbon n3 PUFA)
DHA (22-carbon n-3 PUFA)
Nuts and seeds
High in n3 PUFAs
Some n3 PUFAs but also n6 PUFAs
High in MUFA, low in n6 PUFAs
Peanuts, cashews, brazil and hazel nuts
Some MUFA, but also n6 PUFAs
Cost has been a major impediment to use of fish oil in anti-inflammatory doses. At the time of the major trials of fish oil in RA in the 1980s and 1990s, the cost for 1 g fish oil capsules was typically 30 c per capsule. For 15 capsules of fish oil daily (an average dose for RA trials showing benefit), the annual cost at this price is about A$1650 per annum. For most users this cost is prohibitive, and is particularly discouraging when most other treatments are government subsidized. The Pharmaceutical Benefits Scheme in Australia makes subsidized drugs, irrespective of actual cost, available to consumers at A$4.60 for Health Cardholders and A$28.60 for others for a typical 1-month supply. Fish oil, although available without prescription and with unrestricted access, is thus far more expensive to users than even highly expensive subsidized drugs. Recently, the cost of fish oil capsules has fallen. Products are now available that provide large numbers of capsules at an average cost as low as 10 c per capsule. Although this has reduced the annual cost for an anti-inflammatory dose to about A$550, the cost remains almost twice what non-Health Cardholders pay for most pharmaceutical products. The cost can be reduced substantially by using bottled fish oil taken on juice, a 15 ml daily dose of which costs about A$150 per annum. This makes fish oil somewhat less expensive than many standard medications for non-Health Cardholders but more expensive than the approximately A$55 per annum paid by Health Cardholders per medication below the safety net.
Oils derived from marine fish oil all contain LC n3 PUFAs. Standard fish oil is extracted from fish bodies and typically contains EPA 18% and DHA 12% w/w. Until recently, this was available only in capsules, but now a bottled preparation is available in Australia. Cod liver oil is widely available as both bottled oil and in capsules. It contains approximately 10% EPA and 10% DHA, and so it is also a good source of LC n3 PUFAs. However, at anti-inflammatory doses cod liver oils, which are rich in the fat-soluble vitamins A and D, contain more vitamin A than recommended intakes. Although the amount does not cause symptoms of toxicity, similar doses have been associated with reduced bone density and increased risk for hip fracture in epidemiological studies . This is not a problem with fish body oils, which contain very little of these fat-soluble vitamins.
Pour 30–50 ml juice (e.g. orange, tomato, apple, etc.) into two small 'shot' glasses.
Layer the desired dose of fish oil onto the juice in one glass – do not stir.
Swallow the juice and fish oil with a single gulp, avoiding contact with the lips (where the fish oil can be tasted).
Immediately sip the juice in the other glass slowly through the lips. This will remove any oil from the lips.
Take the fish oil immediately before a solid meal and without further fluid. This avoids floating of the oil on fluid in the stomach and favours mixing of the fish oil with food and passage from the stomach into the intestine. If reflux (repeating taste) becomes a problem, then split the dose before morning and evening meals. Alternatively, take the dose then lie on the left side for at least 15 min. In this position the oil floats into the passage from the stomach to the small intestine.
Fish oil (obtained from the body of the fish) is preferable to cod liver oil, which can deliver undesirable amounts of vitamin A at anti-inflammatory doses.
The repeating taste of fish oil arises from its low specific gravity, which is less than that of water. Thus, fish oil will float on free fluid with the stomach, in the same way that it floats on juice within a glass. Thus, when an eructation occurs to vent the stomach of swallowed gas, fish oil at the gas–fluid interface in the stomach may be partly regurgitated and tasted. This experience can be minimized by avoiding unnecessary fluids at the time of ingestion of fish oil, avoiding aerated drinks and by taking fish oil immediately before a meal. The latter strategy allows fish oil to mix with food, with which it exits from the stomach into the small bowel. These measures are generally effective in avoiding a 'repeating' fish oil taste. In cases where a problem still exists, passage of fish oil into the duodenum can be facilitated by lying in the left lateral decubitus position; this allows the oil to float into the duodenum, which is above the stomach in this position . Some may have a lesser problem with capsules than fish oil on juice but these can also be problematic because fish oil is released from capsules within the stomach. Some patients with persistent oesophageal reflux may not be able to take fish oil.
The odour of fish oil can be minimized by keeping fish oil refrigerated once open and taking it quickly once the fish oil on juice technique is mastered.
Fish oil, like any fat, is rich in calories. However, most people eat to satiety. In our Early Arthritis Clinic, a cohort of 33 RA patients taking fish oil at the rate of 15 ml/day immediately before or during a meal did not increase their mean weight over 1 year; there was a nonsignificant mean change of -0.4 kg from baseline to 1 year. Metabolic studies suggest the LC n3 PUFAs present in fish oil can reduce adipocyte numbers and the contribution of adipose tissue to body mass .
Because most anti-inflammatory drugs can have adverse effects on the foetus, they are generally withdrawn during pregnancy and lactation. Early observations in patients with active RA suggest a tendency toward lessening disease activity in pregnancy. This improvement presumably results from release of immunosuppressive factors that are generated during pregnancy, putatively to prevent immunological rejection of the foetus. In the modern era, in which RA is generally well suppressed by medications, withdrawal of medication in anticipation of and during pregnancy often results in increased disease activity. It is therefore appropriate to consider the safety of fish oil in pregnancy, either as an alternative or as one component of established treatment that may be continued.
LC n3 PUFAs are strongly represented among neural lipids. Neural tissue forms a disproportionately high proportion of body weight in foetuses and, relative to adults, neural development is particularly active in utero and during infancy. n3 PUFAs provided through placental transfer to the foetus or in breast milk, which is rich in LC n3 PUFAs, supports requirements for this development. As a result the possibility of depletion of maternal LC n3 PUFA stores exists. There is a dramatic fall in maternal plasma DHA in the immediate postpartum period, which is a time when relapse or onset of RA is more frequent, especially in women who breast feed [31, 32]. Although there are no studies comparing women receiving fish oil with control women, hypothetically n3 depletion could play a role.
Populations with high maternal n3 intakes have higher infant birth weight . In premature infants, breast feeding and n3-enriched infant formula have been associated with accelerated neural development compared with their counterparts given n3 PUFA poor formula . Although there are no studies into the anti-inflammatory effects of fish oil in pregnancy, symptomatic benefit in RA studies, in which women typically outnumber men, is well established. Thus, there is a rationale for use of fish oil in pregnant and lactating women with RA, and there is no evidence of harm at supplementation levels of at least 2.7 g/day of LC n-3 PUFAs . This dose is provided by 10 ml bottled fish oil (not cod liver oil) or the equivalent 9 or 10 standard fish capsules per day, and is in the range of doses shown to have anti-inflammatory effect. There is no evidence higher doses would be toxic to mother or foetus, but a conservative approach would be to limit the fish oil dose to this level. In pregnancy oesophageal reflux is common. Accordingly, attention to time of dosing (preferably morning or middle of the day and immediately before or during the early phase of a low fat meal) and reduced amount of juice with the dose and its chaser may be especially important.
There are several potentially useful drug–fish oil interactions relevant to the management of arthritis.
As discussed under biochemical rationale (see above), fish oils contain the natural COX inhibitor EPA, which inhibits both COX-1 and COX-2 activity. The different effects of EPA and NSAIDs on synthesis of downstream products are consistent with the known cardioprotective effect of fish oil and increased cardiovascular risk associated with NSAIDs (especially those that are COX-2 selective). Fish oils have been shown to reduce discretionary NSAID use for analgesia by about 50%. Fish oil has not been associated with gastric irritancy. NSAIDs tend to cause a moderate increase in systemic blood pressure, whereas fish oil reduces blood pressure by a similar amount [35, 36].
The most common dose-limiting effects of cyclosporin are hypertension and impaired renal function. These effects appear to be in part thromboxane mediated . Fish oil inhibits TXA2 production and reduces both the hypertensive and nephrotoxic effects of cyclosporin .
Fish oil in anti-inflammatory doses inhibits TNF and interleukin-1 synthesis by peripheral blood mononuclear cells stimulated ex vivo [18, 20, 21]. A rationale therefore exists for concomitant use of fish oil and TNF blockers. To date, this combination has not been evaluated in formal clinical trials.
Gastrointestinal toxicity is common with methotrexate therapy and is often dose limiting. Animal studies show that LC n3 PUFAs reduce loss of appetite, weight loss and gastrointestinal damage associated with methotrexate therapy .
Intolerance to fish oil is not unusual and occurs in about 15% of patients offered this treatment. Unwanted effects include repeating taste, retrosternal burning, diarrhoea, aversion to odour and taste, headache and failure to mask taste. Some patients are unable to accept the idea of taking fish oil. Serious unwanted effects have not been reported.
In a long-term study of fish oil in RA (>3 years), the continuation rate for fish oil was about 80% (unpublished data). This compared favourably with the continuation rates for each of the first-line disease-modifying antirheumatic drugs used concurrently, namely methotrexate, sulpha-salazine and hydroxychloroquine.
Although fish oil has not been associated with any serious acute treatment related syndromes, its long-term use raises several theoretical and practical concerns. These are discussed below.
A dose of 3 g/day EPA plus DHA has been assessed as safe for general consumption . Greenland Inuits consuming their aboriginal diet of sea mammals, sea birds and fish ingest 7 g/day LC n3 PUFAs . These Inuits appear to have a bleeding tendency, which may contribute to an observed increase in apoplexy (cerebral haemorrhage) . The very high consumption of LC n3 PUFAs in this population occurs within the context of a low n6 PUFA intake. The equivalence of AA and EPA in Inuit platelet cell membranes (AA:EPA ratio 1:1) was not reached closely by Australian patients taking 4.5 g fish oil for RA (AA:EPA ratio 4:1) for more than 3 years, although this ratio is substantially different from that in healthy Australian control individuals not taking fish oil and consuming an ordinary diet (AA:EPA ratio 40:1) [19, 43]. The Inuits have a very low frequency of myocardial infarction (relative risk 0.075 compared with Danish control individuals), which appears to be due in major part to dietary PUFAs . They also have a low frequency of inflammatory diseases. For patients with a chronic inflammatory disease such as RA, which is associated with high cardiovascular risk , the reduced cardiovascular risk with and anti-inflammatory effect of fish oil is likely to yield an overall long-term advantage. The disease-modifying effect of fish oil in RA, positive or negative, is unknown. However, the inhibitory effect of anti-inflammatory doses of fish oil on TNF and interleukin-1 synthesis provides the potential basis for a favourable long-term effect on disease progression.
Within the Western context, fish oil supplements have not been associated with an increased bleeding tendency, even in patients taking aspirin or warfarin for antithrombotic effect .
Concerns have been raised that fish oils, which contain highly unsaturated n3 PUFAs, lead to accumulation of lipid peroxides in vessels, which may increase cardiovascular risk. There is no convincing evidence that such a pathological accumulation is aggravated by fish oil. In any case, the overall effect of fish oil is to reduce rather than increase cardiovascular risk .
Methylmercury is an industrial contaminant that accumulates in long-lived fish (e.g. swordfish, marlin, sea perch, shark). Methylmercury is a neurotoxin that impairs neural development, especially in the foetus and infants. Fish consumption has been associated with increased blood and urine mercury [47, 48]. Properly processed fish oils contain very little mercury. Increased blood and urine mercury was not seen in a group of patients taking fish oil at 15 ml/day (4.5 g EPA plus DHA per day) for more than 3 years (unpublished data).
Chlorinated biphenyls (PCBs) are byproducts of industrial synthesis of organic chemicals. They are structurally related to dioxins and are potentially toxic. Industrial processes that produce PCBs have been outlawed because these compounds are poorly biodegradable and they have been found to accumulate in the land and marine food chains. Of continuing concern are polybrominated biphenyl (PBB) fire retardants, the production of which is still allowed. With regard to their poor biodegradability, accumulation and toxic potential, PBBs are similar to PCBs. Although the level of environmental contamination of PBBs is substantially less than that of PCB, their continued production means increasing accumulation, and alternatives are being sought. Halogenated biphenyls can be removed from fish oils by molecular distillation and should be present at low levels in good quality products .
Epidemiological studies show lower frequencies of RA in populations that consume higher amounts of LC n3 fats . However, these differences could be due to incidental unidentified environmental or genetic factors. With regard to the latter, the RA disease susceptibility epitope is not responsible because the Japanese and Inuits, who have high fish intakes and low prevalence of RA, both have relatively high frequencies of DR4 alleles, which confer disease susceptibility [51, 52].
Control of disease activity in systemic lupus erythematosus can be improved with fish oil, as shown in clinical studies and in murine lupus. In the latter, preventive regimens, begun at an age before the disease emerges, can have a strong preventive effect . Considering the safety of fish oil and the increased cardiovascular risk seen in lupus, fish oil seems a reasonable option for treatment of 'minimal lupus', which is defined as the presence of arthralgia and a strongly positive antinuclear antibody. (Fish oil might reasonably be combined with hydroxychloroquine in this setting, and has the advantage of freedom from serious unwanted effects.)
As discussed under clinical evidence for the anti-inflammatory effects of fish oil (see above), fish oil has been found to have therapeutic effects in several inflammatory diseases. Fish oil has been studied most intensively in RA, where there is level 1 evidence for symptomatic improvement . There is level 2 evidence for a strong preventive effect against relapse in Crohn's disease and against progression of renal failure in immunoglobulin A nephropathy [24, 25]. Some types of psoriasis (guttate, pustular) have been shown to improve with oral fish oil given in anti-inflammatory doses [53–55]. Fish oil improves control in systemic lupus erythematosus [26, 56]. In addition to this catalogue of disorders, which share an autoimmune-based inflammation in their pathogenesis, there is strong evidence for cardiovascular benefit with fish oil.
Dietary fish and fish oil have been shown to reduce cardiovascular risk in epidemiological studies and in secondary prevention trials after myocardial infarction. Perhaps the most potent effect of dietary LC n3 PUFAs is to stabilize the myocardial membrane, thereby reducing ventricular fibrillation and sudden death.
The antiarrhythmic effect of LC n3 PUFAs has been demonstrated in vitro in studies of cardiomyocytes challenged by various stimuli . Fish oil or purified n3 fatty acids reduced the incidence of arrhythmias in animal models of ischaemically induced ventricular fibrillation [58, 59]. These findings correlate with the striking reduction in cardiac mortality and, in particular, sudden cardiac death seen with fish oil and diets rich in n3 PUFAs from vegetable sources after myocardial infarction [46, 60, 61]. This effect on sudden death can be seen with under 1 g/day LC n3 PUFAs (i.e. less than the anti-inflammatory dose) .
At anti-inflammatory doses of fish oil other cardiovascular benefits can be seen. These include improved blood pressure control, reduced fasting triglycerides, more rapid clearance of chylomicrons, increased high-density lipoprotein cholesterol, reduced total cholesterol to high-density lipoprotein cholesterol ratio, reduced atheroma (in experimental animals), and improved arterial compliance and flow mediated dilation (for review, see Din and coworkers ).
Importantly, a meta-analysis of large, long-term randomized controlled trials of anti-lipidaemia agents  showed that strategies that increase LC n3 PUFA intake reduce annualized death rates to an extent as least as great as that with statins, which is the only other intervention to have significant benefit. That fish oil is not used more widely to manage cardiovascular risk appears to reflect more the influence of pharmaceutical product marketing on the practice of 'evidence-based medicine' than the merits of fish oil relative to those of commonly used proprietary agents.
In a medical environment in which messages molded by pharmaceutical interests stress the 'need' for NSAIDs, prescribers should consider the NSAID-sparing effects, the lack of serious side effects and the positive health benefits of fish oil. Importantly, recipients should be informed that there is a 'mainstream' evidence base for such a recommendation, thereby distinguishing dietary n3 fats from many other nonprescription items that are grouped loosely as 'complementary medicines'.
Although modest increases in intake of n3 LC PUFAs can reduce cardiovascular risk, relatively large doses (≥ 2.7 g/day EPA plus DHA) are required for anti-inflammatory effects. These doses can be taken efficiently and economically as liquid fish oil on juice. Recipients should be informed that there are multiple strategies for increasing n3 intake, and therefore, no matter what are their usual dietary preferences, there should be an acceptable approach for most individuals.
monounsaturated fatty acid
nonsteroidal anti-inflammatory drug
polyunsaturated fatty acid
tumour necrosis factor
All authors contributed equally to the literature search and content. The final manuscript has been read and approved by all authors.