Eating habits at the population level have changed dramatically in Finland since the early 1970s. Intakes of total and saturated fats were high compared to other European countries(Reference Pietinen, Lahti-Koski and Vartiainen1, Reference Pietinen, Vartiainen and Seppanen2). In 40 years, the proportion of total fat has decreased from 38 %(Reference Pietinen, Vartiainen and Seppanen2) to 33 %(Reference Pietinen, Lahti-Koski and Vartiainen1, Reference Paturi, Tapanainen and Reinivuo3) and that of saturated fat from 21 % to 13 %(Reference Paturi, Tapanainen and Reinivuo3). High-fat milk and butter have largely been replaced by skimmed milk and margarine. Simultaneously, vegetable and fruit consumption has increased substantially(Reference Pietinen, Lahti-Koski and Vartiainen1, Reference Paturi, Tapanainen and Reinivuo3, Reference Puska4), but does not reach the national nutrition recommendations(5).
In the 1980s, the proportion of fat was still high (39–41 %) and consumption of potato, vegetables, cereals and milk products abundant among 18-year-old Finnish men(Reference Räsänen, Ahola and Kara6, Reference Räsänen, Laitinen and Stirkkinen7). After a 21-years’ follow-up among these men, the overall proportion of fat was 36 % and the consumption of vegetables and fruit was higher than in youth(Reference Mikkilä, Räsänen and Raitakari8).
These findings reflect positive changes in the prime everyday i.e. core diet. Despite these improvements, the prevalence of obesity and overweight has increased especially in young adults and adolescents in Finland(Reference Lahti-Koski, Harald and Männistö9). Among 18-year-old boys, the prevalence of overweight (BMI > 25 kg/m2) and obesity (BMI > 30 kg/m2) increased 2·5-fold in 1977–1999(Reference Kautiainen, Rimpelä and Vikat10).
One possible explanation for this development is excess consumption of extra foods that are not recommended in the daily diet e.g. fast food, sweets and soft drinks. In a population-based study from 2002, snacking-type eating was found frequently among Finnish men of whom 36 % ate at least four snacks daily(Reference Männistö, Ovaskainen and Valsta11). A more recent limited study among young men in military service revealed a preference for energy-rich, nutrient-poor foods like pizza and soft drinks(Reference Bingham, Ovaskainen and Tapanainen12). More recent population-based data on dietary habits of Finnish young men is lacking.
Furthermore, neither the overweight epidemic(Reference Lahti-Koski, Harald and Männistö9) nor healthy eating habits are equally distributed in different population groups(Reference Roos, Talala and Laaksonen13). Those with lower socio-economic status are more overweight(Reference Kautiainen, Koivisto and Koivusilta14) and most studies suggest that among adults higher education, occupational class and household income are related with healthier food habits(Reference Lallukka, Laaksonen and Rahkonen15). In addition, adults’ health habits tend to cluster i.e. unhealthy eating, smoking and lack of physical activity are associated(Reference Laaksonen, Talala and Martelin16, Reference Mikkilä, Räsänen and Raitakari17).
Eventually, population-based studies focusing on different aspects of eating habits and their socio-economic background and health behaviour correlates among young men are lacking. The present study aims to describe Finnish young men’s quality of diet, and to determine how background and health behaviour factors explain young men’s eating habits.
Subjects and methods
In Finland, military service is compulsory for 18–29-year-old men and 80 % of them complete it(18). This study was conducted as a part of the DefenceNutri research project, which is a controlled two-phase intervention trial taking place in two garrisons during 2007–2009. The garrisons are Armoured Brigade in Southern Finland (Southern Brigade) and Kainuu Brigade in Northern Finland (Northern Brigade). The study protocol has been approved by the ethics committee of the Hospital District of Helsinki and Uusimaa.
In the present study, baseline data on eating habits of two cohorts of men entering military service in January 2007 and July 2007 are presented. One month before the start of service, questionnaires were sent home to all men (n 6605) starting their military service at the garrisons. Men entering Southern Brigade live mostly in Southern Finland and the majority comes from big and medium size cities. In comparison, Northern Brigade recruits men from Western, Central and Northern Finland, and the majority living in semi-urban areas.
Responding to the questionnaire was possible by Internet or by returning it when entering service. Altogether, 3034 men replied (45·9 %). Incompletely filled questionnaires (n 23) were discarded. In addition, those respondents whose age was ≥ 22 years (n 70) and whose basic education could not be classified (n 36) were not included. The final number of respondents included was 2905. Their characteristics are shown in Table 1.
Characteristics of the study population (n 2905, all male)
*Calculated from self-reported height and weight.
The questionnaire covered sociodemographic background, health behaviours and psychosocial factors. Regarding eating habits, it included questions on food choices and a 36-item FFQ in which consumption was assessed as number of days during the previous week. Items of the questionnaire represented all major food groups of the diet. In the present study, the following data were used: basic education, self-reported weight and height to compute BMI and smoking habits; frequencies of physical exercise and eating breakfast, FFQ, overall monthly eating frequency of fast food (kebab, hamburgers, pizza, French fries and meat pastries), types of bread fat and milk. To control for regional and seasonal dietary variation, garrison and season (winter-summer) were taken into account.
In order to characterize quality of diet of young men, two separate indexes were formed. One was formed to measure daily healthy eating habits (core food index, CFI). It was designed to measure compliance to national nutrition recommendations(5) that promote everyday healthy food choices such as low-fat milk, vegetables and rye bread. The other index was designed to measure consumption of food items not recommended for daily use (extra food index, EFI, Table 2). It was built on previous knowledge on Finnish conscripts’ diet indicating that extra foods such as pizza are the most significant sources of saturated fat and soft drinks of sucrose(Reference Bingham, Ovaskainen and Tapanainen12). Thus, the ultimate purpose of the indexes was to capture the core elements of a healthy diet as defined in dietary recommendations and the most significant sources of extra energy acquired by young men. In addition, aspects to be noticed in formation of indexes (items included in the indexes, continuous scoring and defining purposes of the indexes) were catered for(Reference Waijers, Feskens and Ocke19).
Composition and scoring of core food index (CFI) and extra food index (EFI)
*Number of points equal to consumption frequency as number or days per week.
†Number of points equal to consumption frequency of all fast food as times per month. If consumed more than nine times per month, 10 points were given.
The five dietary factors included in CFI (weekly consumption frequencies of rye bread, fresh vegetables and fruit and berries and type of milk and bread fat) and the four factors in EFI (weekly consumption frequencies of soft drinks, chocolate, candy; monthly consumption frequency of all fast food) and their scoring are shown in Table 2. All items describing weekly consumption were given 1 point for each day the food item was used per week (range: 0–7). For the overall monthly consumption of fast food, all values >9 were combined into one category, resulting in a scale of 0–10 points. In total, the score of CFI ranges between 0 and 30 and of EFI between 0 and 31. Scoring of the indexes gives a high value for CFI when the diet consists of healthy food choices and a high value for EFI when the diet includes plenty of extra foods.
Statistical analyses were performed using Statistical Package for Social Sciences statistical software package version 15 (SPSS Inc., Chicago, IL, USA). First, descriptive statistics of eating habits and correlations between background and health behaviour variables were calculated. Next, univariate associations between index scores and background and health behaviour were analysed using general linear modelling (GLM). Categorical variables were inserted into the model as fixed factors and continuous variables as covariates. In subsequent multivariate analyses, all statistically significant variables from univariate analyses were first incorporated as independent variables to a GLM and the index score as the dependent variable (Model 1). Then, non-significant variables were dropped and only significant variables retained (Model 2). CFI and EFI were analysed separately at all times.
Results
Weekly food consumption frequencies are presented in Table 3.
Food consumption frequencies
Consumption frequency as number of days per week that food item has been used.
*Classified as fast food.
Regarding core foods, rye bread was the most frequently consumed cereal product as 30 % ate it on 6–7 d/week. Daily consumption of fruit and berries and vegetables was infrequent, 8 % and 13 %, respectively. Fresh vegetables were preferred to be cooked ones.
Red meat was the most frequently eaten type of meat. Sausages and chicken were also relatively frequent in the weekly but not in the daily diet. Consumption of fish was rare since 90 % of the men reported eating it on 0–1 d/week. Of dairy products, milk and sour milk were most frequently consumed belonging to the daily diet of 65 % of the men. In addition, cheese was usually consumed daily.
Regarding all fast food classes (French fries, crisps, pizza, kebab, hamburgers, hot dogs, meat pastries and pastries), a clear minority consumed them more than once a week. Candy was preferred to chocolate (46 % v. 26 %, respectively, > 1 d/week) and sugar-sweetened soft drinks to artificially sweetened (‘light’) ones. Beer was the most common alcoholic beverage and others were consumed marginally.
Effects of background and health behaviour factors on CFI are shown in Table 4.
Effect of background and health behaviour factors on CFI
r.c., reference class.
In Model 1, all univariately significant factors were included. In Model 2, all factors significant in Model 1 were included.
The mean CFI value was 14·4. Of background factors, basic education explained CFI the most (R 2 = 0·078) and of health behaviour factors frequency of eating breakfast (R 2 = 0·100). All background and health behaviour factors had significant (P < 0·04) effect on CFI in univariate analyses. CFI score was higher in summer than in winter (P < 0·001) and in Southern Brigade compared to Northern Brigade (P < 0·001). CFI also increased significantly with basic education (P < 0·001 between all classes). BMI was inversely associated with CFI (P = 0·03). Regular smokers had a significantly lower score than in other smoking classes (P < 0·001). As frequencies for physical exercise and eating breakfast increased, so did the score (P < 0·001 between all classes). The score was higher for those drinking beer once a week (P = 0·006) and non-drinkers (P < 0·001) comparing to more frequent beer drinkers.
In the first multivariate model (Model 1), all factors except region, BMI and beer drinking frequency remained significant. The coefficient of determination was 0·205 for the model. In Model 2, all independently significant background and health behaviour factors were included. All factors in the model explained CFI significantly (P < 0·003 for all explanatory variables) and the associations found were similar to those observed in Model 1. The coefficient of determination was similar to that in Model 1 (R 2 = 0·200).
Effects of background and health behaviour factors on EFI are shown in Table 5.
Effect of background and health behaviour factors on EFI
r.c., reference class.
In Model 1, all univariately significant factors were included. In Model 2, all factors significant in Model 1 were included.
The mean EFI was 9·1 with frequencies of drinking beer (R 2 = 0·031) and eating breakfast (R 2 = 0·026) as the most important factors contributing to the score. In univariate analyses, all background and health behaviour factors except region had significant (P < 0·001) effect. The index score was higher in winter (P < 0·001) than in summer. The score was higher for those having comprehensive school as basic education than for those having completed upper secondary school (P = 0·001). Interestingly, EFI was inversely associated with BMI (P < 0·001). Regular smokers had a significantly higher score than in other smoking classes (P < 0·001). Furthermore, the score increased significantly as physical exercise frequency decreased (P < 0·001 between classes). The index score was lowest for those drinking beer once a week. Finally, the mean index score increased also as frequency of eating breakfast decreased. The coefficient of determination remained relatively low (R 2 ≤ 0·03) for all factors.
In Model 1, the background and health behaviour factors, which were significant in univariate analyses, were included. All explanatory variables in the model except basic education remained significant. Effects of season and BMI were similar as in univariate analyses. EFI was higher for occasional smokers than for regular smokers (1–10/d); R 2 for the model was 0·082.
In Model 2, the coefficient of determination (R 2 = 0·083) was equal to that in Model 1. The effect of all factors on EFI was highly significant (P ≤ 0·01 for all explanatory variables).
Discussion
In the present study, examination of consumption frequencies of several core foods and extra foods showed that daily consumption of fruits, berries and vegetables is rare, but rye bread is favoured. In addition, milk, sour milk and cheese form a part of the daily diet. Sugar-sweetened soft drinks are preferred to artificially sweetened ones and candy to chocolate.
When healthy core food eating was analysed by CFI, background factors and other health behaviours had significant associations with core food eating. These eating habits varied notably between different classes of explanatory variables. BMI was negatively associated with CFI in univariate analysis, but not in multivariate analyses. As with CFI, all health behaviour factors were related to EFI, but of background factors; season did not predict this type of eating. Furthermore, individuals with a lower rather than higher BMI ate extra foods significantly more often.
The low use of fruits and berries found in the present study resembles that of 15–24-year-old Finnish men(Reference Helakorpi, Prättälä and Uutela20). Only 8 % used fruit and berries and 13 % fresh vegetables daily or almost daily, which is far from the recommended 400 g/d(5). Fresh vegetables were more frequently consumed than cooked ones, which is in line with previous studies among Finnish men(Reference Paturi, Tapanainen and Reinivuo3, Reference Männistö, Ovaskainen and Valsta11, Reference Helakorpi, Prättälä and Uutela20). Rye bread was the most common bread and similar results have been reported among 25–34-year-old men too(Reference Paturi, Tapanainen and Reinivuo3).
Several international diet quality scores exist, some of which were widely used(Reference Haines, Siega-Riz and Popkin21–Reference Trichopoulou, Kouris-Blazos and Wahlqvist28). However, their application to the present study has limitations as they do not take into account special characteristics of the Finnish diet, e.g. uses of rye bread and berries. Three Finnish indexes also exist. Two of them have been used to evaluate intake of saturated fat(Reference Roos, Ovaskainen and Pietinen29) that offers a constricted scope for use in the present study. The third one was formed to describe the quality of diet of subjects who had been followed from childhood. Its adaptation for our study of young men is not feasible because of our population’s homogeneous sex, age and situation-of-life structure. Formation of CFI and EFI also caters for previous knowledge of a resembling population(Reference Bingham, Ovaskainen and Tapanainen12).
Seasonal differences in eating habits exist as found in the present study. In general, the consumption of fruits, berries and vegetables is high in the summer. Regional differences like higher consumption of vegetables in Southern Finland have been found earlier(Reference Paturi, Tapanainen and Reinivuo3). Previous studies among Finnish men have shown that daily consumption of vegetables is higher among those with higher education and income than among men with low education and income(Reference Roos, Talala and Laaksonen13) and the first-mentioned have lower energy intake(Reference Roos, Prättälä and Lahelma30). The basic education was the second strongest explanatory factor for CFI. Breakfast eating habits explained variation in CFI the most.
An earlier study that followed Finnish children to adulthood found no differences in eating habits based on either the subjects or their parents’ educational level. However, the study also showed that physical activity level was associated with eating habits(Reference Mikkilä, Räsänen and Raitakari8), an association confirmed also in our study, and for both types of eating habits. This same pattern was also extended to other health behaviours. Both CFI and EFI were negatively associated with BMI. Regarding EFI there are several underlying reasons. The vast majority of these men are of normal weight and further analysis requires examination of this association in subgroups according to physical activity. It may be that among young men there are other food items popularly consumed which were not measured here. In addition, consumption quantities that are not reported here may explicate this association. It is worth noticing that the finding does not indicate causality.
There are also some limitations to the present study. First, the response rate is moderate being 46 %. Still, it is at the same level as the current response rate of 15–24-year-old Finnish men in a large health behaviour population study, and responding increases with educational level(Reference Tolonen, Helakorpi and Talala31), age(Reference Eaker, Bergstrom and Bergstrom32, Reference Jackson, Chambless and Yang33) and socio-economic status(Reference Jackson, Chambless and Yang33). Moreover, as 20 % of men do not serve in the military due to medical, ethical and religious reasons(Reference Multimäki, Parkkola and Sourander34), conscripts form a somewhat selected population. Second, the coefficients of determination for both CFI and EFI models are relatively modest. This is probably due to other underlying explanatory factors such as family situation, subjects’ parental education and household income(Reference Laaksonen, Talala and Martelin16). Of these, especially the latter is exposed to change as the study population is in a transition period between youth and independent adult life.
However, despite these limitations, the present study offers new knowledge on young Finnish men’s quality of diet and food choices. Clustering of physical inactivity and smoking have been reported(Reference Laaksonen, Talala and Martelin16, Reference Mikkilä, Räsänen and Raitakari17), but the additional association of drinking beer and breakfast eating habits on quality of diet have not been studied earlier among Finnish young men. Furthermore, the relationship of education and eating habits has been established earlier(Reference Lallukka, Laaksonen and Rahkonen15, Reference Laaksonen, Talala and Martelin16). However, a notable finding is that educational differences were found to affect only core daily eating. It should also be remembered that dietary items included in our indexes represent active personal choices regarding quality of diet.
Future research should explore in more detail the components of young men’s core food and extra food choices. Further research is required to analyse the development of the quality of diet among young men in later adulthood and also during military service.
Acknowledgements
Professor Hely Tuorila and Raisa Valve are thanked for viewing and commenting this paper. Erkki Kronholm, Erkki Alanen, Pauli Puukka and Juhani Mäki are thanked for statistical guidance.
