All-Cause Mortality and Cardiovascular Death between Statins and Omega-3 Supplementation: A Meta-Analysis and Network Meta-Analysis from 55 Randomized Controlled Trials

Statins and omega-3 supplementation have shown potential benefits in preventing cardiovascular disease (CVD), but their comparative effects on mortality outcomes, in addition to primary and secondary prevention and mixed population, have not been investigated. This study aimed to examine the effect of statins and omega-3 supplementation and indirectly compare the effects of statin use and omega-3 fatty acids on all-cause mortality and CVD death. We included randomized controlled trials (RCTs) from meta-analyses published until December 2019. Pooled relative risks (RRs) and 95% confidence intervals (CIs) were calculated to indirectly compare the effect of statin use versus omega-3 supplementation in a frequentist network meta-analysis. In total, 55 RCTs were included in the final analysis. Compared with placebo, statins were significantly associated with a decreased the risk of all-cause mortality (RR = 0.90, 95% CI = 0.86–0.94) and CVD death (RR = 0.86, 95% CI = 0.80–0.92), while omega-3 supplementation showed a borderline effect on all-cause mortality (RR = 0.97, 95% CI = 0.94–1.01) but were significantly associated with a reduced risk of CVD death (RR = 0.92, 95% CI = 0.87–0.98) in the meta-analysis. The network meta-analysis found that all-cause mortality was significantly different between statin use and omega-3 supplementation for overall population (RR = 0.91, 95% CI = 0.85–0.98), but borderline for primary prevention and mixed population and nonsignificant for secondary prevention. Furthermore, there were borderline differences between statin use and omega-3 supplementation in CVD death in the total population (RR = 0.92, 95% CI = 0.82–1.04) and primary prevention (RR = 0.85, 95% CI = 0.68–1.05), but nonsignificant differences in secondary prevention (RR = 0.97, 95% CI = 0.66–1.43) and mixed population (RR = 0.92, 95% CI = 0.75–1.14). To summarize, statin use might be associated with a lower risk of all-cause mortality than omega-3 supplementation. Future direct comparisons between statin use and omega-3 supplementation are required to confirm the findings.


Study Eligibility
The inclusion criteria for studies to be evaluated in the final analysis were as follows: (i) statins or omega-3 supplementation were compared with or added to a placebo; (ii) the sample size and the number of all-cause mortality and CVD death events were reported; and (iii) the follow-up duration was at least one year. RCTs were excluded if the source of omega-3 was from dietary intake or the comparison arm contained omega-6 fatty acids.
Two investigators (T.H. and J.K.) independently assessed articles for the inclusion and exclusion criteria and were responsible for data extraction. Any discrepancies were discussed and resolved through consultations with other investigators (J.-M.K., S.Y.B, and J.-H.C.). Details were recorded about the authors' name; publication year; country; recruitment period; the mean or median follow-up time; body mass index (BMI); demographic information on age and sex; history of CVD, coronary heart disease, myocardial infarction, heart failure, hypertension, dyslipidemia, and diabetes; the mean or median number of smokers and obese subjects; sample size; the daily dose of the intervention; and the number and percentage of outcome events for each treatment arm.

Statistical Analyses
We conducted both direct and indirect comparisons for all-cause mortality and CVD death. In the direct comparison, we investigated the effects of statins or omega-3 supplementation compared with the placebo in a random-effects model using the DerSimonian-Laird method [16]. We additionally performed subgroup analyses by type of prevention, type of statin, and type of omega-3 supplementation. In particular, RCTs in which at least 80% of the study population had any CVD risk factors (hypertension, dyslipidemia, diabetes, smoking, or obesity) were considered to be conducted for primary prevention, whereas those in which at least 80% of the study population had any history of CVD events (CVD, coronary heart disease, myocardial infarction, and heart failure) were considered to be conducted for secondary prevention. The remaining RCTs were considered as mixed population. In the indirect comparison, pooled RRs and 95% CIs were calculated to examine the pairwise comparisons of statins versus placebo, omega-3 supplementation versus placebo, and statins versus omega-3 supplementation in an NMA using a frequentist approach [17].
Heterogeneity across studies was measured by calculating the Higgins I 2 [18]. Substantial heterogeneity was considered to be present if the I 2 value was greater than 50% or the p-value was less than 0.05. Evidence of asymmetry and publication bias was assessed using Begg funnel plots [19] and the Egger test [20], in line with recent recommendations [21]. Publication bias was considered to be present if the funnel plot was asymmetric or if the p-value from the Egger test was lower than 0.05. In this case, pooled estimates from the fixed-effects model would be reported along with those from the random-effects model to counterweight the possible inflation of the therapeutic effect among large and small individual RCTs. For dose-response meta-analysis, the correlated RR estimates across different doses of statins and omega-3 supplementation were calculated using the generalized least-square regression method.
All statistical analyses were performed using Stata SE version 14.0 (StataCorp, College Station, TX, USA).

Literature Search
We identified 1233 articles in the database search ( Figure 1). After irrelevant records were removed, 71 full texts were screened. Of these, 61 were discarded because they addressed irrelevant topics (n = 26), reported inappropriate outcomes (n = 16), or were duplicate or overlapping (n = 19). From the remaining 10 systematic reviews and meta-analyses, 374 RCTs were extracted and accessed to evaluate them for the eligibility criteria. After additionally hand-searching for updated RCTs (n = 3) and excluding ineligible RCTs (n = 322), 55 studies [5, with 36 RCTs of statins and 19 RCTs of omega-3 supplementation were included in the final meta-analysis and NMA.
In the secondary prevention population, 19,805 participants were assigned to receive a statin (n = 3996), omega-3 supplementation (n = 5898), or a placebo (n = 9911). The median age, percentage of male subjects, follow-up duration, and BMI across studies were 68 years old, 78.2%, 3.4 years, and 27 kg/m 2 , respectively. The following distribution was found for patients with a history of CVD events: coronary heart disease, 100%; myocardial infarction, 60%; and heart failure, 100%.
In the secondary prevention population, 19,805 participants were assigned to receive a statin (n = 3996), omega-3 supplementation (n = 5898), or a placebo (n = 9911). The median age, percentage of male subjects, follow-up duration, and BMI across studies were 68 years old, 78.2%, 3.4 years, and 27 kg/m 2 , respectively. The following distribution was found for patients with a history of CVD events: coronary heart disease, 100%; myocardial infarction, 60%; and heart failure, 100%.
In the subgroup analysis by type of statins, risk reduction effects were only observed for pravastatin in terms of all-cause mortality and atorvastatin, pravastatin, fluvastatin, and pitavastatin in terms of CVD death. In contrast, the subgroup analysis by type of omega-3 fatty acids showed similar findings to those of the overall intervention. The pooled estimates and level of heterogeneity did not change much when excluding the REDUCE-IT trial, in which icosapentyl ethyl was used at a high dose of 4 g.
Additional analyses by specific causes of death due to CVD found that statins were associated with lower risks of coronary heart disease death (RR = 0.87, 95% CI = 0.78-0.97) and fatal myocardial infarction (RR = 0.73, 95% CI = 0.57-0.93), but nonsignificant results for fatal stroke and heart failure death ( Figure A10). In contrast, omega-3 supplementation was not associated with any specific causes of CVD death ( Figure A11).
Dose-response meta-analysis for the effect of statins and omega-3 supplementation on all-cause mortality and CV death is summarized in Table A4 and Figures A12 and A13. There was a dose-response relationship between rosuvastatin and all-cause mortality, with a 9% decrement in all-cause mortality (per 10-mg RR = 0.91, 95% CI = 0.80-0.99). There was also dose-response relationship between atorvastatin, pravastatin, simvastatin, fluvastatin, and omega-3 supplementation and CVD death, with per 10-mg RRs (95% CI) of 0.

Network Meta-Analysis
An NMA combining direct and indirect estimates for pairwise comparisons among omega-3 supplementation, statins, and placebo is shown in Table 2. Similar to findings from the meta-analysis, omega-3 supplementation showed a lower risk of all-cause mortality than placebo, but not to a significant extent (RR = 0.97, 95% CI = 0.92-1.03), whereas randomization to omega-3 supplementation reduced CVD death by 9% compared to placebo (RR = 0.91, 95% CI = 0.84-0.99). Statins also demonstrated significant risk reductions for all-cause mortality and CVD death of 12% and 16%, with RRs (95% CIs) of 0.88 (0.84-0.93) and 0.84 (0.78-0.81), respectively. A significant risk reduction was observed for the effects of statin use versus omega-3 supplementation on all-cause mortality (RR = 0.91, 95% CI = 0.85-0.98) but borderline on CVD death (RR = 0.92, 95% CI = 0.82-1.04). Table 2. Relative risks and 95% confidence intervals from network meta-analysis of omega-3 supplementation, statins, and placebo effects on all-cause mortality and cardiovascular disease death.

Discussion
We performed a systematic review, meta-analysis, and NMA to summarize the current evidence on the effects of statin use and omega-3 supplementation on mortality outcomes. In this study, analyses of statins showed a statistically significant 10% reduction in the risk of all-cause mortality and a 14% reduction in the risk of CVD death. Additionally, the meta-analysis showed that omega-3 supplementation did not lead to a significantly lower risk of all-cause mortality, whereas the risk of CVD death was reduced significantly by 8%. In the comparison with omega-3 supplementation, statins were found to significantly reduce all-cause mortality in the total population, borderline in the subgroup of primary prevention and mixed population, but not in the subgroup of secondary prevention. Furthermore, the NMA showed that the effects of statins and omega-3 supplementation on CVD death were borderline in the total population and in the subgroup of primary prevention, but not in the subgroups of secondary prevention and mixed population.
Our findings for the effect of statins are consistent with the latest updated meta-analysis, in which 11% and 20% risk reductions for all-cause mortality (24 RCTs) and CVD death (15 RCTs) were observed with the intervention of statins [8]. More RCTs were included in the current meta-analysis, with a similar effect on all-cause mortality (29 RCTs, 10% versus 11% risk reduction) but a smaller effect on CVD death (26 RCTs, 14% versus 20% risk reduction), although no statistical test was performed to compare the difference. Statins can change plasma levels of total cholesterol (TC), triglyceride (TG), high-density lipoprotein (HDL-C), and low-density lipoprotein (LDL-C) by inhibiting the synthesis of cholesterol in the liver by 3-hydroxy-3-methylglutaryl-CoA (HMG-CoA) reductase [77][78][79][80][81][82][83], thereby reducing the risk of CVD. However, most of the individual RCTs did not detect a significantly beneficial effect of statins on mortality outcomes. Randomization to statins might have led to a decrease in both all-cause mortality and CVD death in the LIPID and 4S trials [48,84]. More additional RCTs observed a significant effect of statins on all-cause mortality than on CVD death [26,37,41,60], but the effect was closer to the null. Since more RCTs found a significant effect, substantial heterogeneity remained among RCTs investigating the association between statin use and all-cause mortality (I 2 = 43.8%, p = 0.006). Furthermore, recent studies found that statin use was associated with a lower risk of mortality in patients receiving clopidogrel (RR = 0.54, 95% CI = 0.40-0.74) [85] and Asian patients with type 2 diabetes [86]. Findings from another prospective cohort study of nearly 86,000 participants also supported the beneficial effect of statins on all-cause mortality (hazard ratio 0.86, 95% CI = 0.77-0.95) and CVD death (hazard ratio 0.75, 95% CI = 0.64-0.89) [87]. Moreover, our additional analyses by specific causes of death by CVD suggested the potential effect of statins on risk reductions of coronary heart disease death and fatal myocardial infarction.
The findings are in line with a recent meta-analysis by Abdelhamid et al., who reported that higher intake of omega-3 fatty acids had little or no effect on all-cause mortality and CVD death [13]. Apart from a different methodological approach, we included more up-to-date RCTs in this study, but excluded several RCTs with issues that might cause discrepancies. Several lipid biomarkers, including TC, TG, HDL-C, and LDL-C have been reported to be associated with CVD risk [88][89][90][91]. Omega-3 fatty acids can decrease plasma TG levels by reducing the production of hepatic very-low-density lipoprotein cholesterol and increasing chylomicron clearance. Furthermore, omega-3 fatty acids can change dysfunctional HDL-C to functional HDL-C and also exert weak TC-and LDL-C lowering effects [92][93][94]. In contrast with the findings of the pooled analysis, most of the individual RCTs did not find that omega-3 supplementation led to a significant reduction of mortality outcomes. The results of the GISSI-P trial stand apart from the null findings of 19 RCTs that examined the association between omega-3 supplementation and all-cause mortality, with an RR of 0.87 (95% CI = 0.77-0.97) (Figures A1 and A3) [40]. Therefore, the pooled effect size for the effect of omega-3 supplementation on all-cause mortality was non-significantly different, as expected. In terms of CVD death, the results from the REDUCE-IT [40], ASCEND [24], and GISSI-P [40] trials, which accounted for 31.4% of the sample size, were only borderline significant, with RRs and 95% CIs of 0.82 (0.67-0.99), 0.82 (0.67-0.99), and 0.84 (0.72-0.97), respectively ( Figures A2 and A4). However, the pooled analysis with a large sample showed that omega-3 supplementation had a significant effect on CVD death, with low heterogeneity (I 2 = 6.0%, p = 0.38). Furthermore, subgroup analyses by specific causes of death due to CVD showed no significant associations between omega-3 supplementation and the deaths of coronary heart disease, myocardial infarction, stroke, and heart failure.
In this study, we found a significantly lower risk of all-cause mortality in the statin group than the omega-3 supplementation group in primary prevention but not in secondary prevention. Because we defined studies of primary prevention as those with high probabilities of CVD risk factors, statin use was therefore hypothesized to reduce the death of CVD risk factor-related diseases. Moreover, statins were observed for their beneficial effects on reducing the mortality from breast, colorectal, kidney, ovarian, and prostate cancers, which contributed to all-cause mortality [95]. However, those effects of omega-3 supplementation have not been adequately investigated. In contrast, the effect of statins and omega-3 supplementation on CVD death reduction was borderline only. The mechanism is unclear, but a possible explanation may be found in the overlap of the pleiotropic effects of statins with the actions of omega-3 supplementation with, including endothelial function improvement, anti-thrombotic effects, and antioxidant effects [96]. Although previous RCTs and meta-analysis examined the effect of combination therapy with statins and omega-3 supplementation versus statins alone in patients with dyslipidemia [97][98][99] or cardiovascular events [92], the effects of statins and omega-3 supplementation have not been investigated. Our methodology was designed to minimize variety in the placebo group, which is the mediating factor in indirect comparisons between statins and omega-3 supplementation. We excluded RCTs if the placebo group contained omega-6 or even low doses of omega-3 to avoid bias in our pooled estimates. Furthermore, a subgroup analysis by primary and secondary prevention was performed to obtain robust findings.
Despite its strengths, there are certain limitations of our study. First, considering the nature of the study, potential heterogeneity in the associations with CVD death remained. Second, due to the lack of data, we were unable to perform the sex-specific meta-analysis to test whether there were different effects in males and females. Lastly, although both EPA alone [75] and the highly purified and stable EPA ethyl ester [28] have been determined to have beneficial effects, a pooled analysis with the other form of EPA + DHA might have introduced heterogeneity due to diversity. Combinations of different statin types might have also led to heterogeneity. Nevertheless, this is an up-to-date study with 278,954 participants that compared the effects of statin use and omega-3 supplementation on mortality outcomes.

Conclusions
In summary, statin use was significantly associated with decreased risks of mortality outcomes, whereas omega-3 supplementation showed nonsignificant or little effect on all-cause mortality and CVD death. Statin use was shown to be more effective in reducing all-cause mortality than omega-3 supplementation. However, the effect was borderline in terms of CVD mortality. Future direct comparisons between omega-3 supplementation and statin use are required to detect the statistical benefits of omega-3 supplementation.

Conflicts of Interest:
The authors declare no conflict of interest.                       . Figure A11. Subgroup meta-analysis for the effect of omega-3 supplementation on specific causes of cardiovascular disease death. Full black quadrilateral represents the point estimate of individual studies and blank with quadrilateral represents pooled estimates. RR, relative risk; CI, confidence interval. Figure A11. Subgroup meta-analysis for the effect of omega-3 supplementation on specific causes of cardiovascular disease death. Full black quadrilateral represents the point estimate of individual studies and blank with quadrilateral represents pooled estimates. RR, relative risk; CI, confidence interval. Pooled estimates for the overall study population and subgroups are highlighted with bold font.