Does childhood nutrition influence adult cardiovascular disease risk?—Insights from the Young Finns Study

Abstract

There is a paucity of detailed information about the role of childhood food patterns or on the impact of individual nutrients on adulthood cardiovascular disease (CVD). We review here the reports that have investigated these questions in the Young Finns Study with its 3596 subjects at baseline, aged 3 to 18 years. All the participants filled in a food habit questionnaire, and half of them provided a 48-hour dietary recall interview. In adulthood, cardiovascular risk factors as well as structural and functional markers of subclinical atherosclerosis were measured, i.e. carotid artery intima media thickness (IMT), and measurements of arterial elasticity and brachial artery endothelial function. Our data demonstrate that dietary patterns can already be identified in childhood. These patterns remain relatively stable over the life-course and associate with cardiovascular risk factors and vascular markers of subclinical atherosclerosis. For example, a traditional dietary pattern characterized by low intakes of fruits and vegetables was associated with elevated increased adulthood IMT especially in men, whereas a diet with a high intake of vegetables was independently associated with increased arterial elasticity in both genders. Our findings and the current literature suggest that childhood nutrition has a significant role in the progression of CVD.

Key Words::

• Adulthood
• cardiovascular disease
• childhood nutrition
• food intake
• intima media thickness
• metabolic syndrome
• vascular health

Key messages

• Obesity and eating habits/dietary patterns track from childhood to adulthood.

• An unhealthy diet in childhood, poor in fruits and vegetables, can be linked in adults to cardiovascular disease (CVD) risk factors, as well as to early vascular changes, i.e. evidence of an elevated risk of CVD.

• The current findings emphasize the importance of meeting already in childhood the dietary guidelines (aimed at promoting cardiovascular health) set by the American Heart Association and American Academy of Pediatrics.

Introduction

Risk factors of atherosclerosis

There is an increasing body of evidence indicating that atherosclerosis actually has its origins in childhood. Data from longitudinal cohort studies have demonstrated that exposure to risk factors in youth is associated with undesirable cardiometabolic outcomes in adulthood (1–4). The factors known to associate with cardiovascular disease (CVD) include age, male gender, or postmenopausal state in women, excess adiposity or overweight and obesity, low physical activity, dyslipidemia (for example, high low-density lipoprotein cholesterol or triglyceride levels and/or low high-density lipoprotein cholesterol), hypertension, diabetes, genetic susceptibility, low fruit, vegetable, and fish/fish oil consumption, high saturated fat, refined carbohydrate, and salt (sodium) intake, and smoking (5,6). For many of these factors, such as hypertension, dyslipidemia, and obesity, an unbalanced diet plays an important role in their development. However, there have been no longitudinal studies published so far that have directly linked childhood nutrition to CVD or to its risk factors or subclinical markers in adulthood. In contrast, in middle-aged and older subjects, there is evidence from several prospective trials that high fruit and vegetable consumption is associated with reduced CVD risk (7,8).

Etiology of atherosclerosis

The progression of arterial atherosclerosis involves a cascade of consecutive steps, starting from endothelial injury, followed by the formation of foam cells, fatty streaks, increased intima media thickness, endothelial dysfunction, and loss of vessel elasticity. Finally, plaques containing fibrous calcium deposits are formed, and they can rupture and release their contents into the bloodstream (9,10). Inflammation (10) and probably also oxidation of low-density lipoprotein (11,12) promote the early phases of this process, leading to an accelerated uptake of low-density lipoprotein by macrophages at the injury site and the first signs of foam cells. Even though detailed molecular mechanisms leading to atherosclerosis are still unclear, the risks associated with a high low-density lipoprotein/high-density lipoprotein ratio, and the preventive pleiotropic role of high-density lipoprotein, are well recognized. High-density lipoprotein may decrease inflammation, prevent low-density lipoprotein oxidation, vascular endothelial cell apoptosis, and thrombosis, and even improve vascular endothelial function (13). Overweight, an indirect indicator of unbalanced nutrition and/or low physical activity, can be linked to high low-density lipoprotein/high-density lipoprotein ratio already in adolescents (14) but also to other mechanisms, such as inflammatory mechanisms (15), which may promote atherosclerosis.

A reduction in obesity is clearly associated with a decreased risk of CVD. It has been recognized in adults that even a slight, 5%–10%, weight loss is an effective way of lowering the risk of CVD (16). However, there is a clear need to clarify the role of childhood nutrition and the changes in nutrition between childhood and adulthood on cardiovascular health. It is plausible that in young subjects, preventive actions could be even more effective than in older subjects.

Study purpose

The purpose of this review is to assess the current available literature to determine whether childhood dietary habits, food consumption, or nutrient intakes are associated in adulthood with 1) dietary habits, 2) known risk factors for CVD, and 3) subclinical markers of atherosclerosis. This review contains literature available from cohort studies in which data have been collected from childhood. In particular, we will focus on data from our own population-based prospective cohort, the Cardiovascular Risk in Young Finns Study (YFS) that began in 1980 (17). The YFS is one of the few studies where data are available from several aspects, i.e. on eating habits (such as food behavior or breast-feeding), food consumption (including the consumption of fruits, vegetables, and sweets), and nutrient intakes (such as selenium intake), as well as many other risk factors assessed in detail both in childhood and in adulthood from the same participants. This review does not contain any unpublished data.

Material and methods

Materials and methods are described in this review to provide the readers with an overview of the Young Finns Study and the methodology used to collect longitudinal nutrition data.

Subjects

In 1980, 4320 children and adolescents aged 3, 6, 9, 12, 15, and 18 years were randomly chosen from the Finnish national population register to obtain a sample that would represent Finnish children and adolescents (17). The girls and boys selected lived in 5 university cities and 12 rural communities in their vicinity. A total of 3596 (83.1%) of those invited participated in the baseline examination. A total of 2991 subjects participated in the 3-year follow-up (1983) and 2799 the 6-year follow-up (1986). Examinations were carried out in 2620 subjects during the 21 year follow-up (2001) and in 2204 subjects during the 27 year follow-up (2007) (17). At baseline, all participants (or the parents of those aged 3 to 9) filled in a dietary behavior questionnaire (18). The questionnaire consisted of questions on habitual eating patterns, special diets, snacking, and on the amount and type of milk and dietary fats consumed. It also included a short non-quantitative food frequency question on the consumption of some foods relevant to the development of CVD, such as fruit and vegetables, milk products, fish, meat, sweets, and soft drinks. The consumption was assessed with the following frequency alternatives: used 1) daily, 2) almost every day, 3) a couple of times per week, 4) about once a week, 5) a couple of times per month, and 6) only occasionally or never (18). The food habit questionnaire was repeated for all of the participants in 1983, 1986, 2001, and 2007.

In addition to the questionnaire, half of the subjects provided more detailed quantitative information on their food intake by a 48-h dietary recall method (n = 1768) which gathered information on foods and beverages consumed by the subjects during the two days prior to the interview. Children aged 3–12 years were interviewed together with their mothers, fathers, or some other accompanying individual (19). The food composition data used in the 1980s were based on the Finnish food composition tables (maintained by the University of Helsinki, later by the National Public Health Institute) and on analytical data obtained from the local food processing industry, details of which have been described previously (20,21). The recall/food frequency questionnaire was repeated after 6, 21, and 27 years.

General principles of data collection

The data collections have included blood tests, physical examinations, questionnaires, and interviews (17). One important criterion in the data collection has been that the data provided can be compared between the study visits, even though some modifications may have been made, for example to the questionnaires to characterize better the status of the subjects with their increasing age. Dietary recalls have been carried out by well-trained nutritionists. In order to decrease random error, it may have been necessary to pool the data from several study visits for statistical analyses, for example to form the average childhood and adulthood diets. In standardized laboratory analyses, methodological developments have taken place within the past 30 years. For this reason, when needed, certain factors have been used to permit a comparison of the laboratory values between the study visits. The data collections and laboratory analyses described in this review have been described in detail in the original publications.

Methods to assess early atherosclerosis

Due to the relatively young age of the study population, there are no hard end-point data values available. Therefore, we have used recognized biomarkers of vascular health, such as carotid intima media thickness (IMT) and arterial pulse wave velocity, to examine the association between childhood risk exposures and arterial health in adults (17).

Carotid IMT has been used widely in cardiovascular epidemiology as a marker of subclinical atherosclerosis, since it correlates with traditional risk factors, and increased IMT values have been shown to predict cardiovascular events (22). IMT values were measured both in 2001 and in 2007, providing the opportunity to assess their progression over a 6-year period.

In addition to carotid artery IMT values, we measured arterial pulse wavelength velocity in the 27-year follow-up using a bio-impedance method (whole body impedance cardiography device, CircMon, JR Medical Ltd, Tallinn, Estonia) (18). Arterial pulse wavelength velocity is a marker of central arterial stiffness, i.e. an elevated velocity value is indicative of decreased elasticity or increased stiffness. Increased arterial pulse wavelength velocity has been shown independently to predict the risk of CVD events (23).

Results and discussion

Tracking of eating behavior and dietary patterns from childhood to adulthood

Since it is difficult to evaluate the quality of the diet using only single or just a few nutrients, there has been growing interest in trying to use nutritional epidemiology to gain a comprehensive whole-diet perspective to examine the role of nutrition on the risk of chronic diseases. The use of dietary patterns has been emphasized since this reflects the reality that foods are not consumed one at a time but in combinations, which can result in complex synergistic or opposite effects. Two approaches can be applied to investigate the dietary patterns: 1) data-driven techniques were used to reveal whether there were differences in food choices between the subjects (for example based on factor or cluster analyses) or 2) scoring-based systems were used to investigate how close the food choices are to patterns determined in advance, such as DASH (Dietary Approaches to Stop Hypertension) or Mediterranean patterns of eating (24–26). Thus far, we have used the population-specific factor analysis.

To test the association between childhood and adulthood nutrition (i.e. possible tracking), we conducted a principal component analysis with 23 food groups (Table I) to identify major dietary patterns at baseline and at the follow-ups of the YFS (27). Two distinct patterns were recognized throughout the study visits (Figure 1). Pattern 1 (named traditional pattern) was positively correlated with consumption of traditional Finnish foods, such as rye, potatoes, milk, butter, sausages, and coffee, and negatively correlated with fruit, berries, and dairy products other than milk. The pattern 1 type of diet was more common among male subjects, smokers, and those living in rural areas (27). Pattern 2 (named health-conscious pattern), predominant among female subjects, non-smokers, and inhabitants of urban areas, was characterized by more health-conscious food choices, such as vegetables, legumes and nuts, tea, rye, cheese, and other dairy products (27). Similar patterns as measured in 1980 were identified in 1986 and also in adulthood in 2001 (Figure 1). In further analyses, there seemed to be tracking of the pattern scores, particularly among subjects who were adolescents at baseline (27). By 2001, all of the subjects were young adults, but only the health-conscious dietary pattern showed a strong positive association with the use of alcoholic beverages (27). This was mainly due to the wine consumption of these health-conscious subjects. However, it must be kept in mind that heavy drinking is often underestimated in any data gathered by questionnaires or recall methods, therefore any findings on the role of alcohol must be interpreted with caution. Of those originally belonging to the uppermost quintile of pattern 1 and 2 scores, 41% and 38%, respectively, were still in the same quintile 21 years later (27). We observed a long-time trend that the eating habits were changing in the direction of the official recommendations, lowering the tracking in general (27). Even though the factor analysis is population-specific, ‘Prudent or Healthy’ and ‘Western’ dietary patterns have been identified with this same method also in other study populations (25,28).

Figure 1. Star plots for the dietary patterns identified in subjects in 1980 (A), 1986 (B), and 2001 (C). Each arm of the star in the graphic presentation illustrates the correlation between the patterns and the different food groups, with a negative correlation (r = −1) at the midpoint and a positive correlation (r = +1) at the outer edge of the constellation (1 = traditional pattern 2 = health conscious pattern). A correlation of zero is indicated by a circle. The dietary patterns significantly tracked from childhood to adulthood. From reference (27) with permission.

Table I. Food groups used in the factor analysis for dietary patterns. From reference (27) with permission.

Food group	Foods included in the group
Rye	Rye bread, rye porridge
Wheat	Wheat bread, pasta
Other cereals	Cereals other than rye and wheat, breakfast cereals, biscuits, starch, rice
Potatoes	Potatoes, potato products
Root vegetables	Root vegetables
Legumes and nuts	Peas, beans, other legumes, nuts, seeds, soy products
Other vegetables	Leaf vegetables, onions, cabbages, tomatoes, cucumber, canned vegetables, mushrooms
Fruit and berries	Fresh fruits, canned fruits, berries, fruit and berry juices
Milk	Milk
Cheese	Cheese
Other dairy products	Cream, sour milk products, ice-cream
Butter	Butter, butter-oil spreads, lard
Margarines and oil	Soft margarine, low-fat spreads, oil
Pork	Pork
Other meat	Beef, game, poultry, lamb, meat products
Sausages	Sausages, frankfurters
Offal	Liver, kidney, other offal
Fish and shellfish	Fish, shellfish, fish products
Eggs	Eggs
Sugars and confectionery	Sugar, syrup, sweets, chocolate
Coffee	Coffee
Tea	Tea
Alcoholic beverages	Alcoholic beverages

Our findings are in line with those presented in other studies, such as the ASH30 Study where food intake in adolescence was a significant although not strong predictor of intake in adulthood more than 20 years later (29). Finnish society has changed dramatically during the last 20–30 years, following greater urbanization, increased level of education, and more and more employees now working in the service sector, resulting in less physical work. Furthermore, in addition to nutritional recommendations and counseling, travel and globalization with the availability of culture-specific diets may have influenced eating habits. Nonetheless, it is interesting to observe that dietary habits still seemed to be deeply ingrained with significant tracking from childhood to adulthood. In our study population, there is also some evidence indicating that eating behavior might be related to socio-demographic factors. In young adults aged 18–24 years, parental education related directly to dietary intake of polyunsaturated fatty acids, and the income level of the family directly with the frequent intake of alcoholic beverages, and inversely with the percentage of dietary energy from fat (30). However, no associations were found between socio-demographic factors (including education of the subjects and their parents) and an index describing the predefined cardiovascular quality of the diet (constructed from the intakes of saturated fat, fiber, and salt) in childhood or in adulthood (19). In the whole-diet approach using dietary patterns, the traditional pattern was found to be significantly less predominant than the health-conscious pattern among subjects with high education or highly educated parents (27). It may be that the differences are manifested in the actual food choices rather than in nutrient intakes, e.g. a high consumption of cheese (correlated with the health-conscious pattern) may result in a similar intake of saturated fat to a high consumption of milk and butter (correlated with the traditional pattern).

Tracking can have an important effect on health outcomes via two routes. Firstly, if detrimental dietary factors are present for a long period of time, their effects accumulate, creating a risk load which may eventually lead to illness or a deterioration in health. This is of particular relevance in degenerative diseases, such as CVD, which insidiously develop over a long period of time. Secondly, the existence of unbalanced dietary factors for a prolonged time may be critical for the disease development, thus increasing the adulthood disease risk. This is illustrated in the development of healthy bones where adolescence is known to represent a critical period. A similar causality may also exist with CVD, as there is some evidence that obesity or life-style factors at puberty may be independently associated with the adulthood CVD risk (31).

Childhood nutrition and adult risk factors

Obesity and adult cardiovascular outcomes

Indirect evidence for the causality between childhood nutrition and possible risk for later CVD emerges from studies that have linked childhood obesity to adulthood cardiovascular outcomes. In the Harvard Growth Study (n =508), being overweight during adolescence was associated with a range of adverse health effects, including coronary artery disease in the follow-up which was carried out 50 years later (32). This causality was confirmed in a large Danish national register-based study that linked childhood obesity (body mass index) to increased risk of adult coronary heart disease in a population of nearly 300,000 Danish schoolchildren (1). In a review with over 700,000 subjects, a statistically significant although weak association was observed in the evaluation of 15 longitudinal studies. It was calculated that every 1-unit increase in body mass index among children and adolescents aged 7–17 years was associated with a relative risk of 1.05 for coronary heart disease in later life. There was no association between body mass index and subsequent coronary heart disease among children who were less than 7 years of age (33). There are some shortcomings in this study, for example there was considerable statistical heterogeneity between the study estimates. There are also some additional findings suggesting that childhood body mass index is poorly or not at all associated with adulthood CVD risk (34,35).

More recently, in an Israeli study with 37,674 healthy men (aged 17 years at baseline) (36), elevated adolescent body mass index predicted both subsequent diabetes and angiography-proven coronary heart disease 17 years later in adulthood in multivariate models adjusted for age, family history, blood pressure, life-style factors, and blood biomarkers. Further adjustment for adulthood body mass index did not dilute the association between adolescent body mass index and subsequent coronary heart disease (36). As compared with the lowest decile, even the body mass index that was considered to be normal increased the risk for the above-mentioned disorders in later life (36).

In many studies, the positive association between childhood body mass index and adulthood CVD risk factors/outcomes has been attributed to tracking of obesity from childhood to adulthood (37–42). In our study population aged 3–18 years at baseline, the correlation coefficient between childhood and adulthood body mass index (21 years later) was 0.30 to 0.65, varying between the age-groups (39). We utilized an age-adjusted multivariable model to assess the link between traditional risk factors in childhood (12–18-year-old children) and IMT measured in 2001 (n =1170) (41). In addition to high low-density lipoprotein cholesterol, the other determinants of increased adulthood carotid IMT were male gender, high systolic blood pressure, smoking, and high body mass index. When these childhood risk factors were adjusted for the corresponding adulthood risk factors, the association of body mass index and smoking did not remain significant, whereas low-density lipoprotein and systolic blood pressure remained independently associated (41). In addition, it was found that adult age and sex-adjusted IMT values in subjects who had been overweight/obese throughout life were comparable to the levels in obese/overweight subjects who were of normal weight as children (39). These findings were recently confirmed in the i3C consortium consisting of our cohort, and the other long-lasting follow-up studies from childhood to adulthood, such as the Bogalusa Heart Study, the Muscatine Study, and the Childhood Determinants of Adult Health study, totaling 6328 subjects (2961 male and 3367 female subjects) with a mean (± SD) age of 11.4±4.0 years at baseline and follow-up of 23.1±3.3 years (42). These data confirmed both the increase in CVD risk outcomes, i.e. carotid artery atherosclerosis, type 2 diabetes, hypertension, and dyslipidemia associated with childhood overweight or obesity and the tracking of adiposity between childhood and adulthood. The risks of these outcomes among overweight or obese children who became non-obese by adulthood were similar to those among persons who were never obese (42). In summary, these findings suggest that childhood obesity may have a negative influence on adult cardiovascular health mediated by adulthood obesity (37–42) since it is known that a high body mass index in childhood is associated with a high risk of obesity in adulthood (43). Due to tracking, most overweight and obese children tend to be obese adults. Therefore, we hypothesized that reducing adiposity in children and adolescents who are overweight or obese could reduce their later cardiovascular risk (42). There is also some evidence that those obese adults who have been normal weight or even below the normal weight as children may have the most adverse levels of blood pressure/hypertension (37,44,45). Such observations may suggest that large changes in adiposity status may pose an additional risk in the development of hypertension.

Although it is clear that adult obesity (body mass index >30 kg/m²) is strongly linked to increased CVD mortality, the role of overweight (body mass index 25–30 kg/m²) is less clear. In the NHANES study, the cause of death information was available for 2.3 million US adults in 2004, with a balanced follow-up (limited follow-up of no more than 15 years for each survey). It was found that the CVD mortality was lower in mildly overweight adults (body mass index 25–29.9 kg/m²), as compared with normal weight adults (46). However, parallel results with a total follow-up (using all follow-up data for all surveys) did not support this primary finding. Nevertheless, also in that study, adult obesity (body mass index ≥30 kg/m²) was strongly linked to increased CVD mortality (46).

Nutrition and adult risk factors

We examined the relations between the two dietary patterns and cardiovascular risk factors in YFS subjects (n =1768 at baseline). Even when adjusting for age, gender, smoking habits, physical activity, total energy intake, and study year, our subjects, both males and females, who had high scores in the traditional pattern from childhood to adulthood (21 years later) had higher total cholesterol, low-density lipoprotein cholesterol, apolipoprotein B, and C-reactive protein levels than those with lower scores (47). With respect to the other factors, for example triglycerides, high- density lipoprotein cholesterol, and apolipoprotein A1, we could not detect any significant associations. On the other hand, the health-conscious pattern from childhood to adulthood was associated with lowered risk factor levels, although this was mainly found in women, i.e. significantly attenuated total cholesterol, LDL cholesterol, apolipoprotein B, and C-reactive protein levels (47).

In the CDAH Study (48), Smith and co-workers examined 2184 subjects and assessed longitudinal associations of breakfast-skipping in childhood and adulthood with cardiometabolic risk factors in adulthood. After adjustment for age, sex, and socio-demographic and life-style factors, those participants who skipped breakfast in both childhood and adulthood had a larger waist circumference and higher fasting insulin, total cholesterol, and low-density lipoprotein cholesterol levels than those who ate breakfast at both time points. Additional adjustments for diet quality and waist circumference attenuated the associations with cardiometabolic variables, but the differences remained significant. The research group concluded that skipping breakfast over a long period may have detrimental effects on cardiometabolic health (48).

In this same study cohort (49), Magnussen and co-workers examined a subset of 539 subjects and evaluated the effect of life-style changes on the stability of plasma/serum lipid and lipoprotein levels from youth to adulthood. The authors concluded that unhealthy life-style changes that occurred between youth and adulthood influenced whether an individual maintained, lost, or developed high-risk blood lipid and lipoprotein levels in adulthood. However, the change in saturated fat intake did not associate with the tracking of plasma/serum lipid or lipoprotein levels from childhood to adulthood (49).

In the Beaver County Lipid Study, Stuhldreher and co- workers found that those subjects who had high-risk lipid levels at baseline but not at follow-up had gained less weight, were less frequent smokers, and adhered to a diet low in saturated fat and cholesterol as compared with those who were hypercholesterolemic in adulthood (50).

It is unclear whether the use of sweets is associated with increased childhood obesity (51), but it may have cumulative long-term health influences. We investigated the associations of body mass index and overweight in adulthood with the consumption 21 years earlier of sweets and sugar-sweetened soft drinks in childhood and also with the change in consumption between childhood and adulthood (52). In women (n =1172), but not in men (n =967), we found evidence for direct associations between increasing frequency of sugar-sweetened soft drinks consumption from childhood or adolescence to adulthood with adult body mass index and overweight. The model was adjusted for age and overweight status in childhood, and education, smoking, and physical activity in adulthood. However, our study did not find any association between childhood consumption of sweets and sugar-sweetened soft drinks with the likelihood of being overweight as an adult (52).

Childhood nutrition and adult subclinical outcomes

Adult arterial pulse wave velocity

We examined arterial pulse wave velocity in adulthood at the 27-year follow-up among a subset of 1622 participants for whom childhood risk factor data were available (18). We found that consumption of vegetables (average age and sex-specific Z scores from the years 1980–1986) in childhood was inversely associated with adulthood arterial pulse wave velocity, and this association remained significant even after adjustment for childhood traditional risk factors (high-density and low-density lipoprotein cholesterol, triglycerides, systolic blood pressure, body mass index, and smoking). This association between childhood vegetable consumption and adulthood arterial pulse wave velocity remained significant also when adjusted for adulthood vegetable consumption, age, and sex. In a sub-analysis, persistently high consumption of both fruits and vegetables (the highest quintiles) from childhood to adulthood was associated with lower arterial pulse wave velocity compared with persistently low consumption (the lowest quintiles) (18).

To the best of our knowledge, this is the first study to have demonstrated the associations between childhood life-style risk factors and adulthood arterial pulse wave velocity, i.e. the previous studies reporting inverse associations between plant-derived compounds and arterial stiffness have been carried out only in middle-aged and older adults (53,54). Our finding especially emphasizes the role of a vegetable-rich diet during childhood in the prevention of CVD later in life. It is also possible that the elevated consumption of fruit and vegetables is an indicator of an overall heart-healthy diet, especially in the 1980s, when food selection in Finland was much more limited (e.g. not all fruits and vegetables were available throughout the year) and daily consumption of fresh vegetables was likely to be a conscious health-promoting choice in the family. Indeed, in the dietary pattern analyses, the consumption of fruit and vegetables was strongly correlated with the health-conscious pattern as well as with other health-promoting life-style factors (27).

Adult carotid intima media thickness (IMT)

In our study population, the carotid IMT values increased with age both in males and in females, the average progression being 46 micrometers per year, which was assessed in subjects aged 24–45 years, reflecting the magnitude of vascular aging.

We investigated which of these childhood risk factors (age and sex-specific Z scores) influenced the 6-year IMT progression (2001–2007) as an outcome variable (40). The strongest inverse correlation was detected in multivariable models with childhood measures of HDL cholesterol, followed by physical activity and fruit consumption (n =1744 male and female subjects aged 3–18 years at baseline). A direct correlation was also observed with body mass index. When taking into account the effects of adulthood risk factors, the association between childhood fruit consumption with adulthood IMT progression remained significant, whereas childhood obesity did not show any independent association. These findings were equal for both genders (40).

In addition, we also used multivariable models to evaluate the effect of the distinctive food patterns on IMT measured in 2001 (n =420 women and 365 men) (55). It was found that in men the high average scores from childhood to adulthood for the traditional pattern, adjusted for age, energy intake, smoking, exercise, education, and finally for systolic blood pressure, low-density lipoprotein cholesterol, and body mass index as possible factors promoting CVD, were associated with increased carotid IMT. In women, the significant association was attenuated when the age and energy intake-adjusted model was further adjusted for smoking, exercise, and education. The effect was particularly clear in a subgroup of participants having a high score for the traditional dietary pattern accompanied by a low score for the health-conscious pattern. The health-conscious dietary pattern displayed no independent associations with IMT (55).

Breast-feeding, selenium fortification, and adult subclinical outcomes

We also investigated the effect of breast-feeding (yes versus no) in infancy on the arterial function and structure 21 years later in adulthood (56). Non-invasive ultrasound was used to measure brachial artery flow-mediated dilatation, carotid artery intima media thickness, and carotid artery compliance in a subset of 1667 young adults. The linear regression models were adjusted for adulthood low-density and high-density lipoprotein cholesterol, systolic blood pressure, waist circumference, brachial vessel size, smoking, and birth-weight. One finding was that maximal brachial artery flow-mediated dilatation was higher in men who had been breast-fed in infancy compared to their formula-fed counterparts. We postulate that adult men who have been breast-fed enjoy better brachial endothelial function compared to men who have been formula-fed (56).

Recently, concern has been raised about possible adverse cardiometabolic effects of high selenium status, such as increased risks of diabetes and hyperlipidemia (57). In Finland, a national fortification program has been in place for selenium since the 1980s. We measured serum selenium in a subset of 1235 young Finns aged 3–18 years at baseline in 1980 prior to the fortification program (the mean ± SD selenium status was 74.3±14.0 ng/mL), and in a subgroup (n =262) at the 6-year follow-up in 1986 following its initiation (the mean selenium status was then 106.6±12.5 ng/mL) (58). After adjusting for age, sex, body mass index, smoking, and parental socio-economic status, we found positive cross-sectional associations between baseline selenium status and serum lipids (serum total cholesterol and high-density and low-density lipoprotein cholesterol), this being in line with previous studies (59,60). However, the significant increase in serum selenium after the implementation of the nationwide fortification program was not associated with any parallel increases in lipid levels.

Strengths and limitations of the Young Finns Study

In summary, the main strength of the YFS is the longitudinal study design with its high participation rate, which made it possible to investigate genetics, environmental factors, life-style factors, anthropometrics, and morbidity in the same participants first as children and then when they grew to be adults. The study subjects are representative of the Finnish general population. Even though the collection of nutritional data is based on eating habit questionnaires and two-day recalls, the frequent follow-ups have ensured the collection of reliable and detailed nutrition data at a population level. It has been possible to pool the data from several study visits for statistical purposes, for example, to devise so-called average childhood and average adulthood diets. One problem causing systematic error, and which we cannot exclude in studies like this, is the possible under-reporting related to the use of unhealthy items, such as sweets. Since the baseline visit was in the early 1980s, it is also unavoidable that there was variability in food intake between subjects, i.e. there were clear fruit and vegetable users, as well as those who consumed a traditional diet. The nutritional differences related to the use of fruits and vegetables have diminished at the more recent follow-ups, as food habits have changed in a more favorable direction towards recommendations.

As a shortcoming, observational studies like the YFS cannot establish causality. Due to the young study population, it is not yet possible to examine the associations between childhood risk factors with disease end-points. In addition, the generalizability of the findings is limited to white European subjects, and the loss of some subjects is an inevitable problem in all long-term studies. Furthermore, original enrolment of the subjects occurred over a wide age-range (3–18 years) resulting in a lack of early childhood data for a large number of the study subjects (17).

Summary

In our study population, two dietary patterns were identified, the traditional one and the health-conscious one. Our results suggest that eating habits/patterns track from childhood to adulthood (Figure 1), and that unhealthy childhood nutrition, i.e. a diet poor in fruits and vegetables, can be linked to adulthood CVD risk factors, and to early vascular changes predicting the risk of CVD (Table II). It is generally believed that nutritional habits are established in childhood, but there have been only limited data to support this hypothesis. Our results are of interest as they suggest that food behavior and food choices are established already in childhood or adolescence, and they affect cardiovascular health in adulthood.

Table II. Summary of the findings in the Young Finns Study (focusing on childhood risk and preventive factors versus adulthood outcomes).

Items investigated	Outcome variable	Findings
Childhood traditional risk factors for arterial health (41)	Intima media thickness (IMT) in young adults	Childhood risk factors found: LDL cholesterol; male gender; systolic blood pressure; smoking; body mass index
Childhood factors linked to arterial health (40)	Progression of IMT (6-year change) in young adults	Childhood preventive factors found: HDL cholesterol; physical activity; fruit consumption (independent). Childhood risk factors found: body mass index; systolic blood pressure
Childhood/longitudinal traditional dietary pattern (47)	Adulthood cardiovascular risk factors	Adulthood/longitudinal factors found: total cholesterol; LDL cholesterol; apolipoprotein B; C-reactive protein
Childhood dietary patterns (traditional versus health-conscious) (55)	IMT in young adults	Increased IMT related to traditional pattern
Childhood use of fruits and vegetables (18)	Arterial pulse wave velocity in adults	Decrease in arterial stiffness—the effect of fruits was independent
Increase in sweet consumption between adolescence and adulthood (sweets and sugar-sweetened soft drinks) (52)	Obesity in adulthood	Women: increased obesity due to soft drinks. Men: no effect
Childhood sweet consumption (52)	Obesity in adulthood	No effect
Breast-feeding in infancy (56)	Endothelial function (EF) in adults	Improved EF compared with formula users in men
Increase in serum selenium levels during childhood and adolescence prior to and following national fortification program (58)	Serum lipid levels in adults	No effect

EF = endothelial function; HDL = high-density lipoprotein; IMT = intima media thickness; LDL = low-density lipoprotein.

On the basis of our findings, the traditional Finnish food pattern containing rye, potatoes, milk, butter, sausages, and coffee was associated with several CVD risk factors, such as total cholesterol, low-density lipoprotein cholesterol, apolipoprotein B, C-reactive protein, as well as elevated arterial stiffness and increased intima media thickness, two markers of elevated CVD risk, in adulthood. On the other hand, fruits and vegetables were especially beneficial foods, being independently linked to adulthood carotid IMT and arterial pulse wavelength velocity, respectively. Thus, the adjustment of models for corresponding adulthood food items did not abolish the beneficial associations found.

Overall, there is evidence that the impact of an unhealthy life-style and obesity during childhood on subsequent adult cardiovascular health may be mainly tracking-based, i.e. the life-style changes between youth and adulthood may determine whether the youth obesity-related CVD risk factors will persist also in adulthood (42). If this is the case, then it would be important to adopt a healthy life-style, which includes balanced nutrition, early in life to decrease the risk of future cardiovascular disease.

We postulate further, based on individual YFS observations, that 1) the increase in consumption of sweet items following childhood may elevate weight in adulthood, at least in females, 2) serum selenium status does not seem to influence serum lipid levels, and 3) breast-feeding for male infants seems to have beneficial effects on brachial endothelial function. Nevertheless, these observations need to be confirmed in other populations.

It may be that the lack of findings linking childhood dietary factors directly to adulthood health is at least partly due to methodological difficulties in capturing the relevant features of the diet, conducting appropriate longitudinal analyses, and to the lack of valid, prospective dietary assessment and proper end-points within the same cohort. It is also possible that the lack of data reflects bias attributable to weak or statistically non-significant findings in other study populations. The YFS will represent a unique opportunity to overcome many of these limitations in the future because of its long follow-up time, good participation rate, and extensive exposure data.

In general, the YFS findings are in good agreement with the recommendations which have accumulated over the past 50 years (5) and with the current dietary guidelines of the American Heart Association and the American Academy of Pediatrics aimed at promoting cardiovascular health (61). It is important to appreciate that international and national recommendations for children are based on the corresponding guidelines for adults, since there is very little long-term data linking nutrition in childhood to the situation when they grow to become adults. Our findings emphasize the importance of adhering to these guidelines already in childhood.

The possible beneficial mechanisms of fruits and vegetables may be mediated through their anti-oxidative, anti-inflammatory, and lipid-lowering effects, their impact on glucose and insulin metabolism, and their influences on vascular endothelial function (5,62–64). In addition, their anti-thrombotic effects (63) may prevent serious CVD complications. It is plausible that appropriate nutritional counseling and optimal follow-up, starting in childhood, could significantly reduce the ever-increasing health care costs and, more importantly, offer a healthier life for all members of society.

Declaration of interest: The authors declare that there is no conflict of interest.