Fish oil and cardiovascular mortality

Omega 3 fatty acids, mainly eicosapentaenoic and docosahexaenoic acids, seem to lower cardiac death by 8%. This meta-analysis includes 14 randomized clinical trials with a total of 72,000 subjects. 

Further benefits (13-29% death reduction) were seen in adults utilizing higher doses of omega 3 (> 1 gram per day) and in those with higher baseline cardiovascular risk; elevated triglycerides (>150 mg/dL), LDL-cholesterol (>130 mg/dL) and non-statin users.

Important to note that these results are to some degree in accordance with the recent Science Advisory from the American Heart Association published in March 2017, suggesting the use of omega three for secondary prevention of coronary heart disease, heart failure and sudden cardiac death.



J Clinical Lipidology


October 2017

Background: Randomized controlled trials (RCTs) assessing use of long-chain omega-3 polyunsaturated fatty acids (LC-OM3), primarily eicosapentaenoic acid (EPA), and/or docosahexaenoic acid (DHA) have shown mixed results.

Objective: The objectives of the study were to update and further explore the available RCT data regarding LC-OM3 supplementation and risk for cardiac death and to propose testable hypotheses for the mixed results obtained in RCTs regarding supplemental LC-OM3 use and cardiac risk.

Methods: A literature search was conducted using PubMed and Ovid/MEDLINE for RCTs assessing LC-OM3 supplements or pharmaceuticals with intervention periods of at least 6 months and reporting on the outcome of cardiac death. Meta-analysis was used to compare cumulative frequencies of cardiac death events between the LC-OM3 and control groups, including sensitivity and subset analyses.

Results: Fourteen RCTs were identified for the primary analysis (71,899 subjects).

In the LC-OM3 arms, 1613 cardiac deaths were recorded (4.48% of subjects), compared with 1746 cardiac deaths in the control groups (4.87% of subjects). The pooled relative risk estimate showed an 8.0% lower risk in the LC-OM3 arms vs controls (P = .015).

Subset analyses showed numerically larger effects (12.9%–29.1% lower risks, all P < .05) in subsets of RCTs with:

Eicosapentaenoic acid + docosahexaenoic acid dosages >1 g/d

Higher risk samples, secondary prevention:

Baseline mean or median TGs ≥150 mg/dL,

LDL cholesterol ≥130 mg/dL,

Statin use <40% of subjects).

Heterogeneity was low (I2 ≤ 15.5%, P > .05) for the primary and subset analyses.


LC-OM3 supplementation is associated with a modest reduction in cardiac death.


More from the publication:

In recent years, several randomized controlled trials (RCTs) have assessed the effects of supplemental long-chain omega-3 polyunsaturated fatty acids (LC-OM3), primarily eicosapentaenoic acid (EPA) and docosahexaenoic acid (DHA), on risk for various types of cardiovascular disease (CVD) events. Results from these studies have been mixed, with some suggesting benefits and others showing neutral effects.

The outcome for which both the observational evidence on LC-OM3 intake or status and RCTs of supplementation appears to show the most consistent association is cardiac death. For example, Rizos et al. conducted a meta-analysis of the effects of supplemental LC-OM3 on the outcomes of all-cause mortality, cardiac death, sudden death, myocardial infarction (MI), and stroke. The only outcome for which the 95% confidence interval (CI) did not cross the null value was cardiac death, with a pooled relative risk (RR) estimate from 13 RCTs of 0.91 (95% CI 0.85, 0.98).

This finding is in agreement with that from a pooled analysis of data from 19 observational cohort studies that evaluated LC-OM3 status based on various biomarkers, including omega-3 fatty acids in total plasma, phospholipids, cholesterol esters and adipose tissue, and coronary heart disease (CHD) risk. Assessment of LC-OM3 status using these biomarkers avoids many of the issues associated with estimation of dietary intake.

The overall pooled RR estimates for fatal CHD per 1 standard deviation increase in EPA and DHA biomarker level were 0.91 (95% CI 0.82-1.00) and 0.90 (95% CI 0.84-0.96), respectively. Results were more consistent for fatal CHD (cardiac death) than for total CHD or non-fatal MI.

The purpose of this meta-analysis and review of research gaps is 2-fold. The first aim is to update and further explore the available RCT data regarding LC-OM3 supplementation and risk for cardiac death. Secondary aims are to briefly review the evidence regarding the effects of LC-OM3 intake on cardiac event risk and propose testable hypotheses for the mixed results obtained in RCTs regarding supplemental LC-OM3 use and cardiac event risk.

The results from this meta-analysis are supportive of the recent Science Advisory from the American Heart Association that concluded LC-OM3 “treatment is reasonable” for

(1) secondary prevention of CHD and sudden cardiac death among patients with prevalent CHD; and

(2) secondary prevention of adverse outcomes in patients with heart failure.

In the present analysis, the pooled RR from 14 RCTs was 0.920 (95% CI 0.861, 0.984; P = .015) with low heterogeneity (I2 = 1.8%) in the primary random effects model. A secondary analysis that included an additional 6 RCTs showed results that were not materially different: pooled RR 0.929 (95% CI 0.876, 0.984; P = .013), with somewhat greater heterogeneity (I2 = 35.3%, P = .061).

One notable feature of LC-OM3 supplementation is the low risk associated with its use. Because of the low risk for adverse effects, even a modest benefit is clinically meaningful.

Larger effects (all P < .05) were observed in subsets of the primary analysis set for RCTs that used:

>1 g/d of EPA + DHA (RR = 0.71),

Had mean or median TG ≥ 150 mg/dL (RR = 0.83) or

LDL-C ≥130 mg/dL (RR = 0.83),

Studied a secondary prevention sample (RR = 0.87), and in which

Baseline statin use was <40% (RR = 0.87).

These results are consistent with the hypothesis that supplemental LC-OM3 may be most efficacious for reducing cardiac death in higher risk individuals.

The present findings are concordant with those of Alexander et al. who found that coronary death was significantly lower in a pooled analysis of 4 secondary prevention RCTs of LC-OM3 (RR = 0.80, 95% CI 0.64, 0.99). The present results do not agree with the conclusions from the 2016 AHRQ systematic review. However, in that report, only the largest studies were evaluated, and exclusion of smaller studies may have materially affected the findings. 

Alexander et al. also analyzed results from prospective cohort studies and reported significantly lower risks for coronary death (9 studies, RR = 0.82, 95% CI 0.69, 0.98) and sudden cardiac death (5 studies, RR = 0.53, 95% CI 0.41, 0.67). In general, results from observational cohort studies have suggested that higher LC-OM3 intake is associated with lower risks for fatal and non-fatal CHD events. For example, Alexander et al. showed pooled RR estimates of 0.77 and 0.81 (both P < .05), respectively, for these outcomes. However, their meta-analysis of RCTs, as well as those completed by Rizos et al. and the AHRQ33 showed small, non-significant reductions in risk for non-fatal CHD events and/or MI.

Thus, for cardiac death, the aggregate RCT and observational evidence has been fairly consistent in showing benefits of higher LC-OM3 intake, particularly in higher risk subsets.

However, the prospective cohort and RCT results have been discordant for the effects of LC-OM3 intake on non-fatal CHD events. These observations suggest several testable hypotheses that warrant exploration.

A possible explanation for the mixed results in RCTs regarding both fatal and non-fatal CHD events relates to dosage. In the present meta-analysis, the reduction in risk for cardiac death was numerically larger in studies that used an EPA + DHA dosage >1 g/d (RR = 0.709) compared with that in the overall primary study sample (RR = 0.920 for both the random and fixed effects models). One large RCT completed in Japan, the JELIS trial, showed a statistically significant 19% reduction in major coronary event risk with an intervention of 1.8 g/d of EPA ethyl esters, but no significant reduction in cardiac death. A post hoc analysis of the relationship of blood EPA level with the treatment effect for the primary composite outcome showed that the benefit was mainly attributable to the subset of those in the active treatment group who had plasma EPA levels ≥150 mcg/L (approximately the top quartile).

Dietary intake of LC-OM3 in Japan is higher than that in the United States and most European countries and the subset of JELIS participants with the highest plasma EPA levels may have been those who consumed the EPA ethyl esters and also had a high background intake of foods containing LC-OM3.

Mozaffarian et al. (2013) found a dose–response relationship between level of plasma phospholipid total LC-OM3 quintile and CHD mortality in the Cardiovascular Health Study (P for trend = .002). Recently, the Omega-3 Acid Ethyl Esters on Left Ventricular Remodeling After Acute Myocardial Infarction investigators showed that 4 g/d of LC-OM3 ethyl esters, compared with a corn oil placebo, significantly improved measures of cardiac structure (fibrosis) and function (left ventricular end systolic volume index) when administered for 6 months after a MI. The magnitudes of the benefit for left ventricular end systolic index rose with increasing quartiles of erythrocyte omega-3 level (P for trend < .0001).

Taken together, the findings from those studies, as well as those of the current meta-analysis and that of Alexander et al. (2017), are consistent with the hypothesis that higher intakes of LC-OM3 (>1 g/d of EPA + DHA) may be required to observe some of the physiological effects that could contribute to lower risks for both fatal and non-fatal CHD events.

Only 28% of the subjects included in the primary meta-analysis dataset participated in RCTs in which the dosage was >1 g/d of EPA + DHA. Furthermore, most of the RCTs assessing cardiovascular outcomes provided no data on baseline and on-treatment biomarkers of LC-OM3 status; thus, it is not possible to evaluate relationships between baseline, or on-treatment LC-OM3 status (reflecting tissue levels), and event risk. Accordingly, additional research is needed to assess larger dosages of LC-OM3 on risks for cardiac death and non-fatal cardiac events, with inclusion of biomarker analyses (eg, total plasma, erythrocyte, phospholipid, or cholesterol ester LC-OM3 levels) to define both baseline status and change in status during treatment.

There are several pathways through which higher LC-OM3 intake could potentially alter risk for fatal and non-fatal cardiac events, including anti-arrhythmic effects (particularly arrhythmias triggered by myocardial ischemia), effects related to cardiac structure and function (eg, fibrosis, myocardial oxygen demand), endothelial and autonomic function (eg, vascular resistance and heart rate), thrombosis, blood pressure, inflammation/resolution, and lipoprotein metabolism.

The mechanisms through which LC-OM3 intake might alter CVD risk have been reviewed elsewhere in detail. The dose–response characteristics for most of these effects have not been fully described. Also, the RCTs completed to date have generally not been designed to test specific mechanistic hypotheses. It is possible that dosages >1 g/d of EPA + DHA are required to produce clinically relevant changes in some of the pathways, such as effects on inflammation and thrombosis. Conversely, some of the benefits associated with higher LC-OM3 intake may be produced mainly in those with low LC-OM3 intake or status, with little or no benefit observed in those with higher baseline intake or status. Such a ceiling effect has been proposed by Mozaffarian et al. regarding effects of LC-OM3 intake on heart rate and susceptibility to ventricular arrhythmia.

One well-documented effect of LC-OM3 is to lower the circulating TG concentration, although dosages >1.5 g/d of EPA + DHA are generally required to produce a clinically meaningful change. A recently published meta-analysis of the effects of TG-lowering drug therapies (fibrates, niacin, EPA ethyl esters) on cardiac or CVD event risk showed that the overall benefit in 10 studies was relatively small (12% reduction), but was larger (18% reduction) in subsets with elevated TGs (≥150 mg/dL), especially if accompanied by low high-density lipoprotein cholesterol (HDL-C; 29% reduction). Two ongoing RCTs are investigating the effects of higher dosage LC-OM3 pharmaceutical products (4 g/d LC-OM3, primarily EPA + DHA ethyl esters or carboxylic acids) on major adverse cardiovascular event rates in high-risk subjects with elevated TGs.

The results of the present meta-analysis provide some reason for optimism because the studies in which subjects had higher baseline TG or LDL-C concentrations showed numerically larger benefits on the incidence of cardiac death than those of the overall primary analysis. In addition to lowering TGs, higher dosages of LC-OM3 have been shown to increase HDL and LDL particle sizes although EPA and DHA show some differences regarding their effects on lipoprotein lipid levels:

  1. EPA and DHA each lower TGs.
  2. DHA tends to raise both LDL-C and HDL-C,
  3. whereas EPA does neither.

The clinical relevance of the above differences in effects on lipoprotein lipids is uncertain at the present.

In the current meta-analysis, the studies in which statin use was low (<40% of participants at baseline) showed a numerically larger benefit than that in the primary analysis. More recently conducted RCTs have had higher prevalence values for use of statins, as well as other cardioprotective agents such as aspirin and other anti-platelet drugs, beta-adrenergic antagonists, and renin-angiotensin-aldosterone system antagonists, reflecting changes in standards of care. Because of overlap in the mechanisms of action of some of these agents and pathways affected by LC-OM3, it may be difficult to demonstrate a benefit of LC-OM3 supplementation, especially at a low dosage, when it is added to other therapies with established cardioprotective properties.

The results from both the present meta-analysis, and that of Alexander et al., suggest that higher risk groups may be more likely to experience reductions in risk for cardiac death with LC-OM3 supplementation, particularly when provided at a higher dosage. VITAL, the Vitamin D and Omega-3 Trial, has 25,874 men and women enrolled and is assessing the effects of 1 g/d LC-OM3 ethyl esters, as well as supplemental vitamin D (2000 IU/d), compared with placebo, using a 2 × 2 factorial design. ASCEND, A Study of Cardiovascular Events iN Diabetes, is another trial assessing the effects of 1 g/d LC-OM3 ethyl esters, with or without aspirin, compared with placebo, using a 2 × 2 factorial design; the trial has randomized 15,480 men and women with diabetes.

Based on the current evidence base, there is reason for concern that VITAL and ASCEND may not show benefits because the LC-OM3 dosage is relatively low and subjects in VITAL are at low average risk.

A limitation of the results from the present meta-analysis is that several of the studies included were small or had suboptimal trial designs. For example, 2 of the largest trials, GISSI-Prevenzione and JELIS, were not placebo controlled. Although this does raise the possibility of bias in ascertainment of event status, this is likely to be less of a concern with the outcome of cardiac death than with non-fatal events. Also, sensitivity analyses and results from observational studies are generally aligned for cardiac death and suggest lower risk associated with higher LC-OM3 intake.

The present meta-analysis of RCTs showed a modest, but statistically significant, benefit of LC-OM3 supplementation on risk for cardiac death (8.0% in the primary analysis). This finding supports the recent Science Advisory from the American Heart Association that concluded LC-OM3 “treatment is reasonable” for

(1) secondary prevention of CHD and sudden cardiac death among patients with prevalent CHD; and

(2) secondary prevention of adverse outcomes in patients with heart failure.

Because of the low risk for adverse effects with LC-OM3 supplementation, even a modest benefit is clinically meaningful.

Subgroup analyses show numerically larger benefits (12.9%–29.1%, all P < .05) in studies that used >1 g/d of EPA + DHA, and in higher risk groups, including those with greater mean or median levels of TGs (≥150 mg/dL) or LDL-C (≥130 mg/dL), secondary prevention study samples, and studies with lower baseline use of statins (which is also a proxy for use of other cardioprotective agents).

These results suggest that additional research is warranted to further evaluate the potential risk reduction with LC-OM3 supplementation at higher dosages and in higher risk samples. Future RCTs should include evaluation of biomarkers of omega-3 status at baseline and during treatment and should be designed to test specific hypotheses about the mechanisms through which benefits might be produced. Four RCTs evaluating CVD event risk with LC-OM3 interventions are ongoing and should provide helpful additional information to guide clinical use of LC-OM3 supplementation or drug therapy.