A Quantitative Analysis of Fish Consumption and Coronary Heart Disease Mortality

A Quantitative Analysis of Fish Consumption and Coronary Heart Disease Mortality Ariane König, PhD, Colleen Bouzan, MS, Joshua T. Cohen, PhD, William E. Connor, MD, Penny M. Kris-Etherton, PhD, George M. Gray, PhD, Robert S. Lawrence, MD, David A. Savitz, PhD, Steven M. Teutsch, MD Abstract:

Although a rich source of n-3 polyunsaturated fatty acids (PUFAs) that may confer multiple health benefits, some fish contain methyl mercury (MeHg), which may harm the developing fetus. U.S. government recommendations for women of childbearing age are to modify consumption of high-MeHg fish to reduce MeHg exposure, while recommendations encourage fish consumption among the general population because of the nutritional benefits. The Harvard Center for Risk Analysis convened an expert panel (see acknowledgments) to quantify the net impact of resulting hypothetical changes in fish consumption across the population. This paper estimates the impact of fish consumption on coronary heart disease (CHD) mortality and nonfatal myocardial infarction (MI). Other papers quantify stroke risk and the impacts of both prenatal MeHg exposure and maternal intake of n-3 PUFAs on cognitive development. This analysis identified articles in a recent qualitative review appropriate for the development of a dose–response relationship. Studies had to satisfy quality criteria, quantify fish intake, and report the precision of the relative risk estimates. Relative risk results were averaged, weighted proportionately by precision. CHD risks associated with MeHg exposure were reviewed qualitatively because the available literature was judged inadequate for quantitative analysis. Eight studies were identified (29 exposure groups). Our analysis estimated that consuming small quantities of fish is associated with a 17% reduction in CHD mortality risk, with each additional serving per week associated with a further reduction in this risk of 3.9%. Small quantities of fish consumption were associated with risk reductions in nonfatal MI risk by 27%, but additional fish consumption conferred no incremental benefits. (Am J Prev Med 2005;29(4):335–346) © 2005 American Journal of Preventive Medicine

Introduction

A

substantial body of evidence supports the hypothesis that regular fish consumption may reduce risks from coronary heart disease (CHD). CHD is the leading cause of death worldwide, accounting for about 12.5% of total deaths.1 In the United States, about 515,000 individuals die from CHD each year, including approximately 250,000 CHDFrom the Harvard Center for Risk Analysis, Harvard School of Public Health (König, Bouzan, Cohen, Gray), Boston, Massachusetts; Division of Endocrinology, Diabetes and Clinical Nutrition, Oregon Health Sciences University (Connor), Portland, Oregon; Department of Nutritional Sciences, Pennsylvania State University (Kris-Etherton), University Park, Pennsylvania; Department of Health Policy and Management, Bloomberg School of Public Health, Johns Hopkins University (Lawrence), Baltimore, Maryland; Department of Epidemiology, School of Public Health, University of North Carolina (Savitz), Chapel Hill, North Carolina; and Department of Outcomes Research and Management, Merck & Co., Inc. (Teutsch), West Point, Pennsylvania Address correspondence and reprint requests to: Joshua T. Cohen, PhD, Harvard Center for Risk Analysis, 718 Huntington Avenue, Boston MA 02115. E-mail: [email protected].

induced sudden deaths.2 Early epidemiologic studies of Inuit and other populations with high rates of seafood consumption first indicated that such diets may decrease the incidence of CHD.3 Later, prospective cohort studies, case– control studies, and clinical trials provided additional, although not always unambiguous, evidence that one or two servings of fish per week may reduce CHD when compared to no fish intake (see review by Schmidt et al.4). The beneficial effects of fish consumption are thought to be largely attributable to the anti-arrhythmic activity of n-3 polyunsaturated fatty acids (PUFAs), for example, through their potential to reduce the risk of ventricular fibrillation. Evidence also suggests that n-3 PUFAs may protect against other diseases, such as stroke,5,6 although some studies hint that n-3 PUFAs could increase risk of stroke.7 Moreover, n-3 PUFAs may be important to fetal development.8 On the other hand, fish is a leading source of exposure to methyl mercury (MeHg), an environmental contaminant that may attenuate the protec-

Am J Prev Med 2005;29(4) © 2005 American Journal of Preventive Medicine • Published by Elsevier Inc.

0749-3797/05/$–see front matter doi:10.1016/j.amepre.2005.07.001

335

tive benefit of n-3 PUFAs against CHD9 –12 and adversely affect fetal development. Because of the potential for MeHg in fish to adversely affect fetal development, the U.S. Food and Drug Administration (FDA) and the U.S. Environmental Protection Agency (U.S.EPA) issued a joint advisory in March 2004 recommending that pregnant women modify their fish consumption.13 However, depending on how they are implemented, interventions to decrease exposure to MeHg may decrease overall fish consumption. For example, Oken et al.14 reported a 17% decrease in fish consumption among pregnant women following the release of the FDA’s 2001 MeHg advisory. Moreover, other members of the population could decrease their fish consumption as an unintended consequence of risk management actions targeting MeHg exposure among women of childbearing age. In order to understand the possible public health ramifications of alternative risk management actions, it is necessary to quantify potential health benefits and risks associated with plausible changes in population fish consumption patterns. This paper evaluates the impact of fish consumption on CHD, including fatal events (fatal sudden cardiac arrests and fatal myocardial infarctions [MIs]), and nonfatal MIs. These effects can then be compared to other risks and benefits associated with changes in fish consumption. Note that this analysis quantifies CHD risk as a function of fish consumption (servings per week). Implicitly, it therefore considers (i.e., “nets out”) the protective effects against CHD (due to n-3 PUFAs) and any adverse impact that increases CHD risk (due to MeHg). Because this approach estimates the net impact of fish consumption on CHD risk, the application of our findings to the U.S. population rests on the assumption that the n-3 PUFA and MeHg concentrations in fish consumed by the populations included here are similar to corresponding concentrations in fish consumed by Americans. Three other papers in this issue develop dose– response relationships between prenatal n-3 PUFA intake and IQ, prenatal MeHg exposure and IQ, and adult fish consumption and stroke incidence.15–17 A fifth paper, also in this issue, combines these results to estimate the aggregate health effects of hypothetical changes in fish consumption on public health.18 The remainder of this paper has three sections. The first section describes the use of observational studies on fish consumption and randomized controlled trials (RCTs) of n-3 PUFA intake to quantify the relationship between (1) fish consumption (servings per week) and CHD-event relative risk in individuals with no preexisting CHD; and (2) n-3 intake (grams per day) and CHD-event relative risk in individuals who do have pre-existing CHD. Studies for this analysis were identified by including a subset of those identified in a recent 336

qualitative review of the literature on this topic.19 The second section reviews the literature on the potential association between chronic low level MeHg exposure and CHD-event relative risk. The final section discusses the findings. We conclude that the evidence is adequate to quantify the relationship between fish consumption and CHD-event relative risk in the general population of individuals with no pre-existing CHD. It is not possible, however, to estimate the extent to which the observed relationship reflects attenuation resulting from exposure to MeHg in fish.

CHD Relative Risk Associated with Fish Consumption or n-3 PUFA Intake This section describes our development of a dose– response relationship for fish consumption or n-3 PUFA intake and CHD events. The analysis is divided by CHD status (individuals free of CHD vs individuals with pre-existing CHD) and by outcome (CHD mortality, and nonfatal MIs). Although angina pectoris and congestive heart failure (other forms of CHD) can also affect a patient’s quality of life, these endpoints are not considered because there is insufficient evidence that they are directly influenced by fish consumption.

Identification of Studies for Inclusion in Analysis Although there is an extensive literature on laboratory studies that elucidate mechanisms and lend credence to a beneficial relationship between fish consumption and CHD mortality risk, only epidemiologic studies provide the information necessary to quantify this relationship. A search of the Medline database was augmented with the findings of a recent review of this literature.19 While the purpose of that review was to evaluate qualitatively the evidence for a causal association between n-3 PUFA intake and cardiovascular disease, our purpose is to aggregate results to develop a quantitative dose–response relationship for a subset of endpoints (i.e., CHD death and nonfatal MI). Studies for inclusion in our analysis were identified by starting with those identified by Wang et al.19 In brief, Wang et al.19 first identified abstracts by searching Medline, Embase, and the Cochrane Central Register of Controlled Trials (4th quarter, 2002). Wang et al.19 also consulted with an expert panel to identify studies. They considered all studies that reported on at least five subjects and excluded studies reported only as letters or as abstracts or which followed subjects with conditions not related to cardiovascular disease. They then eliminated review articles, studies following inappropriate or pediatric populations (subjects aged ⬍19 years), studies that did not mention n-3 PUFA intake, studies involving n-3 PUFA intake exceeding 6 g/day, studies of prospective interventions of ⬍4 weeks in

American Journal of Preventive Medicine, Volume 29, Number 4

duration, studies that did not report outcomes of interest, and studies that reported only n-3 PUFA tissue levels but not intake rates. RCTs were omitted if follow-up was ⬍12 months. This analysis further limits attention to those studies that: (1) reported relative risks for nonfatal MI, or CHD-related mortality (corresponding to the sum of the risks for fatal MI and sudden cardiac death); (2) quantified risk relative to a no intake or very low intake reference group (fish consumption of less than one fish serving per month); (3) followed subjects approximately representative of the general population in terms of CHD risk factors (i.e., we omitted studies that limited attention to populations with particular important risk factors, such as smokers, or populations with protective characteristics, such as vegetarians); (4) had a study design rated by Wang et al.19 as either “A” (least bias; results are valid) or “B” (susceptible to some bias, but not sufficient to invalidate the results), but not “C” (significant bias that may invalidate the results); (5) evaluated the impact of either fish consumption or dietary supplements containing PUFAs found in fish (eicosapentaenoic acid [EPA] or docosahexaenoic acid [DHA], rather than precursors for those compounds). In addition to the above criteria, double counting information must be avoided. Where multiple analyses or multiple publications were available for a single observational study population, attention is limited to a single set of results, with our selection favoring: (1) analyses that are more recent (i.e., additional follow-up); (2) full cohort analyses over nested case– control evaluations; and (3) analyses that measure exposure in terms of total fish consumption (rather than some subset of fish, such as lean or fatty fish). This last criterion is included because our analysis estimates the aggregate impact of changes in population fish consumption (i.e., the net of any n-3 PUFA benefit and MeHg risk). We assume that differences in total fish consumption across a study cohort are a better proxy for this aggregate effect than are differences in consumption across a cohort for some subset of fish. Finally, analyses that controlled for a greater number of potential confounders are also favored. Note that the issues described in this paragraph are generally of much less importance in the context of randomized controlled trials. Where relevant, however, the same criteria are used (e.g., selecting analyses reflecting the most recent follow-up data). Table 1 lists studies eliminated from consideration, along with our rationale for omitting them. Seven observational studies were identified for inclusion in our analysis (Table 2), as were four randomized controlled trial studies (Table 3). Note that all the observational studies listed in Table 2 evaluated the primary prevention of CHD (i.e., they followed subjects not known to have pre-existing CHD), while all the RCT

studies in Table 3 evaluated the secondary prevention of CHD (i.e., they followed subjects who did have pre-existing CHD). The use of observational studies to evaluate the primary prevention of CHD results from the fact that this design is better suited to analyzing smaller risks (CHD events among individuals with no CHD history). Cost considerations generally limit the size and duration of RCTs compared to observational studies. As a result, such studies focus on subjects who have a higher baseline risk for the outcome in question. It is perhaps for this reason that all of the RCT studies satisfying our criteria followed subjects with pre-existing CHD. Finally, note that as illustrated in Table 4, the analyses selected from these observational studies controlled for a similar set of confounders.

Dose–Response Methodology To develop the dose–response relationships between the relative risk of various outcomes and either n-3 intake or fish consumption, the results from each of the relevant studies were first combined into a single data set. For example, to quantify the relationship between fish consumption and the risk of nonfatal MIs in individuals with no pre-existing CHD, the analysis combines the 12 nonreference group, relative-risk values from the Ascherio et al.,47 Hu et al.,50 and Mozaffarian et al.51 studies. These observations, weighted by their statistical precision, are then regressed against fish consumption (servings per week). This analysis assumes that statistical precision is inversely proportional to the squared width of the logtransformed, relative-risk confidence interval. That measure of precision is used because the parameter estimates in a logistic regression are normally distributed after log transformation. Hence, the width of the log-transformed confidence interval is proportional to the estimate’s standard error, and the square of the width is proportional to the estimate’s variance. The variance, in turn, is inversely proportional to the weight assigned an observation when aggregating data for a meta-analysis. For studies reporting exposure in terms of fish consumption, this analysis converts consumption rates expressed as ranges (e.g., “one to three fish servings per month”) into point estimates expressed as average fish consumption servings per week. When lower and upper bounds are specified for a range, the range’s midpoint is used (two fish servings per month in the preceding example, amounting to around 0.5 servings per week). If no upper bound is specified (e.g., “five or more servings per week”), the upper bound value is assumed to be seven fish servings per week. For studies that express fish consumption in terms of grams per day, it is assumed that 100 g of fish is equivalent to one serving. This assumption is consistent with U.S.EPA Am J Prev Med 2005;29(4)

337

338 American Journal of Preventive Medicine, Volume 29, Number 4

Table 1. Studies identified by Wang et al. (2002)19 not considered here for reasons other than design qualitya Reason for omitting study

Study (year)ref

Design 20

Bemelmans (2002) de Lorgeril (1999)21 Dolecek (1992)22 Erkkila (1999)23 Fraser (1992)24 Fraser (1997)25 Gillum (2000)26 Hu (1999)27 Leren (1966)28 Mann (1997)29 Morris (1995)30 Nagata (2002)31 Oomen (2001)32 Osler (2003)33,j Pietinen (1997)34 Sasazuki (2001)35 Singh (2002)36 Siscovick (1995)37 Yuan (2001)38 a

RCT RCT Cohort RCT Cohort Cohort Cohort Cohort RCT Cohort Cohort Cohort Cohort Cohort Cohort Cohort RCT Cohort Cohort

Did not report risk for either nonfatal MI or CHD mortality

Risk not reported relative to very low or no n-3 PUFA intake group

CHD baseline risk differs substantially from risk for general population

Study quality rated as “poor” by Wang (2002)19

Study did not evaluate impact of either fish or n-3 PUFAs in fishb

X

X X

X

X X

Study subsumed by later analysis of same cohort

Xc Xd Xe e

X Xf

Xg Xh i

X

X X Xk X X

Xl X

X

Xm Xn

Also eliminated were studies designated by Wang et al.19 as having a poor design (grade “C”).21,28,36,39 – 44 These analyses evaluated risk in terms of alpha-linolenic acid (ALA) intake, not intake of either fish or n-3 PUFAs found in fish (EPA or DHA). Although humans can convert ALA to EPA and DHA, the conversion rates are slow and not well understood. For this reason, ALA intake may not serve as an effective surrogate for EPA and DHA intake. c This analysis22 reported risks for all-cause mortality only. d This analysis23 evaluated risk for all-cause mortality, cardiovascular mortality, and coronary artery disease mortality. e The samples for these studies24,25 were restricted to Seventh Day Adventists (virtually no current smokers, high levels of exercise, low consumption of fish and beef, high consumption of nuts, whole wheat bread, fruits, and salads). For one study,25 the sample was further restricted to individuals aged ⬎84 years. f Gillum et al.26 evaluated risk for all-cause mortality, cardiovascular disease mortality, and for mortality due to causes other than cardiovascular disease. g Study sample was restricted to health-conscious individuals (vegetarians and the friends of vegetarians). The incidence of ischemic heart disease for the sample was less than one half the rate expected for the population. h Subsumed by Albert et al.45 i Nagata et al.31 evaluated only all-cause mortality, cancer, cardiovascular disease mortality, and all other causes of mortality. j It appears that one of the confidence intervals reported in this analysis33 is in error because it is not symmetric in log space. See the relative risk reported for CHD mortality for all subjects consuming fish two times per month. The central estimate for this interval is 0.98, while the bounds are 0.64 and 1.21. k The reference group in this study33 corresponded to an intermediate level of fish intake relative to other cohort groups (1 meal/week vs ⱕ1 meal/month and 2 meals/month). Although a central estimate for the relative risk can be computed for each exposure group using the lowest-dose group as the reference, it is not clear how to compute the resulting variance and hence the precision weights. l Study sample restricted to smoking males. m Siscovick et al.37 reported risk only for primary cardiac arrest death plus MI. n Yuan et al.38 reported risk estimates separately for acute MI mortality and for other ischemic heart disease mortality, but did not report relative risk for these two causes combined. CHD, coronary heart disease; DHA, docosahexaenoic acid; EPA, eicosapentaenoic acid; MI, myocardial infarction; PUFAs, polyunsaturated fatty acids; RCT, randomized controlled trials. b

Table 2. Studies of subjects without CHD at baseline Fish intake

Study (year)ref

Population 46

Population country

Follow-up (years and person-years) c

Am J Prev Med 2005;29(4)

Kromhout (1985)

872 men aged 40–59 years with no CHD at baseline

Netherlands

Ascherio (1995)47

44,895 male health workers with no known CHD at baseline

U.S.

6 242,029

Daviglus (1997)48

1822 males free of CVD at baseline

U.S.

30 47,153

Albert (1998)45

20,551 male physicians with no MI, cerebrovascular disease, or cancer at baseline

U.S.

11 253,777

Oomen (2000)49

1097 males aged 50–69d free of CHD at baseline

Italy

20

Hu (2002)50

84,688 female nurses with no cancer, angina, MI, or CVD at baseline

U.S.

16 1,307,157

Mozaffarian (2003)51

3910 Medicare enrollees with no known CVD at baseline

U.S.

11 ⬃36,400

a

20

Reported in studya

Our estimateb (servings/ week)

0 g/day 1–14 g/day 15–29 g/day 30–44 g/day ⱖ45 g/day ⬍1/mo 1–3/mo 1/wk 2–3/wk 4–5/wk ⱖ6/wk 0 g/day 1–17 g/day 18–34 g/day ⱖ35 g/day ⬍1/mo 1–3/mo 1–⬍2/wk 2–⬍5/wk ⱖ5/wk 0 0–19 g/day 20–39 g/day ⱖ40 g/day ⬍1/mo 1–3/mo 1/wk 2–4/wk ⱖ5/wk ⬍1/mo 1–3/mo 1/wk 2/wk ⱖ3/wk

0 0.5 1.5 2.6 5.1 0 0.5 1.0 2.5 4.5 6.5 0 0.63 1.8 4.7 0 0.5 1.5 3.5 6 0 0.66 2.1 4.9 0 0.5 1 3 6 0 0.5 1 2 5

Relative risk (95% CI)

Nonfatal MI

1 0.62 (0.39–1.00) 0.80 (0.55–1.17) 0.67 (0.46–0.97) 0.69 (0.46–1.04) 0.96 (0.63–1.47)

1 0.76 (0.60–1.01) 0.75 (0.59–0.96) 0.71 (0.53–0.96) 0.77 (0.54–1.11) 1 0.81 (0.51–1.26) 0.71 (0.44 –1.15) 0.75 (0.46 –1.21) 0.67 (0.42–1.07)

CHD death or fatal MI ⴙ sudden death 1 0.63 (0.32–1.26) 0.56 (0.27–1.15) 0.36 (0.14–0.93) 0.39 (0.13–1.15) 1 0.74 (0.38–1.45) 0.86 (0.50–1.47) 0.71 (0.41–1.21) 0.54 (0.29–1.00) 0.77 (0.41–1.44) 1 0.88 (0.63–1.22) 0.84 (0.61–1.17) 0.62 (0.40–0.94) 1 1.18 (0.59–2.36) 0.82 (0.45–1.51) 0.91 (0.50–1.66) 0.81 (0.41–1.61) 1 0.94 (0.55–1.59) 0.93 (0.53–1.63) 0.67 (0.33–1.39) 1 0.80 (0.56–1.15) 0.65 (0.46–0.91) 0.72 (0.48–1.09) 0.55 (0.33–0.91) 1 0.78 (0.47–1.28) 0.77 (0.45–1.32) 0.53 (0.30 – 0.96) 0.47 (0.27– 0.82)

339

Study does not specify total years of follow-up. Estimated as the midpoint of the lower and upper ends of the range provided in the original study. If only a lower bound is provided, it is assumed the upper bound is seven servings per week. If the original study expressed fish consumption in grams, it is assumed that one serving is 100 grams. c Total years of follow-up not specified. d This analysis omits 553 Dutch subjects from consideration because they were part of the Kromhout et al.46 study. This analysis also omits 1088 Finns because there was no zero-consumption reference group. The lowest consumption group was 0 –19 g/day. CI, confidence interval; CHD, coronary heart disease; CVD, cardiovascular disease; MI, myocardial infarction. b

Table 3. Studies of subjects with pre-existing CHD at baseline Relative risk (95% CI)a Assumed n-3 intake for treatment group (g/day)

Mean follow-up (months)

Population country

Nonfatal MI

CHD death or fatal MI ⴙ sudden death

Study (year)ref

Intervention

Sacks (1995)52

Intervention: oil (2.9 g/day EPA, 1.9 g/ day DHA), n⫽31 Control: placebo, n⫽28

4.8

28

U.S.

0.4 (0.0–5.1)

0.3 (0.0–7.4)b

Leng (1998)53

Intervention: oil (1.7 g/day GLA, 0.3 g/ day EPA), n⫽60 Control: sunflower oil, n⫽60

1.7

24

U.K.

0.7 (0.2–3.4)

RR not reported for fatal MI or sudden death

Nilsen (2001)54

Intervention: oil (1.2 g/day EPA, 2.3 g/ day DHA), n⫽150 Control: corn oil, n⫽150

3.5

18

Norway

RR not reported for nonfatal MI

1.0 (0.4–2.7)

Marchioli (2002)55

Intervention: oil (0.3 g/day EPA, 0.6 g/ day DHA), n⫽5666 Control: no intervention, n⫽5658

0.9

42

Italy

1.0 (0.8–1.2)

0.6 (0.5–0.7)

a

Computed using SAS, version 8.2 for Windows (SAS Institute, Cary NC, 2001) PROC FREQ relrisk output option. Added 0.5 to each cell in the two-by-two table when one cell was empty in order to prevent the width of the confidence interval for the log-transformed relative risk from becoming infinitely wide. CHD, Coronary heart disease; CI, confidence interval; DHA, docosahexaenoic acid; EPA, eicosapentaenoic acid; GLA, gamma-linolenic acid; MI, myocardial infarction; RR, relative risk.

b

estimates.56 The RCTs of individuals with pre-existing CHD report n-3 PUFA intake in grams per day, which is used as the independent variable in a separate regression analysis. The potential mechanism underlying the role of n-3 PUFAs in reducing adverse cardiac events helps to inform the mathematical forms considered for the dose–response relationships. Connor57 reviews the two main mechanisms advanced. First, n-3 PUFA intake may reduce the tendency for arrhythmias to develop. If this is the case, then increased n-3 intake might reduce the risk of sudden death due to cardiac arrest, but would not influence MI incidence or the probability that MIs are fatal. Referring to the Physicians Health Study,45 Connor57 notes evidence of a plateau in CHD sudden-death mortality risk at an n-3 intake level of between 0.3 and 2.7 g/month (0.01 and 0.09 g/day). 340

Such a plateau may exist if, for example, a minimal n-3 PUFA intake rate is sufficient to “stabilize” the myocardium, hence eliminating the potential for further benefits with additional intake. Albert et al.45 review other evidence that reduction of arrhythmia is the main mechanism by which n-3 influences CHD mortality risks. The second mechanism proposed by Connor57 is the potential for n-3 PUFA intake to reduce the formation of atherosclerotic plaques, thus reducing the risk of experiencing an MI. There is no evidence that this mechanism has a plateau at low consumption rates typical among Americans. Albert et al.45 note that results reported by Daviglus et al.48 contradict the hypothesis that n-3 PUFAs act primarily by countering arrhythmias, and hence sudden deaths, although they point out that the discrepancy with the arrhythmia

American Journal of Preventive Medicine, Volume 29, Number 4

Table 4. Confounders controlled in selected observational study analyses Study (year)ref

Age

Body mass index

Hypertension

Hypercholesterolemia

✓

✓

Diabetes

Am J Prev Med 2005;29(4)

Kromhout (1985)46

✓

Ascherio (1995)47 Daviglus (1997)48

✓

✓

✓

✓

✓

✓

✓

✓

✓

✓

Albert (1998)45

✓

✓

✓

✓

✓

Oomen (2000)49 Hu (2002)50

✓

✓

✓

✓

✓

✓

✓

Mozaffarian (2003)51

✓

✓

✓

✓

✓

Type of food consumed

Total energy intake

Smoking

✓

✓

✓ ✓

✓

✓

✓

✓

Alcohol consumption

Skinfold thickness, physical activity, occupation ✓

✓ ✓

✓

✓

✓

✓

✓

✓

✓

Other

✓

Education, religion, abnormal ECG Random aspirin or ␤-carotine treatment group assignment, prior cardiovascular disease, exercise, vitamin use Menopausal status, hormone use, vitamin use, family myocardial infarction history Gender, education, triglycerides

341

Table 5. Relationship between fish consumption and CHD event relative risk: studies of individuals with no pre-existing CHD CHD death

Nonfatal MI

Analysis

Parameter

R2

⌬RR

95% CI

R2

⌬RR

95% CI

Linear regression

Intercept Servings/wk

23%

⫺0.17 ⫺0.039

⫺0.25 to ⫺0.088 ⫺0.066 to ⫺0.011

5.8%

⫺0.27 0.0083

⫺0.34 to ⫺0.21 ⫺0.012 to 0.028

Quadratic regression

Intercept Servings/wk (Servings/wk)2

25%

0.13 ⫺0.085 0.0076

⫺0.26 to 0.002 ⫺0.20 to 0.03 ⫺0.011 to 0.026

45%

⫺0.19 ⫺0.084 0.014

⫺0.27 to ⫺0.12 ⫺0.15 to ⫺0.015 0.004 to 0.025

CHD, coronary heart disease; CI, confidence interval; MI, myocardial infarction; RR, relative risk.

hypothesis may be due to incorrect ascertainment of the exact cause of death. This analysis uses regression of aggregate CHD risk (sudden death plus fatal MI) against fish consumption to investigate the plausibility of the dose–response relationships implied by these two mechanisms. For example, for the CHD mortality risk analysis, a finding that the intercept term is distinct from zero (i.e., its distribution of plausible values is not centered on zero) supports the hypothesis that any level of fish consumption confers protection against this risk compared to eating no fish. A finding that the coefficient for the linear term is distinct from zero supports the hypothesis that further consumption of fish confers incremental protection against this risk. Finally, a sensitivity analysis adds a quadratic term to the linear regression. The quadratic term allows the regression to approximate the dose–response relationship if incremental benefits decrease at higher levels of n-3 intake or fish consumption.

tion among individuals already consuming at least some fish (at least one serving per month) confers additional protection. In fact, the coefficient for the linear term is positive. On the other hand, the confidence interval for the intercept term excludes zero, indicating that consumption of some fish confers protection against nonfatal MIs compared to consuming no fish. Because the linear term has a biologically implausible positive value, this analysis estimates the incremental impact of consuming some fish versus consuming no fish by computing the average relative risk across all fish-consuming groups. That computation suggests that consuming some fish reduces nonfatal MI risk by 25% (95% confidence interval of ⫺0.29 –⫺0.22) compared to consuming no fish. Given the limited number of data points in this regression (13) and the fact that the linear regression did not help to explain the data, the quadratic dose–response relationship is omitted from further consideration.

Individuals with Pre-Existing CHD Results This section discusses in turn the studies of individuals with no pre-existing CHD (observational studies) and the studies of individuals with pre-existing CHD (RCTs).

Our assessment for individuals with pre-existing CHD was complicated by the limited RCT data. As detailed in

Individuals with No Pre-Existing CHD Table 5 summarizes the parameter estimates for the combined results from the studies of individuals with no pre-existing CHD. Figure 1 illustrates the linear regression for CHD mortality risk. The intercept term has a central value of ⫺0.17, suggesting that a low level of fish consumption reduces CHD mortality risk by 17% compared to no fish consumption. The linear term’s coefficient has a central estimate value of ⫺0.039, indicating that each additional fish serving per week reduces risk incrementally by 3.9%, with a confidence interval that excludes zero. The quadratic regression offers little improvement in fit and is therefore disregarded. For nonfatal MIs, the linear regression results do not support the hypothesis that additional fish consump342

Figure 1. Regression of CHD mortality risk versus fish consumption in the general population. Note: The area of each data point is proportional to its statistical weight. The upper and lower bands denote the 95% confidence interval on mean of the predicted value. CHD, coronary heart disease.

American Journal of Preventive Medicine, Volume 29, Number 4

Table 6. Studies of the association between MeHg exposure and CHD Exposure measure(s)

Control for n-3 exposure

CHD outcome

Findings for MI and CHD death

1833 (Salonen) or 1871 (Rissanen) men in eastern Finland 42–60 yrs with no CHD, CVD stroke history, claudication, or cancer at baseline. Follow-up: ⬃7 yrs (Salonen), ⬃10 yrs (Rissanen)

Hair Hg (␮g/g), fish intake (g/day)

No (Salonen) Yes (Rissanen)

Salonen: Acute MI, CHD death, CVD death, and allcause death

CHD mortality: RR⫽1.21 (1.04– 1.40) per ␮g/g Hg in hair

Rissanen: Acute MI, acute chest pain

Total MI: RR⫽1.07 (0.97–1.18) per ␮g/g Hg in hair

Ahlqwist (1999)60

1462 Swedish women aged 38–60 at baseline. Follow-up: 24 years

Serum Hg

No

MI

When controlled for age and education, p ⬎ 0.2 for MI, p⫽0.144 for fatal MI. Correlation ⬍0, suggesting higher Hg exposure reduces risk

Hallgren (2001)61

78 first-ever MI cases from Northern Sweden matched to 156 controls on gender, age, date of health survey, region

Blood Hg

Yes

First MI

No consistent association with Hg exposure

Guallar (2002)10

684 first-ever MI cases matched with 724 controls. Population included men ⱕ70 yrs from any of 8 European countries or Israel

Toenail Hg

Yes

First MI

Highest Hg exposure quintile RR⫽2.16 (1.09–4.29) vs lowest Hg exposure quintile

Yoshizawa (2002)9

470 cases and 464 controls drawn from 33,737 male health professionals with no cancer, MI, angioplasty at baseline. Follow-up: ⬃5 yrs

Toenail Hg

Yes

CHD (fatal CHD, nonfatal MI, coronary-artery bypass surgery, angioplasty

No consistent association with Hg exposure. Highest Hg exposure quintile RR⫽1.03 (0.65–1.65) vs lowest Hg exposure quintile

Study (year)ref

Population and follow-up

Salonen (1995)11 and Rissanen (2000)12

CHD, coronary heart disease; CVD, cardiovascular disease; Hg, mercury; MeHg, methyl mercury; MI, myocardial infarction; RR, relative risk.

Table 3, four satisfactory RCTs have been identified, which collectively provide three data points for each of the two endpoints analyzed (CHD death and nonfatal MI). The assessment is further complicated by the fact that the n-3 PUFA intake rates investigated in the RCTs are far higher than levels corresponding to typical fish intake rates. For example, Carrington and Bolger58 reported average n-3 concentrations on the order of 1% in a market basket of fish typically consumed in the United States. Hence, one fish serving per week of 100 g would correspond to daily n-3 intake of 0.14 g—far less than the 1 to 5 grams of n-3 intake investigated in these trials (Table 3). Because there are only three data points available for this analysis, and because the doses investigated substantially exceed those typically associated with fish consumption in the United

States, we conclude the information available is insufficient for the purpose of quantitatively analyzing the impact of fish consumption on CHD risk for individuals with pre-existing CHD.

Mercury Exposure and CHD Relative Risk Although fish consumption offers nutritional benefits that may protect against CHD events, there is some evidence that MeHg in fish may attenuate those benefits to at least some degree. In particular, MeHg may increase the risk of atherosclerotic disease both by promoting the formation of free radicals and by compromising mechanisms that neutralize these agents.59 In order to identify epidemiologic studies to help quantify the relationship between MeHg exposure and Am J Prev Med 2005;29(4)

343

relative risk for CHD events, the Medline database was searched (title, abstract, or subject heading containing “mercury” and subject heading of either “coronary disease” or “myocardial infarction”; search conducted on February 23, 2004), yielding 41 hits. Of these, five studies were identified that reported on the association between MeHg exposure and CHD events (Table 6). Because of their design and the type of outcomes investigated, we judged it to be inappropriate to use these studies to quantify the extent to which mercury attenuates the relationships described above. First, only one study11 isolates the relationship between MeHg exposure and CHD mortality, and none isolates the relationship between MeHg and nonfatal MIs. Second, the two most recent studies used Hg concentrations in toenail samples as a biomarker. At this time, however, there is no information on how these concentrations relate to fish consumption or to any other biomarker of MeHg exposure. Third, while some of the studies controlled in some way for n-3 intake, others did not (Table 6, column 4). Results from studies that did not control in some way for n-3 intake are informative of the combined impact regardless of the MeHg exposure and n-3 intake that the study participants experienced. Qualitatively, the studies in Table 6 are ambiguous. Results from eastern Finland11,12 and the results reported by Guallar et al.10 support the hypothesis that MeHg exposure increases the risk of CHD events. On the other hand, results reported by Yoshizawa et al.9 and by Hallgren et al.61 revealed no consistent association. Ahlqwist et al.60 found a negative association between MeHg exposure and MI risk.

Discussion This paper describes analyses for the purpose of developing dose–response relationships for individuals who do not have pre-existing CHD (fish consumption and CHD events) and for individuals who do (n-3 intake and CHD events). For the general population (no pre-existing CHD), fish consumption is associated with a decrease in the relative risk of CHD mortality by 17% compared to no fish consumption (less than one fish serving per month). The fact that the confidence interval for this estimate excludes zero supports the hypothesis for the first mechanism described earlier, that is, that n-3 PUFA intake reduces the tendency for arrhythmias to develop. An incremental increase of fish consumption by one serving per week is associated with a further decrease in risk of 3.9%. The confidence interval for this estimate excludes zero, supporting the hypothesis for the second mechanism described earlier, that is, that n-3 PUFA intake reduces the formation of atherosclerotic plaques. The epidemiologic data also suggest that fish consumption is associated with a reduction in nonfatal MI risk of 25% compared to no fish consumption, al344

though no additional benefits accompany further increases in fish consumption. However, the biological mechanism that might give rise to a dose–response relationship with this shape (a reduction in the tendency for arrhythmias to develop) would not influence the incidence of MIs. As a result, we conclude that the impact of fish consumption on nonfatal MI risk is very uncertain. At this time, we conclude that there are insufficient data to develop a reliable dose–response relationship for individuals with pre-existing CHD for either CHD mortality or nonfatal MI risk. Our findings for CHD mortality risk for the general population are similar to those reported for a recent meta-analysis conducted by He et al.62 That analysis reported that for each 20 g/day increase in fish consumption, relative risk for CHD mortality decreases by 7% (95% confidence interval, 1%–13%). By way of comparison, our findings correspond to a 5.5% decrease in relative risk per 20 g fish/day (assuming that one serving per week, that is, 100 g fish/week, corresponds to approximately 14 g fish/day). This analysis noted two differences between this study and the He et al.62 analysis. First, it considered the possibility that the dose–response relationship is nonlinear, and in particular that consumption of any fish confers a large benefit with smaller incremental benefits associated with further fish consumption. This difference probably explains why the incremental risk reduction reported by He et al.62 (7% per 20 g of fish/day) is somewhat larger as the benefit estimated in our analysis (5.5% per 20 g/day). In short, the extra incremental benefit reported by He et al.62 has been absorbed into the initial benefit identified in this paper associated with some versus no fish consumption. Second, there were several differences in the studies included in the two analyses. This analysis omitted two studies that He et al.62 used in their meta-analysis, including Yuan et al.38 because it did not report relative risk for all CHD mortality combined (fatal MIs are broken out separately), and Osler et al.33 because the comparison group was not the lowest exposure group in the study. Our conclusions for the general population are based on studies of fish consumption. Because these studies are observational, their results could reflect the impact of inadequately controlled confounders. As described earlier, however, most of these studies did control for typical CHD risk factors. The observational studies have the advantage of measuring the impact of fish consumption at levels often observed among Americans. Of course, this impact may be mediated by n-3 PUFAs, MeHg, and even other fish constituents not studied here. It is not possible to isolate quantitatively the contributions of these various constituents. It is also possible that the dose–response relationships described here have quadratic components that may imply a decreased incremental benefit at higher levels of fish

American Journal of Preventive Medicine, Volume 29, Number 4

consumption. If a quadratic component does exist, it was not evident in the data analyzed in this paper. It seems that given the available data, use of a linear function to describe the dose–response relationship is adequate at this time. Although an insufficient number of RCTs have been conducted to make meaningful inferences from these studies, this discussion notes several characteristics of this study design. RCTs are less vulnerable than observational studies to problems introduced by confounders. Moreover, they isolate the impact of n-3 PUFAs on the risk of CHD events. The administered doses in these studies, however, are far higher than the corresponding n-3 PUFA intake levels associated with typical rates of fish consumption. The dose–response relationships developed here are useful for the purpose of addressing some of the policy questions central to our overall project. In particular, they can be used to estimate the extent to which increases or decreases in fish consumption affect CHD mortality risk and nonfatal MI risk among members of the general population (i.e., individuals who do not already have CHD). On the other hand, it is not possible to determine how changes in the type of fish that people consume will affect risk. Because different types of fish have different concentrations of both n-3 PUFAs and MeHg,58 it would be useful to know how each component individually influences risk; such information is unfortunately not available. With the limitations just described in mind, we believe that the dose–response estimates developed here serve as a starting point for quantifying the public health implications of changes in fish consumption patterns. The expert panel convened by the Harvard Center for Risk Analysis for this project was chaired by Steven M. Teutsch. In addition to William Connor, Penny Kris-Etherton, Robert Lawrence, and David Savitz, who are co-authors of this paper, the panel consisted of David C. Bellinger, PhD (Department of Neurology, Children’s Hospital, Boston MA), and Bennett A. Shaywitz, MD (Department of Pediatrics and Neurology, Yale University, New Haven CT). This work was supported by a grant from the National Food Processors Association Research Foundation (NFPA-RF) and the Fisheries Scholarship Fund. Member companies of the NFPA-RF may be affected by the findings of research that funded my participation on the panel that wrote this paper.

References 1. Murray CJL, Lopez AD. Alternative projections of mortality and disability by cause 1990 –2020: global burden of disease study. Lancet 1997;349: 1498 –504. 2. American Heart Association. Heart disease and stroke statistics. Dallas TX, 2003. Available at: www.americanheart.org/downloadable/heart/ 1040391091015HDS_Stats_03.pdf. Accessed December 2003. 3. Bang HO, Dyerberg J, Nielsen AB. Plasma lipid and lipoprotein pattern in Greenlandic West-Coast Eskimos. Lancet 1971;1:1143–5.

4. Schmidt EB, Skou HA, Christensen JH, Dyerberg J. N-3 fatty acids from fish and coronary artery disease: implications for public health. Public Health Nutr 2000;3:91– 8. 5. Bang HO, Dyerberg J. The composition of food consumed by Greenlandic Eskimos. Acta Med Scand 1973;200:69 –73. 6. Bang HO, Dyerberg J. Hemostatic function and platelet polyunsaturated fatty acids in Eskimos. Lancet 1979;2:433–5. 7. Orencia AJ, Daviglus ML, Dyer AR, Shekelle RB, Stamler J. Fish consumption and stroke in men: 30 findings of the Chicago Western Electric Study. Stroke 1996;27. 8. Helland IB, Smith L, Saarem K, Saugstad OD, Drevon CA. Maternal supplementation with very long chain n-3 fatty acids during pregnancy and lactation augments children’s IQ at 4 years of age. Pediatrics 2003;111: E39 – 44. 9. Yoshizawa K, Rimm EB, Morris JS, et al. Mercury and the risk of coronary heart disease in men. N Engl J Med 2002;347:1755– 60. 10. Guallar E, Sanz-Gallardo MI, van’t Veer P, et al. Mercury, fish oils, and the risk of myocardial infarction. N Engl J Med 2002;347:1747–54. 11. Salonen JT, Seppanen K, Nyyssonen K, et al. Intake of mercury from fish, lipid peroxidation, and the risk of myocardial infarction and coronary, cardiovascular, and any death in eastern Finnish men. Circulation 1995;91:645–55. 12. Rissanen T, Voutilainen S, Nyyssonen K, Lakka TA, Salonen JT. Fish oil– derived fatty acids, docosahexaenoic acid and docosapentaenoic acid, and the risk of acute coronary events: the Kuopio ischaemic heart disease risk factor study. Circulation 2000;102:2677–9. 13. U.S. Department of Health and Human Services, U.S. Environmental Protection Agency. What you need to know about mercury in fish and shellfish, 2004. Available at: www.cfsan.fda.gov/⬃dms/admehg3.html. Accessed December 2004. 14. Oken E, Kleinman KP, Berland WE, Simon S, Rich-Edwards JW, Gillman MW. Decline in fish consumption among pregnant women after a national mercury advisory. Obstet Gynecol 2003;102:346 –51. 15. Bouzan C, Cohen JT, Connor WE, et al. A quantitative analysis of fish consumption and stroke risk. Am J Prev Med 2005;29:347–52. 16. Cohen JT, Bellinger DC, Shaywitz BA. A quantitative analysis of prenatal methyl mercury exposure and cognitive development. Am J Prev Med 2005;29:353– 65. 17. Cohen JT, Bellinger DC, Connor WE, Shaywitz BA. A quantitative analysis of prenatal intake of n-3 polyunsaturated fatty acids and cognitive development. Am J Prev Med 2005;29:366 –74. 18. Cohen JT, Bellinger DC, Connor WE, et al. A quantitative risk benefit analysis of changes in population fish consumption. Am J Prev Med 2005;29:324 –33. 19. Wang C, Chung M, Balk E, et al. Effects of omega-3 fatty acids on cardiovascular disease. Boston MA: Tufts–New England Medical Center EPC. Prepared for Agency for Healthcare Research and Quality, Rockville MD, 2004 (contract 290-02-0022). Available at: www.ahrq.gov/clinic/ epcsums/o3cardsum.htm. Accessed April 2004. 20. Bemelmans WJ, Broer J, Feskens EJ, et al. Effect of an increased intake of alpha-linolenic acid and group nutritional education on cardiovascular risk factors: the Mediterranean Alpha-linolenic Enriched Groningen Dietary Intervention (MARGARIN) study. Am J Clin Nutr 2002;75:221–7. 21. de Lorgeril M, Salen P, Martin JL, Monjaud I, Delaye J, Mamelle N. Mediterranean diet, traditional risk factors, and the rate of cardiovascular complications after myocardial infarction: final report of the Lyon Diet Heart Study. Circulation 1999;99:779 – 85. 22. Dolecek TA. Epidemiological evidence of relationships between dietary polyunsaturated fatty acids and mortality in the multiple risk factor intervention trial. Proc Soc Exp Biol Med 1992;200:177– 82. 23. Erkkila AT, Sarkkinen ES, Lehto S, Pyorala K, Uusitupa MI. Dietary associates of serum total, LDL, and HDL cholesterol and triglycerides in patients with coronary heart disease. Prev Med 1999;28:558 – 65. 24. Fraser GE, Sabate J, Beeson WL, Strahan TM. A possible protective effect of nut consumption on risk of coronary heart disease. The Adventist Health Study. Arch Intern Med 1992;152:1416 –24. 25. Fraser GE, Shavlik DJ. Risk factors for all-cause and coronary heart disease mortality in the oldest-old. The Adventist Health Study. Arch Intern Med 1997;157:2249 –58. 26. Gillum RF, Mussolino M, Madans JH. The relation between fish consumption, death from all causes, and incidence of coronary heart disease. The NHANES I Epidemiologic Follow-up Study. J Clin Epidemiol 2000;53: 237– 44.

Am J Prev Med 2005;29(4)

345

27. Hu FB, Stampfer MJ, Manson JE, et al. Dietary intake of alpha-linolenic acid and risk of fatal ischemic heart disease among women. Am J Clin Nutr 1999;69:890 –7. 28. Leren P. The effect of plasma cholesterol lowering diet in male survivors of myocardial infarction. Acta Med Scand 1966;466(suppl):1–92. 29. Mann JI, Appleby PN, Key TJ, Thorogood M. Dietary determinants of ischaemic heart disease in health conscious individuals. Heart 1997;78: 450 –5. 30. Morris MC, Manson JE, Rosner B, Buring JE, Willett WC, Hennekens CH. Fish consumption and cardiovascular disease in the Physicians’ Health Study: A prospective study. Am J Epidemiol 1995;142:166 –75. 31. Nagata C, Takatsuka N, Shimizu H. Soy and fish oil intake and mortality in a Japanese community. Am J Epidemiol 2002;156:824 –31. 32. Oomen CM, Ocke MC, Feskens EJ, van Erp-Baart MA, Kok FJ, Kromhout D. Association between trans fatty acid intake and 10-year risk of coronary heart disease in the Zutphen Elderly Study: a prospective population-based study. Lancet 2001;357:746 –51. 33. Osler M, Andreasen AH, Hoidrup S. No inverse association between fish consumption and risk of death from all-causes, and incidence of coronary heart disease in middle-aged Danish adults. J Clin Epidemiol 2003;56: 274 –9. 34. Pietinen P, Ascherio A, Korhonen P, et al. Intake of fatty acids and risk of coronary heart disease in a cohort of Finnish men. The Alpha-Tocopherol, Beta-Carotene Cancer Prevention Study. Am J Epidemiol 1997;145:876 – 87. 35. Sasazuki S. Case– control study of nonfatal myocardial infarction in relation to selected foods in Japanese men and women. Jpn Circ J 2001;65:200 – 6. 36. Singh RB, Niaz MA, Ghosh S, et al. Effect of an Indo-Mediterranean diet on progression of coronary artery disease in high-risk patients (Indo-Mediterranean Diet Heart Study): a randomized single-blind trial. Lancet 2002;360:1455– 61. 37. Siscovick DS, Raghunathan TE, King I, et al. Dietary intake and cell membrane levels of long-chain n-3 polyunsaturated fatty acids and the risk of primary cardiac arrest. JAMA 1995;274:1363–7. 38. Yuan JM, Ross RK, Gao YT, Yu MC. Fish and shellfish consumption in relation to death from myocardial infarction among men in Shanghai, China. Am J Epidemiol 2001;154:809 –16. 39. Egeland GM, Middaugh JP. Balancing fish consumption benefits with mercury exposure. Science 1997;278:1904 –5. 40. Rodriguez BL, Sharp DS, Abbott RD, et al. Fish intake may limit the increase in risk of coronary heart disease morbidity and mortality among heavy smokers. The Honolulu Heart Program. Circulation 1996;94:952– 6. 41. Kromhout D, Feskens EJ, Bowles CH. The protective effect of a small amount of fish on coronary heart disease mortality in an elderly population. Int J Epidemiol 1995;24:340 –5. 42. Singh RB, Niaz MA, Sharma JP, Kumar R, Rastogi V, Moshiri M. Randomized, double-blind, placebo-controlled trial of fish oil and mustard oil in patients with suspected acute myocardial infarction: the Indian experiment of infarct survival. Cardiovasc Drugs Ther 1997;11:485–91. 43. Burr ML, Gilbert JF, Holliday RM, et al. Effects of changes in fat, fish, and fibre intakes on death and myocardial reinfarction: diet and reinfarction trial (DART). Lancet 1989;2:757– 61. 44. Burr ML, Ashfield-Watt PA, Dunstan FD, et al. Lack of benefit of dietary advice to men with angina: results of a controlled trial. Eur J Clin Nutr 2003;57:193–200.

346

45. Albert CM, Hennekens CH, O’Donnell CJ, et al. Fish consumption and risk of sudden cardiac death. JAMA 1998;279:23– 8. 46. Kromhout D, Bosschieter EB, Coulander CL. The inverse relation between fish consumption and 20-year mortality from coronary heart disease. N Engl J Med 1985;312:1205–9. 47. Ascherio A, Rimm EB, Stampfer MJ, Giovannucci EL, Willett WC. Dietary intake of marine n-3 fatty acids, fish intake, and the risk of coronary disease among men. N Engl J Med 1995;332:977– 82. 48. Daviglus ML, Stamler J, Orencia AJ, et al. Fish consumption and the 30-year risk of fatal myocardial infarction. N Engl J Med 1997;336:1046 –53. 49. Oomen CM, Feskens EJ, Rasanen L, et al. Fish consumption and coronary heart disease mortality in Finland, Italy, and The Netherlands. Am J Epidemiol 2000;151:999 –1006. 50. Hu FB, Bronner L, Willett WC, et al. Fish and omega-3 fatty acid intake and risk of coronary heart disease in women. JAMA 2002;287:1815–21. 51. Mozaffarian D, Lemaitre RN, Kuller LH, Burke GL, Tracy RP, Siscovick DS. Cardiac benefits of fish consumption may depend on the type of fish meal consumed: the Cardiovascular Health Study. Circulation 2003;107:1372–7. 52. Sacks FM, Stone PH, Gibson M, Silverman DI, Rosner B, Pasternak RC. Controlled trial of fish oil for regression of human coronary atherosclerosis. J Am Coll Cardiol 1995;25:1492– 8. 53. Leng GC, Lee AJ, Fowkes FG, et al. Randomized controlled trial of gamma-linolenic acid and eicosapentaenoic acid in peripheral arterial disease. Clin Nutr 1998;17:265–71. 54. Nilsen DW, Albrektsen G, Landmark K, Moen S, Aarsland T, Woie L. Effects of a high-dose concentrate of n-3 fatty acids or corn oil introduced early after an acute myocardial infarction on serum triacylglycerol and HDL cholesterol. Am J Clin Nutr 2001;74:50 – 6. 55. Marchioli R, Barzi F, Bomba E, et al. Early protection against sudden death by n-3 polyunsaturated fatty acids after myocardial infarction: time-course analysis of the results of the Gruppo Italiano per lo Studio della Sopravvivenza nell’Infarto Miocardico (GISSI)-Prevenzione. Circulation 2002;105: 1897–903. 56. U.S. Environmental Protection Agency. Exposure factors handbook. Volume II. Food ingestion factors. Washington DC: Office of Research and Development, National Center for Environmental Assessment, 1997 (EPA/ 600/P-95/002Fb; PB98-124233). 57. Connor WE. Importance of n-3 fatty acids in health and disease. Am J Clin Nutr 2000;71:171S–5S. 58. Carrington CD, Bolger MP. An exposure assessment for methylmercury from seafood for consumers in the United States. Risk Anal 2002;22: 689 –99. 59. Guallar E, Aro A, Jimenez FJ, et al. Omega-3 fatty acids in adipose tissue and risk of myocardial infarction: the EURAMIC study. Arterioscler Thromb Vasc Biol 1999;19. 60. Ahlqwist M, Bengtsson C, Lapidus L, Bergdahl IA, Shutz A. Serum mercury concentration in relation to survival, symptoms, and diseases: results from the prospective population study of women in Gothenburg, Sweden. Acta Odontol Scand 1999;57:168 –74. 61. Hallgren CG, Hallmans G, Jansson JH, et al. Markers of high fish intake are associated with decreased risk of a first myocardial infarction. Br J Nutr 2001;86:397– 404. 62. He K, Song Y, Daviglus ML, et al. Accumulated evidence on fish consumption and coronary heart disease mortality. Circulation 2004;109:2705–11.

American Journal of Preventive Medicine, Volume 29, Number 4