Consumption of fish, particularly fatty fish and fish-liver oil, has been associated with reduced risk of fatal and non-fatal CHD in numerous studies( Reference Bucher, Hengstler and Schindler 1 – Reference He, Song and Daviglus 5 ). The cardio-protective effects of fish have largely been attributed to the actions of n-3 long-chain PUFA (LC-PUFA), both EPA and DHA, mainly found in fatty fish. Consequently, numerous national health agencies and organizations have published recommendations of one or two meals of fish per week in order to lower CHD risk( Reference Harris, Kris-Etherton and Harris 2 , Reference Kris-Etherton, Harris and Appel 4 ).
Less is known of the importance of early-life fish and n-3 LC-PUFA intake for prevention of CHD later in life. A study on children’s diet from 1939 did not observe an association between fish consumption in early life and CHD risk later in life( Reference Ness, Maynard and Frankel 6 ). Yet, a Danish study on young women (mean age at baseline was 29·9 years) with median follow-up of 8 years showed that almost no fish and n-3 LC-PUFA intake was associated with increased risk of CHD( Reference Strom, Halldorsson and Mortensen 7 ).
Some risk factors for CHD may develop in young age, implying that early-life exposures may have an impact on CHD risk in later life. For example, studies of fatty streaks in children and young adults have demonstrated that atherosclerosis begins early in life and there is a correlation between the number of CHD risk factors and severity of asymptomatic coronary and aortic atherosclerosis later in life( Reference Berenson, Srinivasan and Bao 8 ). Moreover, little is known about the importance of lean fish consumption for cardiovascular risk.
Nested in an Icelandic population, known for its high intake of lean fish and common use of fish-liver oil( Reference Sekikawa, Steingrimsdottir and Ueshima 9 ), the present study aimed to assess the impact of high lean fish consumption and cod-liver oil intake in adolescence and midlife on the risk of CHD in older women.
Methods
Study design and population
The present study is a prevalence association study nested within the Age, Gene/Environment Susceptibility (AGES)-Reykjavik cohort study of the Icelandic Heart Association( Reference Harris, Launer and Eiriksdottir 10 ). The AGES-Reykjavik Study is a follow-up of the Reykjavik Study, initiated in 1967, with a response rate of 71 %. All men and women born in 1907–1935 and residing in Reykjavik and nearby communities in 1967 were selected; from the 27 281 invited to participate, 19 381 attended( Reference Bjornsson, Bjornsson and Davidsson 11 ). Of the 11 549 cohort members still alive in 2002 when AGES-Reykjavik Study examinations began, 8030 individuals were randomly chosen and invited to the study. From these, 5764 (58 % women) individuals had participated in the AGES-Reykjavik Study by 2006 (71·8 %). The women were 66–96 years old at the time of examinations, average age 76 years. Extensive data were collected in the AGES-Reykjavik Study during clinical examinations, including data on food intake in adolescence, midlife and at present old age( Reference Harris, Launer and Eiriksdottir 10 ). For our analysis we used data from the first clinical examination of the 3326 women who participated, aged 66–96 years (see Fig. 1).
Fig. 1 Selection of participants, originally from the Reykjavik Study and later, the Age, Gene/Environment Susceptibility (AGES)-Reykjavik Study, Iceland, 1967–2006
Assessment of fish and fish-liver oil consumption
Using a validated FFQ designed for the present project( Reference Eysteinsdottir, Gunnarsdottir and Thorsdottir 12 , Reference Eysteinsdottir, Thorsdottir and Gunnarsdottir 13 ), the participants answered questions on current diet, midlife diet (between the ages of 40 and 50 years) and adolescent diet (between the ages of 14 and 19 years). Two questions on fish consumption in the FFQ were used for the analysis. One concerned frequency of fish as a main meal and the other frequency of fish as a topping on bread and in salad. Possible response categories were: (i) ‘never’; (ii) ‘less than once a week’; (iii) ‘1–2 times a week’; (iv) ‘3–4 times a week’; (v) ‘5–6 times a week’; (vi) ‘daily’; and (vii) ‘more than once a day’. Total fish consumption was estimated by combining the two questions on fish consumption into one variable, giving the amount of fish for bread topping a relative value of 40/150 of that for fish as main meal. The relative value was obtained from average portion sizes from Icelandic national nutrition surveys, 40 g of fish for bread toppings and 150 g for fish as main meal. Numerical values for portions of fish were thus calculated, adding the frequency of fish as bread toppings and as main meals. Total fish consumption was divided into three groups: high (>4 portions/week), moderate (2–4 portions/week) and low (<2 portions/week). The FFQ did not include questions on fish type, but cod and haddock were the predominant fish species in the early 20th century and also today, accounting for 80–85 % of total fish consumption in Iceland( Reference Gunnarsdottir, Gunnarsdottir and Steingrimsdottir 14 – Reference Sigurjonsson 17 ). Intake of salted and smoked fish was evaluated in the FFQ, using a separate question. Salted and smoked fish was included in the total fish consumption. Information on salted and smoked fish intake for each time period was used in adjustment analysis.
Frequency of intake of fish-liver oil (cod liver) supplements (liquid or capsules), hereafter referred to as fish-liver oil, was assessed for each period of life, using the same response categories as for fish, omitting the last option of ‘more than once a day’. Responses were further categorized as seen in Table 1. Cod-liver oil is traditionally a common supplement in Iceland( Reference Steingrimsdottir, Thorgeirsdottir and Olafsdottir 16 ) and the common dose is 10 ml/d. According to the Icelandic food composition database, the amount of EPA is 0·75 g in 10 ml of cod-liver oil and the amount of DHA is 1·0 g/10 ml( 18 ).
Table 1 Characteristics of women aged 66–96 years with and without CHDFootnote *, Age, Gene/Environment Susceptibility (AGES)-Reykjavik Study, Iceland, 2006 (n 2966)
SBP, systolic blood pressure; DBP. diastolic blood pressure.
* Due to missing cases, the number of women included in the table varies from 2761 to 2966.
† P values are based on the χ 2 test, except for age, where the independent-sample t test was used.
‡ Status at entry to the AGES-Reykjavik Study.
§ Status at entry to the Reykjavik Study.
|| Data from FFQ on midlife and adolescent diet from the AGES-Reykjavik Study.
Validation of the FFQ
For midlife dietary habits, retrospective food consumption of 56–72-year-old participants (107 women) was estimated by comparing the results in the AGES-FFQ with detailed dietary data (an hour-long interview on dietary habits in the past 3 months) gathered from the same individuals 18–19 years previously in a 1990 national nutrition survey. The strongest correlation was found for fish-liver oil (r=0·56, P<0·001) while the correlation coefficient for fish consumption was 0·28 (P=0·004)( Reference Eysteinsdottir, Gunnarsdottir and Thorsdottir 12 ). Validity of the early-life dietary assessment has not been, and cannot be, investigated. Yet, the participants provided information on early-life residency and the dietary data importantly show similar residency-dependent variation in dietary habits as documented in contemporary studies by Sigurjonsson( Reference Sigurjonsson 17 ). For example, in our data we can see that those who reported early-life residency in coastal villages also reported higher fish and fish-liver oil intakes compared with those raised in rural areas or Reykjavik in the early-life period, corresponding to the results from the 1939 dietary survey.
CHD – outcome assessment
An end point for CHD was obtained from hospital records. The records were systematically reviewed according to the MONICA (Monitoring of trends and determinants in cardiovascular disease) protocol based on the incidence of myocardial infarction (MI), coronary artery bypass surgery (CABG) or percutaneous coronary intervention (PCI). Participants were defined as having a CHD event if MI, CABG or PCI had occurred at entry to AGES-Reykjavik. The quality of the registration in Iceland according to MONICA methods has previously been described elsewhere( Reference Tunstall-Pedoe, Kuulasmaa and Amouyel 19 ).
Covariate assessment
Information on midlife diabetes, total blood cholesterol, hypertension and BMI was obtained from the Reykjavik Study, as the concurrent information on these risk factors might be biased due to ageing. Diabetes was either self-reported or based on glucose value (≥126 mg/l) after overnight fasting. Total blood cholesterol was categorized into high (≥6 mmol/l) and low (<6 mmol/l). All blood chemistry was measured after overnight fasting. Blood pressure was measured with a mercury sphygmomanometer and the mean value of two blood pressure measurements from separate occasions used. Blood pressure measurements were divided into no hypertension, pre-hypertension and hypertension. BMI was categorized by obesity status (see Table 1).
From the AGES-Reykjavik Study we retrieved information on age at entry, educational level, smoking status, family history of heart disease, physical activity in adolescence and midlife, alcohol consumption in midlife and dietary habits in youth and midlife. Family history of coronary disease was recorded if father, mother, siblings or children had coronary thrombosis. Information on self-reported frequency of moderate or vigorous physical activity in three time periods (20–34 years, 35–49 years and 50–65 years of age) was pooled and categorized into never, rarely, occasionally, moderately and often. Dietary covariates were consumption of fruit, vegetables and amount of spread used on bread. For fruit and vegetables there were seven responses possible; the same as described earlier for fish. There were four response options for usual amount of spread used on bread, with pictures showing buttered bread with the respective amounts. In Iceland, butter was the predominant spread used on bread during the study period.
Statistical analyses
We excluded 360 women from the analysis due to missing data regarding heart disease, fish and/or fish-liver oil consumption, leaving 2966 women in our analysis.
We used χ 2 tests to estimate different characteristics of women with and without CHD, of fish and fish-liver oil consumers in midlife, and of fish and fish-liver oil consumers in adolescence. P values ≤0·05 were considered statistically significant. We used logistic regression to calculate odds ratios and 95 % confidence intervals of CHD by differential fish and fish-liver oil consumption, for midlife and adolescent consumption separately.
The first multivariable model was adjusted for age (as a continuous variable) at entry to AGES-Reykjavik. The second model was adjusted for age, education, smoking status, physical activity, alcohol consumption, fish consumption (for the outcome of the fish-liver oil analysis) and fish-liver oil consumption (for the fish analysis). The third model was adjusted for the same variables as in the second model, adding vegetable and fruit consumption, as well as the amount of spread used on bread (see categories in Table 1). The categories for fruit and vegetables in adolescence are different from the midlife categories due to different consumption pattern and availability of foods. In the fourth model additional adjustments were made for diabetes, blood pressure, total blood cholesterol and BMI in midlife. This was done only for midlife consumption since these covariates were not available for adolescence.
The statistical software package PASW version 18, release version 18·00 (2009) was used for all statistical analyses.
Results
Out of 2966 women, 234 (7·9 %) had previously known CHD at entry into the AGES-Reykjavik Study, while 2732 did not have CHD according to MONICA registration. Major characteristics of women with and without CHD are presented in Table 1.
Table 2 presents multivariable analysis results for the association between CHD and fish and fish-liver oil consumption in adolescence. No statistically significant association was observed between fish consumption and CHD (>2 portions/week v. ≤2 portions/week), while all three models showed a significant protective association between adolescent fish-liver oil intake and CHD. Compared with women with no intake of fish liver oil, those who consumed fish-liver oil at least three times weekly had an OR of 0·61 (95 % CI 0·45, 0·81) for CHD in model 1, 0·62 (95 % CI 0·45, 0·84) in model 2 and 0·62 (95 % CI 0·45, 0·85) in model 3.
Table 2 Odds ratio estimates (and 95 % confidence intervals) for CHD in women aged 66–96 years by fish and fish-liver oil consumption in adolescence, Age, Gene/Environment Susceptibility (AGES)-Reykjavik Study, Iceland, 2006 (n 2966)
Ref., reference category.
* Adjustments were made for age, education, smoking status, physical activity, alcohol consumption, family history of heart disease, fish consumption (for fish-liver oil) and fish-liver oil consumption (for fish).
† Additional adjustments were made for consumption of fruit and vegetables and amount of spread used in adolescence.
Table 3 presents multivariable analysis results for fish and fish-liver oil consumption in midlife. No statistically significant association was observed between fish consumption (>2 portions/week v. ≤2 portions/week) and CHD, while midlife fish-liver oil consumption was associated with decreased risk of CHD. Using women with no intake of fish-liver oil as a reference group, those with fish-liver oil consumption of at least three times weekly had an OR of 0·63 (95 % CI 0·48, 0·84) to develop CHD in model 1, 0·67 (95 % CI 0·49, 0·91) in model 2, 0·68 (95 % CI 0·50, 0·92) in model 3 and 0·68 (95 % CI 0·50, 0·94) in model 4. Additional adjustment for salted and smoked fish consumption did not alter our results (data not shown). Testing for interaction between fish and salted and smoked fish yielded a P value of 0·11 for the adolescent period and 0·87 for the midlife period. No significant association was found between fish intake and CHD when stratified by high and low intake of salted and smoked fish (see online supplementary material, Supplemental Table 1).
Table 3 Odds ratio estimates (and 95 % confidence intervals) for CHD in women aged 66–96 years by fish and fish-liver oil consumption in midlife, Age, Gene/Environment Susceptibility (AGES)-Reykjavik Study, Iceland, 2006 (n 2966)
Ref., reference category.
* Adjustments were made for age, education, smoking status, physical activity, alcohol consumption, family history of heart disease, fish consumption (for fish-liver oil) and fish-liver oil consumption (for fish).
† Additional adjustments made for consumption of fruit and vegetables and the amount of spread used in midlife.
‡ Additional adjustments were made for cholesterol level, diabetes, hypertension and BMI, all measured in midlife.
Pooling fish-liver oil consumption in both midlife and adolescence showed that frequent (three times weekly or more) intake of fish-liver oil in both adolescence and midlife was associated with lower risk of CHD compared with two times weekly or less in both life periods (OR=0·60; 95 % CI 0·43, 0·84). No significant association was found in the group with infrequent intake in adolescence but high frequency in midlife or vice versa (Table 4). Adjustments were made for age, education smoking status and alcohol consumption in midlife, physical activity in midlife and early adulthood, and family history of heart disease. Sixty-seven per cent of the participants had high intake of fish-liver oil in both adolescence and midlife.
Table 4 Odds ratio estimates (and 95 % confidence intervals) for CHD in women aged 66–96 years by longitudinal fish-liver oil consumption, Age, Gene/Environment Susceptibility (AGES)-Reykjavik Study, Iceland, 2006 (n 2966)
Ref., reference category.
Low, fish-liver oil consumption two times weekly or less; high, fish-liver oil consumption three times weekly or more.
* Adjustments made for age, smoking, education, family history of heart disease and alcohol consumption.
Discussion
In the present study in a female population with high consumption of lean fish and consistent use of fish-liver oil, we found clear support for the hypothesis that fish-liver oil consumption reduces the risk of CHD in women aged 66–96 years. Our data did not show any protection by fish consumption against CHD, neither during the adolescent period nor in midlife.
Our findings of a reduced risk of CHD in women reporting prolonged consumption of fish-liver oil starting in early life may be explained by several factors. Studies have shown that risk factors for CHD such as dyslipidaemia, hypertension, obesity and insulin resistance can develop early and affect formation of atherosclerosis, and therefore increase CHD risk( Reference Berenson, Srinivasan and Bao 8 , Reference Mahoney, Burns and Stanford 20 , Reference Magnussen, Niinikoski and Juonala 21 ). Fish-liver oil is rich in n-3 LC-PUFA, the physiological effects of which can be both long term and short term. n-3 LC-PUFA’s lowering effect on plasma TAG concentration, blood pressure and platelet aggregation might require months or years before clinical outcomes are evident( Reference Mozaffarian 22 ). Our estimates appear to be independent of major risk factors for CHD and stratifying the data by cholesterol (high/low), diabetes (yes/no) and hypertension (yes/no/pre-hypertension) did not reveal any differential association (data not shown). However, fish-liver oil is not only rich in n-3 LC-PUFA but also vitamin D. Although the effect of vitamin D on CHD is not clear, its effect, or the effects of other nutrients found in fish-liver oil, cannot be excluded as confounding factors in our analysis( Reference Anderson, May and Horne 23 , Reference Hsia, Heiss and Ren 24 ).
Not finding the putative protective effect of fish consumption on CHD risk in our study population may be explained by several factors. In this population with high fish consumption, there is a lack of a reference group of women consuming fish seldom or never. In our study, the lowest fish intake category that could be used for reference is defined as those who consumed fish twice or less per week and within that category only ten participants never consumed fish in adolescence and four participants never consumed fish in midlife. Studies have shown 20–36 % reduced risk of CHD death when two fish meals are consumed per week compared with no fish or less than once weekly consumption( Reference Mozaffarian and Rimm 3 , Reference He, Song and Daviglus 5 ), while no benefit has been reported by further increase in fish consumption( Reference Mozaffarian and Rimm 3 ). Similarly a Danish study on young women only found increased risk for CVD among women who consumed between 0 and 3 g fish/d( Reference Strom, Halldorsson and Mortensen 7 ). A beneficial threshold level for fish consumption might thus already have been reached by our reference group. On the other hand, haddock and cod, the most common fish consumed in Iceland during the 20th century( Reference Gunnarsdottir, Gunnarsdottir and Steingrimsdottir 14 – Reference Sigurjonsson 17 ), are lean species containing only small amounts of n-3 LC-PUFA( Reference Gunnarsdottir, Tomasson and Kiely 25 ). Lean fish has not consistently been associated with protection against CHD( Reference Kromhout, Bosschieter and de Lezenne Coulander 26 – Reference Mozaffarian, Lemaitre and Kuller 29 ), suggesting that n-3 LC-PUFA or other fat-soluble substances in fish may be the important factor for this effect. Interestingly, cod-liver oil contributes 42 % of total intake of n-3 LC-PUFA in the present diet of Icelanders, compared with 40 % from fish and seafood( Reference Sekikawa, Steingrimsdottir and Ueshima 9 ). Also, information on cooking methods and condiment use are not available. Stick margarine, high in both saturated and trans-fatty acids, was commonly used with fish in the years 1980–2000 approximately( Reference Steingrimsdottir, Thorgeirsdottir and Aegisdottir 15 ), possibly masking any putative benefit of fish consumption in midlife. Alternatively, the presence of environmental pollutants such as mercury or polychlorinated biphenyls in fish cannot be excluded, counteracting a possible benefit of fish intake. This, however, remains an unlikely explanation since insignificant levels of these pollutants have been measured in the Atlantic cod fish, including haddock( Reference Mozaffarian and Rimm 3 ), the most common fish type consumed in Iceland during the exposure windows of our study. Also, salted fish was a significant proportion of fish intake in Iceland in former times, which could possibly affect our results. Adjustment for intake of salted fish in the present study did not change our results. Further, doing a stratified analysis for salted fish consumption revealed no significant outcome. However, women with high consumption of salted fish who also reported to consume more than four portions of fish weekly (with at least one portion being salted fish) showed an insignificant positive association with CHD, seen both for midlife and the adolescence period. Interestingly, women with low consumption of salted fish who reported to consume more than two portions fish weekly showed an insignificant inverse association in adolescence. This suggests that the proportion of salted fish in the total fish consumption might be of great importance in the present study and could be one of the reasons why no association was found between high total fish consumption and CHD in our analysis. Another possible reason for not finding an association with fish intake is the low validity of the retrospective question of fish intake in midlife (0·28, P=0·004)( Reference Eysteinsdottir, Gunnarsdottir and Thorsdottir 12 ). In spite of being considered acceptable, the validity may not be sufficient to detect a putative beneficial association.
To our knowledge, few studies are available on early-life dietary factors and CHD risk in later life. The Boyd Orr cohort study, a prospective study of English children’s diet from 1939, did not find any beneficial relationship between diet in childhood, including fish, and CHD later in life( Reference Ness, Maynard and Frankel 6 ). The reference group in that study also had a relatively low percentage of people consuming low amounts of fish and the participants could therefore, as in our study, have reached a potential beneficial threshold of fish intake. A more likely explanation, however, may be that the estimated consumption in childhood was based on household, rather than individual consumption.
The strength of our study includes the detailed and valid assessment of outcome as well as access to a wide selection of covariates. The fact that our results on fish-liver oil are robust through multiple adjustments suggests that our findings may indeed be valid, especially considering that Icelandic women are a low-risk population for CVD( Reference Aspelund, Thorgeirsson and Sigurdsson 30 ). Few studies have been able to provide data on early-life diet combined with detailed ascertainment of CHD outcomes later in life and we believe our study is unique in that aspect. Also long-term use of cod-liver oil has not been studied previously to our knowledge.
Our study is likely to suffer somewhat from recall bias since the women were asked to recall their diet many decades earlier. However, a previous US-based study showed that food-related memory from childhood over four decades later can be as accurate as from current diet, especially for food items eaten rarely or daily( Reference Dwyer and Coleman 31 ). Our own data show that the validity of the questions on fish-liver oil and fish consumption was similar for midlife consumption and recent consumption, r=0·56 (P<0·001) and 0·28 (P=0·004), respectively( Reference Eysteinsdottir, Gunnarsdottir and Thorsdottir 12 ). Additionally, despite the rich covariate selection and multiple adjustments, the influence of unmeasured confounders cannot be excluded.
The Icelandic Heart Association has a long reputation of valid end-point assessment of CHD. Nevertheless, we cannot exclude that the non-case group might include women with unrecognized non-fatal MI. However, additional analysis where the end point was based on electrocardiogram, responses to Rose Angina questionnaire and questions concerning previous heart procedures at entry revealed similar results (data not shown).
Conclusion
With few existing studies on early-life dietary factors and CHD, our study provides important evidence for the potential preventive role of fish-liver oil consumption throughout life on the development of CHD in women. Our results suggest that moderate prolonged fish-liver oil consumption initiated in early life may be protective against the development of CHD in women. If confirmed in future studies, preferably with prospective ascertainment of fish-liver oil consumption, these findings may have significant public health implications.
Acknowledgements
Financial support: This present analysis received no specific grant from any funding agency in the public, commercial or not-for-profit sectors. The AGES-Reykjavik Study was funded by the National Institutes of Health (contract number N01-AG-12100); the Intramural Research Program of the National Institute on Aging; the Icelandic Heart Association; and the Icelandic Parliament. The funding agencies (National Institute on Aging, Icelandic Heart Association and Icelandic Parlament) for the AGES-Reykjavik Study had no role in the design, analysis or writing of this article. Conflict of interest: None. Authorship: A.H., J.E.T., U.A.V. and L.S. were responsible for formulating the research questions and the design of the study. A.H. and J.E.T. were responsible for analysing the data. All authors were responsible for writing the article. Ethics of human subject participation: This study was approved by the Icelandic National Bioethics Committee (VSN: 00-063) and the Institutional Review Board covering research for the National Institute on Aging, USA.
Supplementary material
To view supplementary material for this article, please visit http://dx.doi.org/10.1017/S1368980015001020
