Hostname: page-component-7c8c6479df-ph5wq Total loading time: 0 Render date: 2024-03-28T14:53:07.430Z Has data issue: false hasContentIssue false

Fish consumption in relation to other foods in the diet

Published online by Cambridge University Press:  08 June 2011

Anu W. Turunen*
Affiliation:
National Institute for Health and Welfare, Department of Environmental Health, Neulaniementie 4, FI-70210Kuopio, Finland
Satu Männistö
Affiliation:
National Institute for Health and Welfare, Department of Chronic Disease Prevention, Mannerheimintie 164a, FI-00300Helsinki, Finland
Anna L. Suominen
Affiliation:
National Institute for Health and Welfare, Department of Environmental Health, Neulaniementie 4, FI-70210Kuopio, Finland University of Turku, Institute of Dentistry, Lemminkäisenkatu 2, FI-20520Turku, Finland
Pekka Tiittanen
Affiliation:
National Institute for Health and Welfare, Department of Environmental Health, Neulaniementie 4, FI-70210Kuopio, Finland
Pia K. Verkasalo
Affiliation:
National Institute for Health and Welfare, Department of Environmental Health, Neulaniementie 4, FI-70210Kuopio, Finland
*
*Corresponding author: Anu W. Turunen, fax +358 20 610 6499, email anu.turunen@thl.fi
Rights & Permissions [Opens in a new window]

Abstract

Our aim was to investigate whether fish consumption is associated with the consumption of other healthy foods. The study population consisted of 2605 men and 3199 women from the nationally representative Health 2000 survey and 114 professional fishermen and 114 fishermen's wives (the Fishermen substudy) in Finland. Dietary data were collected using a calibrated (i.e. determined to have relative validity) FFQ. Model-adjusted means for food consumption and P values for linear trend were calculated across fish consumption tertiles. Those with the highest fish consumption had the highest consumption of vegetables, fruit and berries, potatoes, oil and wine even after adjusting for other food groups. The consumption of red meat and sausages had a tendency to decrease across fish consumption tertiles but the associations were inconsistent in the study populations. In conclusion, fish consumption had a positive linear association with the consumption of some other healthy foods such as vegetables, fruit, berries, and oil both in the general population of Finland and in a population with high fish consumption. Additional adjustment for other food groups had a clear effect on some of the studied associations. Therefore, when evaluating the health effects of fish consumption, confounding by other foods characterising a healthy diet needs to be considered.

Type
Full Papers
Copyright
Copyright © The Authors 2011

According to official nutrition recommendations, it is advisable to eat fish at least twice per week(Reference Lichtenstein, Appel and Brands1, 2). Fish consumption and long-chain PUFA (n-3 PUFA) intake are thought to be important for human health(Reference Riediger, Othman and Suh3Reference Calder and Yaqoob5), and to protect especially from CVD(Reference Mozaffarian6, Reference He7), diabetes(Reference Djoussé, Gaziano and Buring8Reference Nettleton and Katz10) and possibly some cancers(Reference Gonzalez and Riboli11, Reference Wolk12) although the conclusions seem partly contradictory(Reference Kaushik, Mozaffarian and Spiegelman13, Reference Hooper, Thompson and Harrison14). Despite the large body of evidence, it has been hypothesised that at least a part of the observed beneficial effects of fish consumption could be explained by an overall healthy diet, and fish consumption may even be a surrogate for healthy lifestyle in general(Reference He7, Reference Suominen-Taipale, Partonen and Turunen15). This implies that the postulated health benefits of fish consumption would not be achieved by eating fish alone.

Many epidemiological studies on the associations between fish consumption and chronic diseases such as CVD(Reference Ascherio, Rimm and Stampfer16Reference Mozaffarian, Lemaitre and Kuller19) including cerebrovascular diseases(Reference Montonen, Jarvinen and Reunanen20Reference Iso, Rexrode and Stampfer23) and cardiovascular mortality(Reference Hu, Cho and Rexrode24Reference Albert, Hennekens and O'Donnell27), diabetes(Reference Patel, Sharp and Luben9) and cancer(Reference Sugawara, Kuriyama and Kakizaki28Reference Chavarro, Stampfer and Hall32) have reported associations between fish and other foods as baseline characteristics to describe their study populations. These studies have provided descriptive data on, for example, the association between the consumption of fish and vegetables but have mostly not accounted for the effects of confounding factors such as lifestyle and other foods in the diet. To the best of our knowledge, the specific association between fish consumption and the overall diet has not been the main research question in previous studies, and has thus not gained the attention it deserves to provide strong evidence for dietary recommendations.

In the present study, we wanted to investigate whether fish consumption is associated with the consumption of some other foods, especially those that are considered healthy based on official nutrition recommendations. In other words, we wanted to explore if there are some common features in the diets of those who eat a lot of fish. The analyses were conducted in a unique population with high fish consumption where the effects of fish could be most easily seen. In addition, the analyses were repeated in a large sample of the general population of Finland.

Methods

Study populations

The nationally representative Health 2000 health examination survey (the Health 2000 survey) was conducted during 2001–2002 and coordinated by the National Institute for Health and Welfare in Finland (THL, merged from the former National Public Health Institute (KTL) and the National Research and Development Centre for Welfare and Health (STAKES))(Reference Aromaa and Koskinen33). The main study was carried out in a population aged 30 years or over, and it included an interview, several questionnaires and a health examination. A total of 5998 participants completed an FFQ(Reference Heistaro34), and 2605 men and 3199 women had all the required data for the present study.

The Nutrition, Environment and Health study (the Fishermen study) was conducted during 2004–2005 and coordinated by THL. A total of 1427 professional fishermen, their wives and other family members answered a self-administered health questionnaire. Of those, 309 volunteers, aged 22–74 years, and living near Helsinki and Turku study centres participated in a health examination study (the Fishermen substudy) including, for example, blood sampling, basic measurements and an FFQ(Reference Turunen, Verkasalo and Kiviranta35). Of those, 114 professional fishermen and 114 fishermen's wives had all the required data for the present study. In our previous study, we showed that fish consumption among the fishermen and their wives was approximately 1·5-fold when compared with that of the general population(Reference Turunen, Verkasalo and Kiviranta35), which attests our hypothesis of a population with high fish consumption.

The present study was conducted according to the guidelines laid down in the Declaration of Helsinki, and the protocols of the Health 2000 survey and the Fishermen study were approved by the ethical committee of the Hospital District of Helsinki and Uusimaa. A written informed consent was obtained from all participants.

Dietary data

In both studies, diet was assessed by the same calibrated (i.e. determined to have relative validity) self-administered FFQ designed to cover the whole diet over the past 12 months(Reference Männistö, Virtanen and Mikkonen36, Reference Paalanen, Männistö and Virtanen37). The FFQ consisted of 128 commonly used and/or nutritionally important food items, mixed dishes and alcoholic beverages based on the national FINDIET study(Reference Paturi, Tapanainen and Reinivuo38). Serving sizes were specified by natural units (for example, serving, slice, glass, cup) or weight/volume measures, and the nine response options ranged from ‘never or seldom’ to ‘six or more times per day’ (for more information, see the Appendix in our earlier publication(Reference Turunen, Mannisto and Kiviranta39)). Daily consumption of foods (g/d) and the intake of energy (MJ/d) and nutrients were calculated by the national Fineli® Finnish Food Composition Database(Reference Reinivuo, Hirvonen and Ovaskainen40). For the purpose of the present study, food items and mixed dishes were combined into twenty food groups based on culinary use, nutrient profile and nutritional relevance (see Supplementary Appendix; available online at http://www.journals.cambridge.org/bjn).

Other covariates

Age (years) was calculated at the time of sampling in the Health 2000 survey and at the time of the health examination in the Fishermen substudy.

Data on education, marital status and smoking were obtained from a structured interview in the Health 2000 survey and from the self-administered health questionnaire in the Fishermen substudy. With regard to education, two questions were used in the Health 2000 survey: ‘What is your basic education?’ with eight response options from ‘less than elementary school’ to ‘high school’ and ‘What is your highest education after basic education?’ with eleven response options from ‘no vocational education’ to ‘doctoral degree’. In the Fishermen substudy, education level was determined by the question ‘What is your education?’ with eight response options from ‘less than elementary school’ to ‘academic degree’. The final education variable was constructed similarly in both studies, and it consisted of three classes: basic, intermediate and high education. The final marital status consisted of three classes: married or cohabiting, unmarried, and divorced/separated or widowed. Regarding smoking, the following questions were asked: ‘Have you ever smoked?’, ‘Have you smoked at least 100 times?’, ‘Have you ever smoked regularly (i.e. daily for at least 1 year)?’ and ‘When did you last smoke?’. The final smoking variable consisted of three classes: never, occasional or ex-, and daily smoker.

Data on physical activity were obtained from the self-administered health questionnaire both in the Health 2000 survey and in the Fishermen substudy. Free-time physical activity was determined by the question ‘How often do you exercise in your free time so that the duration is at least half an hour and you get at least mildly out of breath?’ with response options ranging from ‘daily’ to ‘a couple of times per year or less often’. Physical activity while commuting was determined by the question ‘How many minutes do you walk or cycle while going to work?’ with response options ranging from ‘not at all’ to ‘2 hours per day or more’. The final physical activity variable consisted of three classes: sufficient, intermediate and sedentary.

Weight (kg) and height (cm) were measured during the health examination by trained research personnel. Weight in light clothing was measured to an accuracy of 0·1 kg using a bioimpedance device or digital scales in the Health 2000 survey, and digital scales in the Fishermen substudy. Height was recorded using a wall-mounted stadiometer to an accuracy of 0·5 cm.

Statistical analyses

Statistical analyses to produce Tables 1–3 were performed using the SAS statistical software package (version 9.2; SAS Institute Inc., Cary, NC, USA). SAS survey procedures were used to account for the sampling design of the Health 2000 survey data. In addition, a post-stratification weight was used to adjust for the oversampling of the 80-year-old and older individuals, and for the non-response to the FFQ(Reference Djerf, Laiho, Kehtonen and Heistaro41). For the Fishermen substudy, basic SAS procedures were used. For the categorical variables in Table 1, however, the SAS SURVEYFREQ procedure was used also for the Fishermen substudy since the basic SAS FREQ procedure does not yield 95 % CI for the multinomial proportions.

Table 1 Background data of the Health 2000 survey and the Fishermen substudy participants

(Mean values or percentages with 95 % confidence intervals)

Table 2 Model-adjusted food consumption (g/d) by fish consumption tertiles (g/d) among the Health 2000 survey men and women

(Geometric means and 95 % confidence intervals)

* n 868, geometric mean 14, range 0–29 g/d.

n 868, geometric mean 38, range 30–48 g/d.

n 869, geometric mean 75, range 49–557 g/d.

§ n 1065, geometric mean 13, range 0–29 g/d.

n 1068, geometric mean 37, range 30–46 g/d.

n 1066, geometric mean 72, range 46–561 g/d.

** Model 1: age and total energy intake.

†† Model 2: age, BMI, education, marital status, smoking, physical activity and other food groups.

Table 3 Model-adjusted food consumption (g/d) by fish consumption tertiles (g/d) among the fishermen and the fishermen's wives

(Geometric means and 95 % confidence intervals)

* n 38, geometric mean 36, range 10–49 g/d.

n 38, geometric mean 69, range 50–93 g/d.

n 38, geometric mean 137, range 94–463 g/d.

§ n 38, geometric mean 27, range 0–42 g/d.

n 38, geometric mean 55, range 43–72 g/d.

n 38, geometric mean 102, range 74–282 g/d.

** Model 1: age and total energy intake.

†† Model 2: age, BMI, education, marital status, smoking, physical activity and other food groups.

The SAS GLM procedure was used to produce model-adjusted consumption of different food groups by fish consumption tertiles for Tables 2 and 3. We used three different adjustments: age and total energy intake (model 1); age, total energy intake and lifestyle factors (BMI, education, marital status, smoking and physical activity) (data not shown); and age, lifestyle factors and the consumption of eighteen other food groups (model 2). Total energy intake was omitted from model 2 due to multicollinearity caused by simultaneous inclusion of energy and all food groups. Energy and continuous food group variables were transformed according to natural logarithm log(x+1), which clearly improved normality of the variable distributions and fulfilled the model assumptions. Antilogarithms were taken from the arithmetic means of the log-transformed variables, and the resulting geometric means and their 95 % CI were reported. In addition, P values for linear trends across fish consumption tertiles were calculated. The coefficients for linear contrasts needed for the calculation of these P values were produced by SAS/IML software(42) because geometric means for the fish consumption tertiles were not equally spaced. All the results were reported separately for the sexes and the two studies. To check for multicollinearity, pairwise Pearson correlation coefficients and collinearity diagnostics, namely tolerance, variance inflation factor and condition index, were calculated.

As sensitivity analyses, we applied an additive model with thin-plate regression spline in the multiple generalised cross-validation package (mgcv) for R Statistical Software version 2.9.1(Reference Wood43, 44). An additive model is a non-parametric extension of a linear model for Gaussian response and allows the data to ‘speak for themselves’ because a smoothing function does not assume a rigid form for the dependence. It can be used to explore the relationships between the dependent variable and the independent variables, for example, to visually assess linearity of the studied associations. In the sensitivity analyses of the present study, each food group was treated as a response variable (at a time), continuous fish consumption as a smoothed predictor, and all other covariates (as in model 2) as parametric predictors. As a result of sensitivity analyses, scatter plots with regression curves and approximate 95 % CI were drawn, and the most important observed associations were visualised in Figs 1 and 2. In the figures, the plotted points are partial residuals, the solid curve is the additive model fit, and the dashed curves represent the approximate 95 % CI. The fit is named as s(log_fish, edf), where edf is the estimated degrees of freedom describing the wiggliness of the fit. When edf for the smooth is close to 1, the curve fits to a straight line. However, the CI typically becomes wider (and the uncertainty increases) towards the ends of the curve due to decreasing number of observations, and, therefore, only the central part of the curve is usually reliable.

Fig. 1 Adjusted smoothed associations between fish consumption and the consumption of selected foods (vegetables, (a) and (b); fruit and berries, (c) and (d)) among the Health 2000 survey men (n 2605; (a) and (c)) and women (n 3199; (b) and (d)). Associations were adjusted for age, BMI, education, marital status, smoking, physical activity and other food groups, and were produced by an additive model with a thin-plate regression spline. All the FFQ variables were transformed according to log(x+1). The solid curve is the additive model fit and the dashed curves represent the approximate 95 % CI. The fit is named as s(log_fish, edf), where edf is the estimated degrees of freedom describing the wiggliness of the fit. The plotted points are partial residuals.

Fig. 2 Adjusted smoothed associations between fish consumption and the consumption of selected foods (red meat, (a) and (b); sausages, (c) and (d)) among the Health 2000 survey men (n 2605; (a) and (c)) and women (n 3199; (b) and (d)). Associations were adjusted for age, BMI, education, marital status, smoking, physical activity and other food groups, and were produced by an additive model with a thin-plate regression spline. All the FFQ variables were transformed according to log(x+1). The solid curve is the additive model fit and the dashed curves represent the approximate 95 % CI. The fit is named as s(log_fish, edf), where edf is the estimated degrees of freedom describing the wiggliness of the fit. The plotted points are partial residuals.

Results

The fishermen were, on average, 4 years older and had a higher proportion of individuals having only basic education when compared with the general population men in the Health 2000 survey (Table 1). The proportion of daily smokers was lower among the fishermen and their wives than in the general population. The geometric mean for fish consumption was over twofold among the fishermen when compared with the general population men, and 1·6-fold among the fishermen's wives when compared with the general population women.

Both in the general population and among the fishermen and their wives, those with the highest fish consumption had the highest consumption of vegetables, fruit and berries, potatoes, oil and wine (Tables 2 and 3). All these associations were relatively consistent regardless of adjustments. Red meat and sausage consumption had a tendency to decrease across fish consumption tertiles. The inverse association between fish and red meat was more evident among the fishermen and their wives, whereas the inverse association between fish and sausages was more evident among the women in both studies. In general, adjustment for lifestyle factors in addition to age and total energy intake (data not shown) had only a minor effect on the results. An additional adjustment for other food groups attenuated the observed associations. For example, when adjusted only for age, total energy intake and lifestyle factors, those with the highest fish consumption also had the highest consumption of poultry and the lowest consumption of liquid milk products, and sugar and confectionery, but these associations practically disappeared after adjusting for other food groups.

The regression curves produced by the additive model gave reassurance that a linear trend test could be applied for the studied associations. In addition, the shapes of the curves supported the conclusions made based on the model-adjusted means. For example, the regression curves showed a clear positive association between smoothed fish consumption and the consumption of fruit and berries (Fig. 1), and a clear negative association between smoothed fish consumption and the consumption of sausages (Fig. 2). On the other hand, the regression curves showing the association between smoothed fish consumption and red meat consumption (Fig. 2) were relatively flat, and the 95 % CI appeared to include zero. The regression curves for the fishermen and their wives are not shown due to the small number of observations.

Discussion

In the present study, those with the highest fish consumption had the highest consumption of vegetables, fruit and berries, potatoes, oil and wine. These trends were essentially the same regardless of adjustments and study population. Red meat consumption had a tendency to decrease across fish consumption tertiles, especially among the fishermen and their wives, whereas sausage consumption decreased across fish consumption tertiles especially among the women in the general population and the fishermen's wives. When adjusted for lifestyle, those with the highest fish consumption had the lowest consumption of liquid milk products and sugar and confectionery in both studies but these associations practically disappeared after adjusting for other food groups.

We used a calibrated FFQ(Reference Männistö, Virtanen and Mikkonen36, Reference Paalanen, Männistö and Virtanen37) on the whole diet, which is a primary method to measure usual long-term food consumption(Reference Willett and Willett45). Although absolute food consumption was reported in the present study, it should be noted that an FFQ is designed only to rank participants according to their dietary intake. With regard to multivariate modelling, a high number of covariates in a regression model may cause instability in the estimates. In the present study, however, the correlations between the covariates were relatively low, and collinearity seemed not to be a problem. To increase the validity of the results, the analyses were performed in two study populations: in a large nationally representative population with an exceptionally high response rate and in a unique population with high fish consumption.

In the present study, fish consumption seemed to have a strong positive linear association with vegetable, fruit and berry consumption even after adjusting for other food groups. This association was seen also among the fishermen, despite the fact that among them, high fish consumption has been thought to be an occupational characteristic and to derive from tradition and easy availability of fish. Similar positive linear associations have also been seen in all the previous studies that reported age-adjusted or unadjusted means as baseline characteristics(Reference Patel, Sharp and Luben9, Reference Ascherio, Rimm and Stampfer16Reference Mozaffarian, Longstreth and Lemaitre21, Reference Iso, Rexrode and Stampfer23Reference Chavarro, Stampfer and Hall32). Additionally, fish typically loads to a prudent dietary pattern together with vegetables and fruit in dietary pattern analyses(Reference Pala, Sieri and Masala46Reference Heidemann, Schulze and Franco48).

Consuming more of one protein source usually means consuming less of some other source of protein. In the majority of previous studies, those with the highest fish consumption had the lowest meat consumption(Reference Ascherio, Rimm and Stampfer16, Reference Mozaffarian, Lemaitre and Kuller19, Reference Mozaffarian, Longstreth and Lemaitre21, Reference Iso, Rexrode and Stampfer23, Reference Hu, Cho and Rexrode24, Reference Chavarro, Stampfer and Hall32), but in some studies the direction of the association was the opposite(Reference Iso, Kobayashi and Ishihara17, Reference Sugawara, Kuriyama and Kakizaki28Reference Kobayashi, Tsubono and Otani31), possibly due to combining all types of meat (red and white) in one variable. In the present study, a negative association between fish and red meat consumption was observed but it was more distinguishable among the fishermen and their wives than in the general population. The explanation for this might be that fishermen and their wives have fish consumption high enough to replace other sources of protein in their diet. Further, there was an inverse association between fish and sausage consumption in the general population and among the fishermen's wives. Overall, this association was more distinctive among the women than among the men, which may be due to the fact that women are typically more health conscious than men, and they may prefer fish over sausages. In addition, energy intake and the total amount of food consumed are usually smaller among women and therefore fish may be able to partially replace other types of meats in their diet. With regard to poultry consumption in previous studies, those who had the highest fish consumption had the highest consumption of poultry(Reference Iso, Kobayashi and Ishihara17, Reference Iso, Rexrode and Stampfer23, Reference Albert, Hennekens and O'Donnell27, Reference Engeset, Andersen and Hjartaker29, Reference Albert, Campos and Stampfer49). In the present study, this association was seen only in the general population, although the positive linear association practically disappeared especially among the women after adjusting for other food groups.

In some previous studies, those with the highest fish consumption had the lowest consumption of dairy products(Reference Ascherio, Rimm and Stampfer16, Reference Albert, Hennekens and O'Donnell27, Reference Chavarro, Stampfer and Hall32), but in some studies, the direction of the association was the opposite(Reference Iso, Kobayashi and Ishihara17, Reference Iso, Rexrode and Stampfer23). In only one previous study, the consumption of sweets was reported by fish consumption groups, and the association seemed to be negative(Reference Ascherio, Rimm and Stampfer16). In the present study, there was a negative linear association between fish consumption and the consumption of liquid milk products and sugar and confectionery when adjusted only for lifestyle factors but the association practically disappeared after adjusting for other food groups.

Alcohol consumption was typically the highest among those who had the highest fish consumption in some(Reference Patel, Sharp and Luben9, Reference Oomen, Feskens and Rasanen26, Reference Wolk, Larsson and Johansson30, Reference Chavarro, Stampfer and Hall32) but not all(Reference Anderson, Nettleton and Herrington18, Reference Mozaffarian, Lemaitre and Kuller19) previous studies. The positive linear association was especially evident among the Italian men(Reference Oomen, Feskens and Rasanen26). Similarly, in the present study, the more the general population consumed fish, the higher was their wine and spirit consumption. A positive linear association between fish and wine consumption was also seen among the fishermen and their wives. This may be seen, together with vegetable and fruit consumption, as an indication of the Mediterranean-style diet(Reference Willett, Sacks and Trichopoulou50) also among the Finnish fish consumers. Additionally, oil consumption was positively associated with fish consumption, which is also concordant with the Mediterranean-style diet.

The above referenced epidemiological studies reported associations between fish and other foods as either age-adjusted(Reference Ascherio, Rimm and Stampfer16, Reference Mozaffarian, Longstreth and Lemaitre21, Reference Hu, Cho and Rexrode24, Reference Yamagishi, Iso and Date25, Reference Albert, Hennekens and O'Donnell27, Reference Wolk, Larsson and Johansson30) or crude(Reference Patel, Sharp and Luben9, Reference Iso, Kobayashi and Ishihara17Reference Montonen, Jarvinen and Reunanen20, Reference Iso, Rexrode and Stampfer23, Reference Oomen, Feskens and Rasanen26, Reference Sugawara, Kuriyama and Kakizaki28, Reference Engeset, Andersen and Hjartaker29, Reference Kobayashi, Tsubono and Otani31, Reference Chavarro, Stampfer and Hall32). This is understandable since they reported food consumption means across fish consumption groups as baseline characteristics of their study populations. The aim of the present study was specifically to study the associations between fish and other foods, and, therefore, adjustments were essential. Adjusting for total energy intake is needed since, at least for some food groups, the more an individual eats one food, the more he or she tends to eat other foods too. Adjusting for lifestyle and all other food groups enables us to see the independent remaining effect after the confounding effect of lifestyle and the other food groups has been removed. In the present study, adjusting for other food groups attenuated the observed associations but the majority of them remained distinguishable even after the diet adjustment. This persistence can be seen as an indication of relatively strong and consistent associations. For the most part, our observations were parallel with the observations of the above referenced studies, suggesting that the present results are generalisable to other populations. Overall, diet associated with fish consumption appears to be relatively universal across populations regardless of differences in social and cultural circumstances(Reference Balder, Virtanen and Brants51) and dietary habits.

In summary, fish consumption seemed to have a positive linear association with the consumption of some other healthy foods such as vegetables, fruit, berries and oil both in the general population of Finland and in the population with high fish consumption. Additional adjustment for other food groups had a clear effect on some of the studied associations. Therefore, when evaluating the health effects of fish consumption, confounding by other foods characterising a healthy diet needs to be considered.

Acknowledgements

The present study was supported by the Academy of Finland (project no. 77008, 205324, 206950 and 124286), the Finnish Cancer Organisations, the Yrjö Jahnsson Foundation and the Juho Vainio Foundation. No funding source had any influence over the design, conduct or reporting of the present study. The authors would like to thank the volunteers and the research staff of the Fishermen study and the Health 2000 health examination survey.

A. W. T., S. M., A. L. S. and P. K. V. were responsible for conception and design, and acquisition of the data. A. W. T. and P. T. analysed and interpreted the data. A. W. T. drafted the manuscript. All authors were involved in reviewing the manuscript critically for important and intellectual content and gave final approval.

There are no conflicts of interest to declare.

References

1 Lichtenstein, AH, Appel, LJ, Brands, M, et al. (2006) Diet and lifestyle recommendations revision 2006: a scientific statement from the American Heart Association Nutrition Committee. Circulation 114, 8296.CrossRefGoogle ScholarPubMed
2 National Nutrition Council (1998) Finnish Nutrition Recommendations Committee Report 1998, 7th ed. Helsinki: Ministry of Agriculture and Forestry.Google Scholar
3 Riediger, ND, Othman, RA, Suh, M, et al. (2009) A systemic review of the roles of n-3 fatty acids in health and disease. J Am Diet Assoc 109, 668679.Google Scholar
4 Sidhu, KS (2003) Health benefits and potential risks related to consumption of fish or fish oil. Regul Toxicol Pharmacol 38, 336344.Google Scholar
5 Calder, PC & Yaqoob, P (2009) Omega-3 polyunsaturated fatty acids and human health outcomes. Biofactors 35, 266272.CrossRefGoogle ScholarPubMed
6 Mozaffarian, D (2008) Fish and n-3 fatty acids for the prevention of fatal coronary heart disease and sudden cardiac death. Am J Clin Nutr 87, 1991S1996S.CrossRefGoogle ScholarPubMed
7 He, K (2009) Fish, long-chain omega-3 polyunsaturated fatty acids and prevention of cardiovascular disease – eat fish or take fish oil supplement? Prog Cardiovasc Dis 52, 95114.CrossRefGoogle ScholarPubMed
8 Djoussé, L, Gaziano, JM, Buring, JE, et al. (2011) Dietary omega-3 fatty acids and fish consumption and risk of type 2 diabetes. Am J Clin Nutr 93, 143150.Google Scholar
9 Patel, PS, Sharp, SJ, Luben, RN, et al. (2009) Association between type of dietary fish and seafood intake and the risk of incident type 2 diabetes: the European Prospective Investigation of Cancer (EPIC)-Norfolk cohort study. Diabetes Care 32, 18571863.Google Scholar
10 Nettleton, JA & Katz, R (2005) n-3 Long-chain polyunsaturated fatty acids in type 2 diabetes: a review. J Am Diet Assoc 105, 428440.Google Scholar
11 Gonzalez, CA & Riboli, E (2006) Diet and cancer prevention: where we are, where we are going. Nutr Cancer 56, 225231.CrossRefGoogle ScholarPubMed
12 Wolk, A (2005) Diet, lifestyle and risk of prostate cancer. Acta Oncol 44, 277281.Google Scholar
13 Kaushik, M, Mozaffarian, D, Spiegelman, D, et al. (2009) Long-chain omega-3 fatty acids, fish intake, and the risk of type 2 diabetes mellitus. Am J Clin Nutr 90, 613620.CrossRefGoogle ScholarPubMed
14 Hooper, L, Thompson, RL, Harrison, RA, et al. (2006) Risks and benefits of omega 3 fats for mortality, cardiovascular disease, and cancer: systematic review. BMJ 332, 752760.CrossRefGoogle ScholarPubMed
15 Suominen-Taipale, AL, Partonen, T, Turunen, AW, et al. (2010) Fish consumption and omega-3 polyunsaturated fatty acids in relation to depressive episodes: a cross-sectional analysis. PLoS One 5, e10530.Google Scholar
16 Ascherio, A, Rimm, EB, Stampfer, MJ, et al. (1995) Dietary intake of marine n-3 fatty acids, fish intake, and the risk of coronary disease among men. N Engl J Med 332, 977982.CrossRefGoogle ScholarPubMed
17 Iso, H, Kobayashi, M, Ishihara, J, et al. (2006) Intake of fish and n3 fatty acids and risk of coronary heart disease among Japanese: the Japan Public Health Center-Based (JPHC) Study Cohort I. Circulation 113, 195202.CrossRefGoogle ScholarPubMed
18 Anderson, JS, Nettleton, JA, Herrington, DM, et al. (2010) Relation of omega-3 fatty acid and dietary fish intake with brachial artery flow-mediated vasodilation in the Multi-Ethnic Study of Atherosclerosis. Am J Clin Nutr 92, 12041213.CrossRefGoogle ScholarPubMed
19 Mozaffarian, D, Lemaitre, RN, Kuller, LH, et al. (2003) Cardiac benefits of fish consumption may depend on the type of fish meal consumed: the Cardiovascular Health Study. Circulation 107, 13721377.Google Scholar
20 Montonen, J, Jarvinen, R, Reunanen, A, et al. (2009) Fish consumption and the incidence of cerebrovascular disease. Br J Nutr 102, 750756.CrossRefGoogle ScholarPubMed
21 Mozaffarian, D, Longstreth, WT Jr, Lemaitre, RN, et al. (2005) Fish consumption and stroke risk in elderly individuals: the Cardiovascular Health Study. Arch Intern Med 165, 200206.CrossRefGoogle ScholarPubMed
22 He, K, Rimm, EB, Merchant, A, et al. (2002) Fish consumption and risk of stroke in men. JAMA 288, 31303136.CrossRefGoogle ScholarPubMed
23 Iso, H, Rexrode, KM, Stampfer, MJ, et al. (2001) Intake of fish and omega-3 fatty acids and risk of stroke in women. JAMA 285, 304312.CrossRefGoogle ScholarPubMed
24 Hu, FB, Cho, E, Rexrode, KM, et al. (2003) Fish and long-chain omega-3 fatty acid intake and risk of coronary heart disease and total mortality in diabetic women. Circulation 107, 18521857.CrossRefGoogle ScholarPubMed
25 Yamagishi, K, Iso, H, Date, C, et al. (2008) Fish, omega-3 polyunsaturated fatty acids, and mortality from cardiovascular diseases in a nationwide community-based cohort of Japanese men and women the JACC (Japan Collaborative Cohort Study for Evaluation of Cancer Risk) Study. J Am Coll Cardiol 52, 988996.CrossRefGoogle Scholar
26 Oomen, CM, Feskens, EJ, Rasanen, L, et al. (2000) Fish consumption and coronary heart disease mortality in Finland, Italy, and The Netherlands. Am J Epidemiol 151, 9991006.Google Scholar
27 Albert, CM, Hennekens, CH, O'Donnell, CJ, et al. (1998) Fish consumption and risk of sudden cardiac death. JAMA 279, 2328.Google Scholar
28 Sugawara, Y, Kuriyama, S, Kakizaki, M, et al. (2009) Fish consumption and the risk of colorectal cancer: the Ohsaki Cohort Study. Br J Cancer 101, 849854.CrossRefGoogle ScholarPubMed
29 Engeset, D, Andersen, V, Hjartaker, A, et al. (2007) Consumption of fish and risk of colon cancer in the Norwegian Women and Cancer (NOWAC) study. Br J Nutr 98, 576582.CrossRefGoogle ScholarPubMed
30 Wolk, A, Larsson, SC, Johansson, JE, et al. (2006) Long-term fatty fish consumption and renal cell carcinoma incidence in women. JAMA 296, 13711376.CrossRefGoogle ScholarPubMed
31 Kobayashi, M, Tsubono, Y, Otani, T, et al. (2004) Fish, long-chain n-3 polyunsaturated fatty acids, and risk of colorectal cancer in middle-aged Japanese: the JPHC study. Nutr Cancer 49, 3240.CrossRefGoogle ScholarPubMed
32 Chavarro, JE, Stampfer, MJ, Hall, MN, et al. (2008) A 22-y prospective study of fish intake in relation to prostate cancer incidence and mortality. Am J Clin Nutr 88, 12971303.Google Scholar
33 Aromaa, A & Koskinen, S (editors) (2002) Terveys ja toimintakyky Suomessa. Terveys 2000 –tutkimuksen perustulokset (Health and Functional Capacity in Finland. Health 2000 – Survey of Basic Results). Helsinki: Kansanterveyslaitos.Google Scholar
34 Heistaro, S (2008) Methodology Report. Health 2000 Survey. Report no. B26. Helsinki: National Public Health Institute.Google Scholar
35 Turunen, AW, Verkasalo, PK, Kiviranta, H, et al. (2008) Mortality in a cohort with high fish consumption. Int J Epidemiol 37, 10081017.CrossRefGoogle Scholar
36 Männistö, S, Virtanen, M, Mikkonen, T, et al. (1996) Reproducibility and validity of a food frequency questionnaire in a case–control study on breast cancer. J Clin Epidemiol 49, 401409.CrossRefGoogle Scholar
37 Paalanen, L, Männistö, S, Virtanen, MJ, et al. (2006) Validity of a food frequency questionnaire varied by age and body mass index. J Clin Epidemiol 59, 9941001.CrossRefGoogle ScholarPubMed
38 Paturi, M, Tapanainen, H, Reinivuo, H, et al. (editors) (2008) Finravinto 2007 – tutkimus (FINDIET 2007 – Survey). Helsinki: Kansanterveyslaitos.Google Scholar
39 Turunen, AW, Mannisto, S, Kiviranta, H, et al. (2010) Dioxins, polychlorinated biphenyls, methyl mercury and omega-3 polyunsaturated fatty acids as biomarkers of fish consumption. Eur J Clin Nutr 64, 313323.CrossRefGoogle Scholar
40 Reinivuo, H, Hirvonen, T, Ovaskainen, ML, et al. (2010) Dietary survey methodology of FINDIET 2007 with a risk assessment perspective. Public Health Nutr 13, 915919.Google Scholar
41 Djerf, K, Laiho, J, Kehtonen, R, et al. (2008) Weighting and statistical analysis. In Methodology Report Health 2000 Survey. Report no. B26, pp. 182200 [Heistaro, S, editor]. Helsinki: National Public Health Institute.Google Scholar
42 SAS Institute Inc. (2008) SAS/IML® 9.2 User's Guide. Cary, NC: SAS Institute Inc.Google Scholar
43 Wood, SN (2006) Generalized Additive Models: an Introduction with R. Boca Raton, FL: Chapman and Hall.Google Scholar
44 R Development Core Team (2009) R: A Language and Environment for Statistical Computing. Vienna: R Foundation for Statistical Computing.Google Scholar
45 Willett, WC (1998) Food frequency methods. In Nutritional Epidemiology, pp. 74100 [Willett, WC, editor]. New York: Oxford University Press.Google Scholar
46 Pala, V, Sieri, S, Masala, G, et al. (2006) Associations between dietary pattern and lifestyle, anthropometry and other health indicators in the elderly participants of the EPIC-Italy cohort. Nutr Metab Cardiovasc Dis 16, 186201.Google Scholar
47 van Dam, RM, Rimm, EB, Willett, WC, et al. (2002) Dietary patterns and risk for type 2 diabetes mellitus in U.S. men. Ann Intern Med 136, 201209.Google Scholar
48 Heidemann, C, Schulze, MB, Franco, OH, et al. (2008) Dietary patterns and risk of mortality from cardiovascular disease, cancer, and all causes in a prospective cohort of women. Circulation 118, 230237.Google Scholar
49 Albert, CM, Campos, H, Stampfer, MJ, et al. (2002) Blood levels of long-chain n-3 fatty acids and the risk of sudden death. N Engl J Med 346, 11131118.Google Scholar
50 Willett, WC, Sacks, F, Trichopoulou, A, et al. (1995) Mediterranean diet pyramid: a cultural model for healthy eating. Am J Clin Nutr 61, 1402S1406S.Google Scholar
51 Balder, HF, Virtanen, M, Brants, HA, et al. (2003) Common and country-specific dietary patterns in four European cohort studies. J Nutr 133, 42464251.Google Scholar
Figure 0

Table 1 Background data of the Health 2000 survey and the Fishermen substudy participants(Mean values or percentages with 95 % confidence intervals)

Figure 1

Table 2 Model-adjusted food consumption (g/d) by fish consumption tertiles (g/d) among the Health 2000 survey men and women(Geometric means and 95 % confidence intervals)

Figure 2

Table 3 Model-adjusted food consumption (g/d) by fish consumption tertiles (g/d) among the fishermen and the fishermen's wives(Geometric means and 95 % confidence intervals)

Figure 3

Fig. 1 Adjusted smoothed associations between fish consumption and the consumption of selected foods (vegetables, (a) and (b); fruit and berries, (c) and (d)) among the Health 2000 survey men (n 2605; (a) and (c)) and women (n 3199; (b) and (d)). Associations were adjusted for age, BMI, education, marital status, smoking, physical activity and other food groups, and were produced by an additive model with a thin-plate regression spline. All the FFQ variables were transformed according to log(x+1). The solid curve is the additive model fit and the dashed curves represent the approximate 95 % CI. The fit is named as s(log_fish, edf), where edf is the estimated degrees of freedom describing the wiggliness of the fit. The plotted points are partial residuals.

Figure 4

Fig. 2 Adjusted smoothed associations between fish consumption and the consumption of selected foods (red meat, (a) and (b); sausages, (c) and (d)) among the Health 2000 survey men (n 2605; (a) and (c)) and women (n 3199; (b) and (d)). Associations were adjusted for age, BMI, education, marital status, smoking, physical activity and other food groups, and were produced by an additive model with a thin-plate regression spline. All the FFQ variables were transformed according to log(x+1). The solid curve is the additive model fit and the dashed curves represent the approximate 95 % CI. The fit is named as s(log_fish, edf), where edf is the estimated degrees of freedom describing the wiggliness of the fit. The plotted points are partial residuals.

Supplementary material: File

Turunen Supplementary Material

Turunen Supplementary Material

Download Turunen Supplementary Material(File)
File 33.3 KB