OUP user menu

Fish intake and acute coronary syndrome

Lone J. Bjerregaard, Albert M. Joensen, Claus Dethlefsen, Majken K. Jensen, Søren P. Johnsen, Anne Tjønneland, Lars H. Rasmussen, Kim Overvad, Erik B. Schmidt
DOI: http://dx.doi.org/10.1093/eurheartj/ehp375 29-34 First published online: 14 September 2009


Aims To study the effect of fish consumption on the risk of acute coronary syndrome (ACS) in healthy subjects.

Methods and results This Danish follow-up study included 57 053 men and women between 50 and 64 years. Intake of lean and fatty fish was estimated from a detailed and validated food frequency questionnaire. Potential cases of ACS were identified through nationwide medical databases. A total of 1122 cases of ACS were verified during a mean follow-up period of 7.6 years. Among men, intake of fatty fish was associated with a lower risk of ACS. For men in the highest quintile of fish intake compared with the lowest quintile, the hazard ratio was 0.67 (95% confidence interval: 0.53–0.85). The inverse association was observed for intakes >6 g of fatty fish per day with no obvious additional benefit observed for higher intakes. Intake of lean fish was not associated with ACS. There were few cases of ACS and results were not consistent in women.

Conclusion In conclusion, a modest intake of fatty fish was associated with a lower risk of ACS in middle-aged men, whereas no consistent associations were observed among women.

  • Acute coronary syndrome
  • Coronary heart disease
  • Fish
  • n−3 PUFA
  • Nutrition
  • Prospective study
  • Follow-up study
See page 15 for the editorial comment on this article (doi:10.1093/eurheartj/ehp478)


A very low incidence of coronary heart disease (CHD) was reported in Greenland Eskimos more than 35 years ago which was believed to be due to their extremely high intake of seafood rich in n−3 polyunsaturated fatty acids (PUFA).1,2 Since then, data from other populations with a very high intake of seafood such as Alaskan Natives3 and Japanese fishermen4 have supported the hypothesis that marine n−3 PUFA may protect against atherosclerosis, thrombosis, and CHD.5

Studies in Western populations with a low or moderate intake of fish have been associated with lower coronary mortality,610 while the effect of seafood on non-fatal myocardial infarction (MI) and other manifestations of CHD has been less consistent.11

Few studies have examined the effect of fish consumption on acute coronary syndrome (ACS) which includes patients with unstable angina pectoris, ST-elevation MI, non-ST-elevation MI, and fatal MI.12 Recently, however, significantly lower blood levels of n−3 PUFA were found in patients admitted with ACS compared with controls,13,14 suggesting a possible protective effect of n−3 PUFA against ACS.

We have examined the association between lean and fatty fish intake and the risk of ACS in a large prospective study in Denmark. Information on fish consumption was obtained using a detailed and carefully validated food frequency questionnaire allowing separate effects of fish high and low in n−3 PUFA to be investigated.


Study cohort

The Diet, Cancer and Health cohort was established by the Danish Cancer Society in 1993–97 when 57 053 participants enrolled into a prospective study. Patients were, for this study, followed until the end of 2003, with a mean follow-up period of 7.6 years. The study participants were aged between 50 and 64 years, born in Denmark, and without a cancer diagnosis registered in the Danish Cancer Registry at baseline. The study was conducted in accordance with the Helsinki Declaration II and approved by the regional Ethics Committees. The study design has been reported in detail elsewhere.15

Dietary assessment

Information on the dietary habits during the previous year was obtained from a detailed, semi-quantitative 192-item food frequency questionnaire. Participants were asked to estimate their daily intake of foods that come in natural units, such as slices of bread, pieces of fruit, and cups and glasses of different beverages. For other foods, such as mixed dishes, a sex-specific portion size was calculated based on the results from a calibration study. By multiplying the frequencies of intake by the portion size, an individual average intake in grams/day of all foods and nutrients was calculated. A detailed description of the questionnaire and its validation has been published previously.16,17

A total of 24 foods and recipes in the food frequency questionnaire covered intake of fresh and processed fish specified according to preparation method. The different species of fish were categorized as either lean or fatty fish according to their content of n−3 PUFA below or above 1 g/100 g based on data from the Danish Food Composition Databank (Table 1).

View this table:
Table 1

Fish in questionnaire

Lean fish <1 g n−3 per 100 gIntake in % of total fishFatty fish >1 g n−3 per 100 gIntake in % of total fish
Cod/pollack15Caviar (Danish)<1
Cod roe9


Participants completed a lifestyle questionnaire regarding their health status, lifestyle habits, and social factors. Non-fasting serum total cholesterol and blood pressure were measured at baseline along with anthropometrical measurements.15

Acute coronary syndrome

The Danish National Patient Registry covers all hospital discharge diagnoses in Denmark since 1977 and all discharge diagnoses from outpatient clinics since 1995. Identification of cases with ACS has been described in detail elsewhere.18 Briefly, we identified cases of ACS in the cohort who were registered with an incident diagnosis of ACS (ICD-8: 410.00–410.99 and 427.27; ICD-10: I20.0, I21.0–I21.9, and I46.0–I46.9) during the study period. All medical records were reviewed by one of two physicians (A.M.J. and S.P.J.) to ensure that each case fulfilled the criteria of ACS used by the American Heart Association and the European Society of Cardiology.12 Since the first symptom of ACS may be fatal MI, we also included participants who were coded in the Causes of Death Registry as dying from ACS. Regarding sudden deaths outside hospitals, only cases with witnessed symptoms suggestive of coronary origin, a history of CHD, or indicative autopsy findings were included, whereas subjects who died suddenly without these characteristics were not included.

Statistical analyses

The association between fish intake and the risk of ACS was assessed using Cox' proportional hazard models with age as the time axis. Time in study was included as time-varying binary covariate, allowing the hazard ratio (HR) to change after 1 year in the study. For model diagnostics, we used a log-rank test in addition to graphical inspection of smoothed scaled Schoenfeld residuals. Analyses were performed separately for men and women.

In separate crude models, total intake of fish, fatty fish, and lean fish was included as categorical variables where the intake was split into quintiles according to the distribution of intake among cases. Total ACS and fatal MI were considered as outcomes. In adjusted models, we included baseline values of established risk factors for ACS: school education (short: ≤7 years, medium: 8–10 years, and long: ≥11 years); smoking [never, former, and current (1–14, 15–25, and >25 g of tobacco per day), and duration of smoking]; alcohol intake (0, 0.1–4.9, 5.0–9.9, 10.0–19.9, 20.0–39.9, and >40.0 g/day); body mass index (<25, 25–30, and ≥30 kg/m2); history of diabetes mellitus (yes, no); systolic blood pressure; serum cholesterol and physical activity (<30 and ≥30 min/day); dietary intake of fruits (g/day) and vegetables (g/day); total energy intake (MJ/day); and dietary intake of saturated fat (g/day), monounsaturated fat (g/day), and polyunsaturated fat (g/day). For women, we also adjusted for menopausal status (pre- or post-menopausal). In supplementary analyses, we stratified by the intake of fish oil capsules (yes, no). We used Stata version 9.2 (Stata Corporation, College Station, TX, USA) for statistical analyses.


A total of 160 725 potential study participants were invited and the invitation was accepted by 27 178 men and 29 875 women. Participants who did not appear in the Cancer Registry but were identified with a cancer diagnosis in The Danish National Patient Registry before inclusion in the study (n = 606) were excluded. We also excluded 1013 participants with a diagnosis of ACS prior to enrolment into the study. Furthermore, 103 participants were excluded due to missing medical records together with 20 patients where a medical-related procedure may have caused ACS. Finally, we excluded 1085 participants for whom information on one or more covariates used in adjusted analyses was missing. This left a total of 54 226 participants (25 573 men and 28 653 women) eligible for analysis. During 7.6 years of follow-up, a total of 1122 cases (854 men and 268 women) were verified.

Table 2 shows the sex-specific baseline characteristics of the cohort. The median intake of fish was 35 g/day among women and 42 g/day among men. The prevalence of several established risk factors for cardiovascular disease (age, high cholesterol levels, smoking, and overweight) was higher among cases compared with the cohort as a whole.

View this table:
Table 2

Baseline characteristics

Men cohortWomen cohort
Number25 57328 653
Age (years)56 (53; 60)56 (53; 60)
Body mass index (kg/m2)26.1 (24; 28)24.8 (23; 28)
Total cholesterol (mmol/L)5.9 (5.2; 6.7)6.2 (5.4; 7.0)
Smoking status
 % (n) non-smoker26 (6713)44 (12 631)
 % (n) former smoker34 (8754)23 (6671)
 % (n) current smoker <15 g/day11 (2696)15 (4365)
 % (n) current smoker 15–25 g/day18 (4475)15 (4255)
 % (n) current smoker >25 g/day11 (2935)3 (731)
Level of school education
 % (n) ≤7 years34 (8754)31 (8920)
 % (n) 8–10 years42 (10 666)50 (14 404)
 % (n) ≥11 years24 (6153)19 (5329)
Dietary variables
 Daily total energy intake (MJ)11 (9; 13)9 (7; 10)
 Alcohol intake g/day20 (11; 40)9 (3; 17)
 Daily total energy intake (MJ) excl alcohol10 (8; 12)8 (7; 10)
 Pct. energy: carbohydrate46 (42; 49)48 (45; 52)
 Pct. energy: protein18 (16; 20)18 (16; 19)
 Pct. energy: fat37 (33; 40)34 (30; 37)
 Intake of SFA (g/day)36 (29; 45)28 (22; 35)
 Intake of MUFA (g/day)32 (26; 40)23 (19; 29)
 Intake of PUFA (g/day)15 (12; 19)12 (9; 15)
 Intake of fruitsa and vegetablesb (g/day)310 (205; 443)392 (267; 554)
Total intake of fish (g/day)42 (28; 60)35 (24; 51)
Fish according to fat content (g/day)
 Fatty fish ≥1 g n−3 PUFA15 (9; 25)12 (7; 21)
 Lean fish <1 g n−3 PUFA24 (16; 36)21 (14; 30)
 Fish oil supplements, % (n)16 (4038)17 (4937)
  • Values are medians (25th–75th percentiles) unless otherwise stated. SFA, saturated fatty acids; MUFA, monounsaturated fatty acids; PUFA, polyunsaturated fatty acids.

  • aExcl. fruit juices.

  • bExcl. potatoes and legumes.

The sex-specific HRs are given for crude and adjusted analyses of fish intake (Table 3). For total fish intake, no consistent associations were found, whereas we identified a protective association for fatty fish consumption in men. For lean fish, there was no effect in men, but the results may suggest a tendency for a protective association in women. Results from adjusted analyses did not differ substantially from crude analyses and allowing the HR to change after 1 year of follow-up indicated no effect modification by time. When we considered fatal MI (n = 175) as an outcome, the associations were similar to those of intake of fish (total, fatty, and lean fish) and ACS, but the confidence intervals (CIs) were wider and no association was statistically significant (data not shown). Stratification according to intake of fish oil capsules did not significantly change the risk estimates (data not shown).

View this table:
Table 3

Fish intake and acute coronary syndrome

Men (854 cases)Women (268 cases)
HR (95% CI)HR (95% CI)HR (95% CI)HR (95% CI)
Total fish intake in quintilesTotal fish intake in quintiles
 Q1 (0–24 g/day)11 Q1 (0–22 g/day)11
 Q2 (25–35 g/day)0.92 (0.74; 1.14)0.96 (0.77; 1.19) Q2 (23–31 g/day)0.98 (0.67; 1.43)1.06 (0.72; 1.56)
 Q3 (36–47 g/day)0.90 (0.73; 1.11)0.98 (0.78; 1.22) Q3 (32–41 g/day)1.03 (0.71; 1.51)1.16 (0.78; 1.73)
 Q4 (48–64 g/day)0.89 (0.72; 1.10)0.98 (0.78; 1.23) Q4 (42–54 g/day)1.04 (0.71; 1.51)1.14 (0.76; 1.71)
 Q5 (>64 g/day)0.82 (0.66; 1.02)0.87 (0.69; 1.10) Q5 (>55 g/day)0.82 (0.56; 1.20)0.85 (0.55; 1.32)
Fatty fish intake in quintilesFatty fish intake in quintiles
 Q1 (0–6 g/day)11 Q1 (0–5 g/day)11
 Q2 (7–11 g/day)0.72 (0.58; 0.89)0.74 (0.60; 0.92) Q2 (6–10 g/day)0.78 (0.53; 1.13)0.83 (0.56; 1.22)
 Q3 (12–17 g/day)0.64 (0.52; 0.79)0.67 (0.54; 0.83) Q3 (11–15 g/day)0.91 (0.62; 1.32)1.03 (0.70; 1.53)
 Q4 (18–27 g/day)0.64 (0.52; 0.79)0.69 (0.55; 0.87) Q4 (16–23 g/day)0.89 (0.61; 1.30)1.02 (0.68; 1.52)
 Q5 (>27 g/day)0.63 (0.51; 0.78)0.67 (0.53; 0.85) Q5 (>23 g/day)0.71 (0.48; 1.04)0.78 (0.51; 1.19)
Lean fish intake in quintilesLean fish intake in quintiles
 Q1 (0–14 g/day)11 Q1 (0–12 g/day)11
 Q2 (15–21 g/day)1.01 (0.81; 1.24)1.06 (0.86; 1.32) Q2 (13–19 g/day)0.70 (0.48; 1.01)0.76 (0.52; 1.12)
 Q3 (22–28 g/day)0.97 (0.78; 1.20)1.02 (0.82;1.26) Q3 (20–25 g/day)0.83 (0.57; 1.21)0.90 (0.61; 1.34)
 Q4 (29–39 g/day)1.00 (0.81; 1.24)1.07 (0.86; 1.34) Q4 (26–33 g/day)0.95 (0.65; 1.38)1.03 (0.69; 1.54)
 Q5 (>39 g/day)0.98 (0.80; 1.21)1.02 (0.81; 1.28) Q5 (>33 g/day)0.75 (0.51; 1.09)0.78 (0.51; 1.20)
  • Incidence rate ratios and 95% CIs.

  • aAdjusted for education, smoking, alcohol intake, body mass index, history of diabetes mellitus, systolic blood pressure, serum cholesterol, physical activity, dietary intake of fruits and vegetables, total energy intake, dietary intake of saturated fat, monounsaturated fat, and polyunsaturated fat. Analyses in women were also adjusted for menopausal status.


In this prospective cohort study, intake of fatty fish was associated with a significant 30% lower risk of ACS in men during a mean follow-up period of 7.6 years when comparing subjects in the lowest quintile (≤6 g fatty fish/day) with those in the higher quintiles of intake. No clear dose–response relationship was, however, found, and a higher intake of fatty fish was not associated with a further lowering of the risk of ACS. Among men, no association was observed for lean fish intake. For women, a tendency for a lower risk of ACS was observed in the top quintile of intake of both fatty and lean fish. However, the number of female cases was low (n = 268) and the CIs were wide.

The strengths of this study include the follow-up study design, the detailed and validated food frequency questionnaire, the large number of cases, and the strict validation of the diagnosis ACS. The food frequency questionnaire used in our study was self-administered and has been validated and tested against two 7-day weighed diet records.16,17 Furthermore, a biomarker substudy from this cohort comparing fatty acid profiles in adipose tissue biopsies with dietary intake of fatty acids showed significant correlations between the content of marine n−3 PUFA in adipose tissue and that calculated from the food frequency questionnaire.19

Some limitations also need to be considered. The food frequency questionnaire data representing the average diet the previous year are determined with some uncertainty. All data were only assessed at baseline and the participants' diets and risk factors may have changed during the study. There were few cases of fatal MI and few cases in women which make interpretation of results in these two groups difficult. Also, a diagnosis of ACS outside hospital is dubious, but accounted for <10% (111/1122) of the cases in the present study. Moreover, it was not possible to define whether the cases of fatal MI also fulfilled the criteria of sudden cardiac death (death within 1 h from onset of cardiac symptoms). Therefore, the study provides little information about a possible effect of fish consumption on arrhythmias and sudden cardiac death. Finally, individuals with a higher fish intake generally exercise more, smoke less, and are less likely to be overweight.20 Adjustment for lifestyle including diet did, however, not substantially change the associations, and residual confounding is therefore not a likely explanation for the observed associations.

The effect of fish consumption on CHD has been studied in other large epidemiological studies. Thus, in the Nurses’ Health Study21 comprising female nurses followed for 16 years, women in the highest quintile of fish intake (≥5 times a week) had a 21% lower risk of death from CHD compared with those in the lowest quintile (fish intake less than once a month). We have no explanation for the difference in findings between this very large study and our study. Most other trials in women5,6,11 have included relatively few cases and the findings have not been consistent.

Fish consumption was related to a relatively low risk of total mortality and sudden cardiac death, but not with a lower risk of MI or cardiovascular mortality in the Physicians’ Health Study of 20 551 healthy men followed for 11 years.22 There were 194 cases of MI and chest pain in a study of Finnish men initially free of CHD during 10 years of follow-up, with 44% fewer events in subjects in the highest quintile of serum n−3 PUFA compared with those in the lowest quintile.23 In a study of Chinese men, those who ate ≥50 g of fish/shellfish per week had a 44% lower risk of fatal MI compared with those who consumed less fish.24 Furthermore, in a study of healthy Japanese men and women aged 40–59 years followed for 11 years, there was a 37% lower risk of fatal and non-fatal CHD and a 56% lower risk of definite MI in those in the highest quintile of fish intake (median 180 g/day) compared with those in the lowest quintile (median 23 g/day).4 In contrast, no difference was seen in fatal or non-fatal CHD between the group with the highest fish intake (mean intake 119 g/day) and the group with no intake of fish in the Health Professionals Study of men aged 40–75 years during 6 years of follow-up.25

In a meta-analysis by He et al.,6 subjects who consumed fish at least once a month had a lower mortality from CHD than individuals with a lower intake of fish. Thus, each 20 g/day higher intake of fish was related to a 7% lower risk of CHD mortality. A recent intervention trial from Japan, the JELIS trial,26 included subjects (both with and without CHD) with hypercholesterolaemia who were randomized to treatment with low-dose statin with or without 1.8 g of purified eicosapentaenoic acid. The primary endpoint was a composite of fatal and non-fatal MI, unstable angina pectoris, percutaneous coronary interventions, or coronary artery bypass grafting during a mean follow-up period of 4.6 years. There was a borderline significant 18% reduction in major coronary events in the group of subjects without CHD at baseline supplemented with eicosapentaenoic acid.

In secondary intervention trials with hard clinical endpoints, fatty fish7 and fish oil capsules8,26,27 had a beneficial effect on various endpoints, e.g. total mortality,7 coronary mortality,8 sudden cardiac death,8 MI,27 and major coronary events.26 In contrast, patients with stable angina pectoris randomized to fatty fish twice-weekly had an increased risk of coronary mortality and sudden cardiac death.28

There are many explanations for the discordant results obtained, not least differences in study design and methodology. Differences in the type of fish consumed and preparation methods9,29 may further explain inconsistency in the relation between fish intake and CHD across populations. Specific data on the different types of fish consumed in the studies are limited. Unlike most other studies, we have differentiated between lean and fatty fish, distinguishing according to the content of n−3 PUFA and not according to the total amount of fat in fish. However, although sardines and tuna contain <1 g/100 g of n−3 PUFA based on The Danish Food Composition Tables and accordingly were classified as lean fish by us, they may in other studies be classified as fatty fish. However, tuna and sardines contributed little to the total fish intake in this Danish population (Table I) and a putative different classification of them did not affect the results (data not shown). Another explanation for the conflicting data from epidemiological studies may relate to an effect of pollutants. Some species of fish may contain environmental contaminants, where especially methylmercury has been brought into focus for its potential adverse effects on CHD.5,23,30 On the other hand, the content of selenium in fish may protect against a toxic effect of methylmercury.23,30

Regarding mechanisms, marine n−3 PUFA are incorporated into cell membrane phospholipids where they affect membrane function, cell signalling, arrhythmic properties, and a variety of other processes.5,3032 Indeed, some data have indicated antithrombotic4,24,26,27 or antiarrhythmic5,8,22,33 effects of marine n−3 PUFA, but results have not been uniform.11 Furthermore, a stabilizing (anti-inflammatory) effect of marine n−3 PUFA on human atherosclerotic plaques have also been reported.34 Marine n−3 PUFA also lower plasma triglycerides, impair platelet and leucocyte reactivity, lower heart rate, and may slightly reduce blood pressure,5,31,32 although not convincingly shown for the very modest intake of fatty fish found to be protective in the present study. Finally, recent data suggest that fish consumption may decrease the risk of heart failure, a strong predictor of coronary events.35,36

Several effects of marine n−3 PUFA may thus have contributed to the negative association between intake of fatty fish and ACS observed in the present study, perhaps not least a reduced incidence of clinically important plaque rupture/erosion with supervening thrombosis.

In conclusion, our study showed that a moderate intake of fatty fish is associated with a lower risk of ACS in middle-aged healthy men. Results among women were less consistent and further studies with more female cases are needed to clarify if fatty fish should be recommended for the prevention of ACS in women.


The Diet, Cancer and Health study was funded by the Danish Cancer Society. This study was funded by the Svend Andersen Foundation and the Danish Heart Association.

Conflict of interest: none declared.


View Abstract