The increasing effort to mitigate the environmental and nutritional challenges posed by the current food systems in Western countries has highlighted the need for individuals to adopt more plant-based diets(Reference Röös, Bajželj and Smith1). Legumes are of particular value in plant-based diets. Leguminosae or Fabaceae commonly known as legumes or pea family are crop plants with edible seeds that have been used for many years for human and animal consumption. Legumes include crops that are produced for fresh consumption as vegetables (green peas and green beans), some as dry grains (pulses: e.g. dry peas, beans and lentils), some for oil extracting purposes (soybean) and some for sowing purposes (alfalfa and clover)(2).
Legume consumption varies worldwide. Although its consumption has declined all around the world(3), they are still highly consumed in some areas such as South Asia, the Middle East and the Mediterranean area(Reference Miller, Rangarajan and Gupta4). Whereas, in Western Europe, their consumption remains low(Reference Micha, Khatibzadeh and Shi5). In the Nordic countries, including Finland, legume consumption is especially low(Reference Kaartinen, Tapanainen and Männistö6,Reference Steib, Johansson and Hefni7) compared with the Eat Lancet Commission’s reference diet (75 g/d), so called planetary health diet, which refers to both human health and environmental sustainability(Reference Willett, Rockström and Loken8). According to the latest national food consumption survey in Finland, FinDiet 2017, the mean consumption for legumes among women is only 13 g/d and among men 12 g/d, with cooked green peas (mainly in pea soup) being the most common leguminous food eaten (45 % share), followed by green beans, kidney beans and soya mince (each 10 %)(Reference Valsta, Kaartinen and Tapanainen9,Reference Kaartinen, Tapanainen and Maukonen10) . Legumes are a valuable source of complex carbohydrates and slowly digestible starch such as oligosaccharides and phenolic compounds as well as fibre(Reference Hall, Hillen and Garden Robinson11,Reference McCrory, Hamaker and Lovejoy12) and are rich in nutrients such as riboflavin, thiamine and especially folate(Reference Campos-Vega, Loarca-Piña and Oomah13). Therefore, regular legume consumption may reduce the risk of CVD and risk factors such as blood pressure, inflammation and lipid profiles as well as the risk of colorectal cancer(Reference Marventano, Izquierdo Pulido and Sánchez-González14–Reference Fechner, Fenske and Jahreis16).
Although considerable research has been devoted to the association between legume consumption and health outcomes, only a limited number of international studies have investigated legume consumption in relation to the consumption of other foods and intake of nutrients(Reference Steib, Johansson and Hefni7,Reference Mudryj, Yu and Hartman17–Reference Perera, Russo and Takata19) . Such studies would be of importance when planning epidemiological research around legumes and interpreting their results. Moreover, characterising diets of individuals with high legume consumption can play an important role in the identification of healthy diets already adopted in the population and thereby increasing public health knowledge of legume consumption and improving recommendations.
Two studies from the US using the National Health and Nutrition Examination data reported that, compared with non-consumers, consumers with the highest legume consumption had a higher intake of main legume-derived nutrients such as folate, Fe and Mg, as well as carbohydrates and a lower intake of total fat(Reference Mitchell, Lawrence and Hartman18,Reference Mitchell, Marinangeli and Pigat20) . A Canadian study(Reference Mudryj, Yu and Hartman17) using the 2004 Canadian Community Health Survey observed a positive association between legume consumption and energy, folate and Zn intake as well as the consumption of fruits and vegetables. Additionally, they found an inverse association for cholesterol intake for the highest legume consumers compared with non-consumers, whereas no association was observed with fats. These three studies mainly included pulses (dry legume grains). In the Swedish population, a positive association was observed between high consumption of legumes (compared with non-consumers) and the intake of energy, fibre, folate, Mg, potassium and Fe, using the national Riksmaten Survey data(Reference Steib, Johansson and Hefni7). Additionally, using the principal component analysis, they observed an association between high consumption of legumes and healthier food choices (such as fruits, nuts, seeds and tea) among women. They included pulses, fresh legumes, peanuts, soya products, sprouts and mixed meals with legumes in their study.
This study, for the first time in Finland and second in Europe (first study conducted in Sweden(Reference Steib, Johansson and Hefni7)), aims to investigate how legume consumption is associated with the consumption of other foods and intake of nutrients in adult men and women to strengthen the knowledge on the role of legumes in prevailing diets/dietary patterns in Nordic countries and Europe. Primarily, the findings of this study have implications for nutritional research, studying the associations between legume consumption and health outcomes. Further implications are for Finland and other countries to better characterise legume consumers.
Methods
Study population
FinHealth 2017 is a nationally representative study that gathered information on the health and well-being of the adult population in Finland(Reference Koponen, Borodulin and Lundqvist21,Reference Borodulin and Sääksjärvi22) . The FinHealth 2017 Study was conducted between January and May 2017 and comprised a thorough health examination and several health questionnaires including a FFQ. An invitation letter to a health examination along with the first health questionnaire was sent to the eligible sample obtained from the Finnish population register (n 10 247, age range 18 years and over). Of the eligible participants, 58 % participated in the health examination (n 5952). Of those who participated in the health examination, 89 % returned the FFQ.
After the exclusion of incomplete and incorrectly filled FFQ (n 119), those who withdrew written consent (n 7) and those outside the daily energy intake cut-off points (0·5 % of participants at both ends of the sex-specific daily energy intake distributions) (n 51), the final population for this study comprised 5125 participants (2875 women and 2250 men)(Reference Borodulin and Sääksjärvi22,Reference Männistö, Kaartinen, Maukonen, Borodulin and Sääksjärvi23) .
The study was conducted according to the guidelines laid down in the Declaration of Helsinki and the ethical guidelines of the Finnish Institute for Health and Welfare and Tampere University. All procedures involving human subjects were approved by the Ethics Committee of the Hospital District of Helsinki and Uusimaa (Reference 37/13/03/00/2016). Written informed consent was obtained from all subjects(Reference Borodulin, Borodulin and Sääksjärvi24).
Dietary data
All food group and nutrient data for this study were derived from a semi-quantitative FFQ enquiring about the participants’ habitual food consumption over the past 12 months. The reproducibility and validity of the FFQ, referring to the general adult population in Finland, has been documented and found to be reliable and valid for epidemiological studies(Reference Männistö, Virtanen and Mikkonen25–Reference Kaartinen, Tapanainen and Valsta27). The FFQ consists of 134 items on foods, mixed dishes and beverages generally consumed in Finland. The frequency of consumption was categorised into ten groups, ranging from never to at least 6 times a day. The sex-specific portion sizes were specified by natural units (for example, serving, slice, glass and cup) or weight/volume measures(Reference Borodulin and Sääksjärvi22). Daily consumption of foods (ingredient level, g/d) and nutrient intakes (g/d) were calculated using the National Food Composition Database (FINELI®) and the FINESSI software of THL(Reference Reinivuo, Hirvonen and Ovaskainen28).
The main exposure variable, legumes, was composed of three main categories: peas and beans, soya products and legume-based meat alternatives. The first category includes green and dry peas, beans and lentils, either frozen or cooked and prepared, including canned products. The soya products include tofu, crushed soya and other soya-based products. Legume-based meat alternatives refer to those products in which non-soya legumes are the main ingredients, such as fava bean-based meat alternative product. Of these three categories, the first two were the most important at the time of FinHealth 2017, since there are traditional dishes with peas still in common use in Finland, and vegan/vegetarians have used peas, beans and soya for a long time. Whereas, the plant protein products had just entered the market(Reference Jallinoja, Niva and Latvala29). Peanuts were included in nuts and seeds (not in legumes) due to their usual mode of consumption in the Finnish diet.
Food groups were chosen to characterise best the Finnish diet. Some of the food group variables in the same category were merged, to ease the analysis process and interpretation of results. For instance, among cereals, rye, oat and barley were merged and regarded as the healthy option, as they are the main whole grain sources in the Finnish diet, and wheat and rice as the less healthy option. In Finland, wheat is mostly consumed as refined, in the form of white wheat bread, pastries and sweets. Rice consumption is quite low, and polished rice is the most commonly consumed type(Reference Valsta, Kaartinen and Tapanainen9). Additionally, for milk products, products such as milk and yoghurt were categorised into liquid milk products. Cheese was categorised separately. This was done to examine whether there was, in general, a different pattern with regard to liquid milk products or cheeses in relation to legume consumption. Moreover, all vegetable-based fat spreads, such as margarine and oil (vegetable-based oil) regarded as healthy options, were merged as vegetable-based fat spreads and oil. Whereas, butter and other butter–oil mixtures, regarded as unhealthy options, were categorised as butter and butter-based fat spreads. The nutrients for the analysis were chosen based on the nutrient content of legumes, the nutrient intake status of the Finnish population and previous studies(Reference Steib, Johansson and Hefni7,Reference Valsta, Kaartinen and Tapanainen9,Reference Mudryj, Yu and Hartman17,Reference Mitchell, Lawrence and Hartman18,Reference Mitchell, Marinangeli and Pigat20) .
Socio-demographic and lifestyle factors and anthropometric measures
The socio-demographic and lifestyle data utilised in this study were obtained primarily from self-administered questionnaires(Reference Sääksjärvi, Borodulin, Borodulin and Sääksjärvi30). Other background data such as age and sex (for the FinHealth 2017 study) were obtained from the sampling frame (national population register)(Reference Koponen, Juolevi, Rissanen, Borodulin and Sääksjärvi31). Total years of education were categorised into tertiles (low, middle and high), taking into account the participants’ sex and birth year. This was done to adjust for the increase in school years and changes in the Finnish educational system during the past decades. Leisure time physical activity was initially assessed based on four categories: inactive (activities that are not physically straining like reading and watching TV), moderately active (activities like walking, light home gardening, fishing and cycling several h/week), active (activities such as running, swimming, fitness training, cross-county skiing and strenuous gardening several hours a week) and highly active (including competitive sports regularly and several times/week like running, cross-country skiing and ball games). The last two categories were merged because of the low number of participants in the highly active category. Smoking was initially categorised into four groups: non-smokers (never smoked or did not smoke regularly), quit ≥6 months ago, quit <6 months ago and current smoker. Because of the low number of participants who quit ≥6 months ago and quit <6 months ago, these categories were combined and renamed as ‘former smokers’.
Height (cm) and weight (kg) of the participants were measured during the health examinations by specially trained nurses, according to international standard protocols(Reference Tolonen32); height was measured to the closest 0·1 cm measure and weight was measured with participants wearing light clothes and no shoes to the closest 0·1 kg(Reference Borodulin, Råman, Borodulin and Sääksjärvi33). Body mass index (BMI) was calculated as weight (kg) divided by the squared height (m2).
Statistical analyses
Almost all participants consumed at least some legumes in the past year, since legumes are contained in some of the dishes included in the FFQ. There were only two participants who did not consume any legumes. Therefore, a separate non-consumer group was not included in this study. However, the two non-consumers were included in the lowest quartile of legume consumption. There are significant differences in dietary patterns in general, as well as legume consumption between women and men in Finland(Reference Valsta, Kaartinen and Tapanainen9,Reference Prättälä, Paalanen and Grinberga34) . Hence, analyses were conducted separately for women (n 2875) and men (n 2250). The participants (n 5125) were classified by sex, into quartiles (Q) of legume consumption (Q1–Q4).
Socio-demographic and lifestyle characteristics, including age, educational level, smoking status, leisure time physical activity level and BMI, were first compared across legume quartiles. Age and BMI were used as continuous variables. Median and interquartile range were calculated for quantitative variables and percentages for categorical variables. The P value of the associations was calculated using Kruskal–Wallis test for continuous variables and the χ 2 test for categorical variables.
Since the distribution of the food level outcome variables and nutrients were positively skewed, log10 transformation was chosen based on the shape of residuals’ distributions. To conduct logarithmic transformation and geometric means, “1” was added to all the values of those variables that had a value zero. To obtain the energy-adjusted geometric means of each food group and nutrient outcome (since the outcome variables were log transformed) for each quartile, legume quartiles were used as a categorical variable and each outcome and energy (kJ/d), as a continuous variable in the analysis of covariance.
To determine the association between legume consumption and other food groups and nutrients, multivariable linear regression was firstly used for all the variables. All food groups and nutrients were adjusted for energy intake by including energy intake in the models. This was done to focus on diet quality rather than absolute intake, thereby removing confounding by energy intake. We constructed two statistical models. The energy-adjusted model was constructed using each food group or nutrient at a time as the dependent variable and the medians of legume quartiles and energy intake (kJ/d), as independent continuous variables. The multivariable model was constructed using each food group or nutrient at a time as the dependent variable and the medians of legume quartiles, energy intake (kJ/d), age and BMI as independent continuous variables and educational level, smoking status and leisure-time physical activity as independent categorical variables.
For variables nuts and seeds, rye, oat and barley, liquid milk products, butter and butter-based fat spreads, fish and fish products, poultry and alcohol, the distributions of residuals of linear regression models were not fully normalised after transformation. Therefore, for these variables, the results were checked using multinomial logistic regression. Multinomial models were created by categorising these variables into ten ordered groups, each including approximately 10 % of cases. After that, both energy-adjusted and multivariable models were created using the categorised variable as dependent variable. All analyses were performed using the IBM SPSS statistical software, version 26.
Results
In this study, 56 % of the participants were women. The un-adjusted median legume consumption was low in both women (8·9 g/d, interquartile range 5·1–15·7) and men (9·6 g/d, interquartile range 5·8–15·7). However, there was a substantial variation in legume consumption: the median of the highest quartile was nearly eight times higher than that of the lowest quartile for both sexes. The medians for the Q1–Q4 were 3·0, 6·8, 11·3 and 26·0 g/d for women and 3·3, 7·5, 12·2 and 25·8 g/d for men, respectively. High legume consumption was positively associated with younger age, higher educational level and higher leisure-time physical activity level. Legume consumption was not statistically significantly associated with smoking and BMI (Table 1).
Table 1. The socio-demographic and lifestyle characteristics of the participants (women (n 2875) and men (n 2250)) by the quartiles of total legume consumption (median (interquartile range (IQR) or numbers (percentage %))
Q, quartile.
* Crude legume consumption range in each quartile (g/d) for women is: ≤5·09, 5·10–8·88, 8·89–15·66 and ≥15·67 and for men: Q1 (≤5·76), Q2 (5·77–9·61), Q3 (9·62–15·70) and Q4 (≥15·71).
† The P value of the associations was calculated using Kruskal–Wallis test for continuous variables and χ 2 test for categorical variables. P values are significant at 0·050.
‡ Educational level was obtained by categorising total years of education into tertiles, considering the sex and birth year.
§ Only leisure time physical activity was included.
There was a positive association between legume consumption and several food groups, such as fruits and berries, vegetables, nuts and seeds and vegetable-based fat spreads and oil (Table 2). For instance, the difference between the mean consumption of vegetables in the highest and lowest quartile for women was 151 g/d. Among men in the highest legume consumption quartile, the mean consumption of vegetables, 274 g/d, was twice that of the lowest quartile, 137 g/d. The highest quartile of legume consumers had the lowest intake of cereals. Additionally, there was a negative association between legume consumption and red and processed meat and butter and butter-based fat spreads in both sexes. For instance, for red and processed meat, the difference in the mean consumption in the lowest and highest quartiles was 17 g/d for women and 6 g/d for men.
Table 2. Food consumption (g/d) by legume consumption quartiles (g/d) for women (n 2875) and men (n 2250) (energy-adjusted geometric means and their 95 % confidence intervals (CI))
Q, quartile.
* Quartile medians (g/d) are not adjusted for energy intake since energy (kJ/d) was used in the model.
† The P values for the energy-adjusted model were obtained from multivariable linear regression using each food as a continuous dependent variable at time and the median of each legume quartile and energy (kJ/d) as continuous independent variables. P values are significant below 0·050.
‡ The P values for the multivariable model were obtained from multivariable linear regression using each food as a continuous dependent variable and the median of each legume quartile, energy (kJ/d), age and BMI (kg/m²) as continuous independent variables and education, smoking, physical activity as categorical independent variables. P values are significant below 0·050.
§ The geometric means were calculated using legume quartiles as a categorical variable, including each food and energy (kJ/d), as a continuous variable in the analysis of the covariance.
|| To conduct log transformation and geometric means, ‘1’ was added to these variables, as they included zero consumption values.
Legume consumption was positively associated with cheese in the energy-adjusted model, but this association attenuated (to non-significant) in the multivariable model. Additionally, the inverse association for rye, oat and barely was significant in the energy-adjusted model only. There was a negative association between legume consumption and potato among women. Among men, this association was positive and became significant in the multivariable model. Furthermore, the inverse association for chocolate and sweets in men became significant in the multivariable model (among women significant in both models).
The results of the multinomial models were essentially similar to the results of the linear regression models. For some variables, the statistical significance of the P values changed. For females, this occurred for variable rye, oat and barley in the multivariable model. For males, changes were in the following variables: rye, oat and barley (multivariable model), liquid milk products (energy-adjusted and multivariable model), butter and butter-based fat spreads (energy-adjusted model and multivariable model) and poultry (energy-adjusted model).
The associations between legume consumption and the intake of nutrients are presented in Table 3. The associations were statistically significant for most of the studied nutrients in both sexes. Legume consumption was positively associated with the intake of protein, PUFA, fibre, vitamins A, E and C, folate, thiamine, Mg, Fe, Zn and iodine in both models for both sexes. For example, the difference in the mean folate intake among women between the highest and lowest legume quartile was 93 µg/d. Among men, this difference was 79 µg/d, respectively. Legume consumption was also positively associated with MUFA intake (only among men) in the energy-adjusted model that attenuated to non-significant in the fully adjusted model. A positive association was also observed for riboflavin, but this changed to non-significant in men in the fully adjusted model. Higher legume consumption had a positive association with vitamin B12 intake only in men. Legume consumption was also positively associated with salt intake.
Table 3. Energy and nutrient intakes by the quartiles of legume consumption (g/d) for women (n 2875) and men (n 2250) (energy-adjusted geometric means and their 95 % confidence intervals (CI))
Q, quartile.
* Quartile medians (g/d) are not adjusted for energy intake since energy (kJ/d) was used in the model.
† The P values for the energy-adjusted model were obtained from multivariable linear regression using each nutrient as a continuous dependent variable at time and the median of each legume quartile and energy (kJ/d) as continuous independent variables. P values are significantly below 0·050.
‡ The P values for the multivariable model were obtained from multivariable linear regression using each nutrient as a continuous dependent variable and the median of each legume quartile (continuous), energy (kJ/d), age and BMI(kg/m²) as continuous independent variables and education, smoking, physical activity as categorical independent variables. P values are significantly below 0·050.
§ The geometric means were calculated using legume quartiles as a categorical variable, including each food and energy (kJ/d), as a continuous variable in the analysis of the covariance.
|| The values for energy variable are not additionally adjusted for energy.
On the other hand, there was a negative association between legume consumption and the intake of some nutrients such as SAFA and sucrose (for sucrose only significant among women). For instance, the difference in the mean SAFA intake between women in the lowest and highest quartile of legume consumption was 2·2 g/d. Among men, this difference was 1·8 g/d, respectively. Among the vitamins studied, legume consumption and vitamin D levels had a negative association among women in the energy-adjusted model. This difference attenuated to non-significant in the multivariable model. Overall, the differences in the results between the energy-adjusted model and multivariable model were small (Tables 2 and 3). However, in some cases, adjusting for socio-demographic and lifestyle variables changed the associations to non-significant or vice versa as described above.
Discussion
Higher consumption of vegetables, fruits and berries and lower consumption of red and processed meat were observed at the higher legume consumption levels. These differences in food consumption were also reflected in differences in the intake of nutrients between the legume consumption groups. Higher intake of fibre, folate and thiamine in women and men and riboflavin in women, as well as lower intake of saturated fatty acids and sucrose (for sucrose significant only in women), were observed at higher legume consumption levels.
Similar to earlier studies(Reference Steib, Johansson and Hefni7,Reference Mudryj, Yu and Hartman17,Reference Mitchell, Lawrence and Hartman18) one of the main findings of this study was that higher legume consumption was associated with healthier food choices. The highest quartile of legume consumers, regardless of their sex, consumed considerably more fruits and berries, vegetables, nuts, seeds and vegetable-based fat spreads and oil and less butter and butter-based fats spreads, as well as, red and processed meat, compared with the lowest quartile. In contrast to the findings of Mitchell et al.(Reference Mitchell, Lawrence and Hartman18) regarding cereal consumption in the USA population, a negative association was observed between legume consumption and total cereals. Among those food group variables for which multinomial regression models were constructed, although the P values changed, the overall associations described with odd ratios showed essentially similar type of associations as seen in linear regression models. The results of the linear regression for those variables in question should be interpreted with caution.
The results of this study are in line with the findings of the Swedish study,(Reference Steib, Johansson and Hefni7) and a US study that legume consumption is positively associated with the intake of fibre, folate, Fe, potassium and Mg. Unlike the findings of a US study that observed a negative association for riboflavin and no change in thiamine intake between the highest legume consumption and the non-consumers(Reference Mitchell, Marinangeli and Pigat20), our findings showed that legume consumption was positively associated with the intake of thiamine and riboflavin (for riboflavin only in women). These findings are of importance, since compared with the average requirements for vitamins and nutrients, at least one-fifth of the Finnish population has insufficient intake of folate, riboflavin and thiamine. Furthermore, among the Finnish adult population, fibre and carbohydrate intakes are lower than recommended in two-thirds of the population(Reference Valsta, Kaartinen and Tapanainen9). Interestingly, higher legume consumption was positively associated with vitamin B12 intake among men. This could be partially explained by the higher consumption of other animal products such as fish and cheese among men across quartiles. Unlike Mitchell et al., (Reference Mitchell, Lawrence and Hartman18), our findings revealed a significantly positive association between legume consumption and PUFA intake. Similar to our findings, the Canadian study(Reference Mudryj, Yu and Hartman17) reported higher intake of PUFA and MUFA among legume consumers and the highest in the third quartile compared with non-consumers. Based on our results, higher legume consumption was positively associated with MUFA intake only among men in the energy-adjusted model. These associations suggest that in the Finnish food culture higher legume consumption goes along with better food choices (food group results) and better nutrient intake profile (nutrient results). Therefore, the adoption of such diets more widely has the potential to improve public health nutrition. However, this needs to be further studied in different research settings.
Regarding nutrients that are of concern in the diet of the Finnish adult population, salt intake was significantly higher in the highest legume quartile in the present study. Similar findings have been observed by the Canadian study(Reference Mudryj, Yu and Hartman17). This could be possibly due to the generally high salt content in the Finnish legume food recipes and new legume-based meat alternatives, in addition to the salt content of the other food groups that are positively associated with legume consumption. Moreover, based on the findings of this study, the iodine intake was significantly higher in the highest quartile. Another potential concern is the lower vitamin D intake among women in the highest quartile compared with the lowest quartile (significant only in energy-adjusted model). This could be partly explained by the significantly lower consumption of liquid milk products among the women in the highest quartile of legume consumption since fortified milk products are one of the major food sources of vitamin D in Finland. Among men, although vitamin D intake increased across quartiles (highest in the third quartile), the association was not statistically significant.
Legume consumption might vary across demographic groups in different populations. There was a negative association between legume consumption and age in both men and women. Most previous studies examining legume consumption and age, however, have reported a positive association(Reference Steib, Johansson and Hefni7,Reference Mudryj, Yu and Hartman17–Reference Perera, Russo and Takata19) . On the other hand, a US study (Reference Mitchell, Marinangeli and Pigat20) showed that most of the legume consumers are aged 31–70 years. In contrast, among the Canadian population,(Reference Mudryj, Yu and Hartman17) no association was found when adjusted for energy intake. Contrary to the US studies (Reference Mitchell, Lawrence and Hartman18–Reference Mitchell, Marinangeli and Pigat20) and similar to the Swedish population(Reference Steib, Johansson and Hefni7), there was a positive association between legume consumption and level of education. However, most of the Swedish participants had a level of education higher than the elementary level, and their legume quartiles were adjusted for energy intake when assessing the socio-demographic and lifestyle characteristics(Reference Steib, Johansson and Hefni7). Our findings showed that legume consumption was positively associated with leisure-time physical activity, but no association was found with smoking and BMI. The present study was the only study that examined the association between legume consumption and leisure-time physical activity and considered it as a confounder. In line with earlier findings in Sweden and Canada(Reference Steib, Johansson and Hefni7,Reference Mudryj, Yu and Hartman17) , although the total energy intake was higher in the highest quartile of legume consumption, BMI did not differ significantly across the quartiles of legume consumption. Our results could be partially explained by the higher intake of healthier food groups and higher leisure-time physical activity in the highest quartile. Overall, more research is needed to confirm these findings, especially regarding age, educational level and physical activity. Our results suggest that legume consumption might vary across demographic groups across the Finnish population. In fact, it has been reported that the highest educated group in Finland obtains more nutrients not only from vegetables, fruit and berries but also from legumes and nuts compared with the lowest educated group(Reference Valsta, Tapanainen and Kortetmäki35).
There are several differences between this study and previous studies. Earlier studies utilised methods of data collection other than FFQ for nutritional data, such as 24 h dietary recall. Hence, they compared consumers to non-consumers as well as across consumption quartiles and reported absolute intakes. Despite the use of different methods, meaningful and similar associations were identified. Additionally, due to considerable variations in the dietary guidelines of countries, legumes are categorised differently in the literature. The Canadian(Reference Mudryj, Yu and Hartman17) and USA studies(Reference Mitchell, Lawrence and Hartman18–Reference Mitchell, Marinangeli and Pigat20) looked specifically at pulses (dry legume grains), whereas the Swedish study(Reference Steib, Johansson and Hefni7) included pulses, fresh legumes, peanuts, soya products, sprouts and mixed meals with legumes in their study. In our study, all types of legumes currently consumed were considered, due to overall low legume consumption in this population. It is important to note that food pattern and food culture differences might account for differences between the studies.
A limitation of this study is the self-reported data; hence, there is a possibility for over- and under-reporting. Another limitation is the lack of absolute measures, as we utilised FFQ and semi-quantitative data. Therefore, we were not able to compare our results with recommendations and dietary guidelines. Nevertheless, FFQ is the most appropriate method for assessing dietary habits over a long period for epidemiological study purposes and is designed for studies concentrating on the association of diet with various health outcomes(Reference Willett36). The strengths of this study are the validated FFQ, the large population-based sample, as well as measured body weight and height. Additionally, the data collected for the FinHealth 2017 were detailed and varied, allowing us to consider several food and nutrient variables and confounders in this study. The relatively high participation rate (58 %) in this study, in comparison with other international studies, increases the generalisability of these findings to the Finnish adult population. Moreover, the present study was the only study that adjusted the association between legume consumption and food groups and nutrients for central socio-demographic and lifestyle confounders.
Overall, in line with the results for food groups, legume consumption is associated with better nutrient intake profile in the Finnish adult population. The results thus reinforce the general view that diets with higher legume consumption appear to be more balanced. In some cases, when the association was additionally adjusted for socio-demographic and lifestyle confounders, the associations lost their significance such as in cheese or gained significance for potato and chocolates and sweets (both among men). Therefore, these confounders are central when studying diet and dietary patterns. Moreover, the results have implications for the planning of future epidemiological research on the associations of legume consumption and health outcomes and their interpretation.
In conclusion, more uniform studies are needed from Nordic countries on legume consumption to confirm recent findings regarding the association of legumes with other healthy and less healthy aspects of diets. Additionally, this study offers important background information for planning intervention studies and programs as well as stakeholders wishing to increase legume consumption. For instance, health policy actions or future interventions could benefit from these results. Overall, increasing legume consumption is an important goal, also from the food system perspective. Legumes maintain ethical synergies by contributing to nutrition security while promoting human health as well as having beneficial environmental impact(Reference Alsaffar37). Characterising legume consumers can play an important role in increasing public health knowledge of legume consumption and improving national and international dietary recommendations.
Conclusions
Overall, our findings suggest higher legume consumption has positive association with the consumption of foods considered healthy as well as nutrient intake profile, in the adult population of Finland. At the same time, higher legume consumption is associated with higher salt intake in both women and men. Increased legume consumption appears to be also associated with higher education and a healthier lifestyle. Additionally, taking into consideration the potential of legumes in Finnish food system and agriculture(Reference Stoddard, Hovinen and Kontturi38) and according to the EAT Lancet Commissions guidelines(Reference Willett, Rockström and Loken8), the production and consumption of legumes in the general adult population of Finland need to be promoted and increased. The practical relevance of our findings needs to be elucidated and studied further, e.g. in longitudinal settings, so that dietary recommendations for legumes could include better advice to mitigate or enhance their impacts.
Acknowledgements
The authors acknowledge the experts and staff of FinHealth 2017 Study organisation and the Fineli team for keeping the food composition database up to date. The authors thank the participants of the FinHealth 2017 Study.
This research was funded by the Strategic Research Council at the Academy of Finland (Grant Nos. 327699, 352483 and 327698). The Strategic Research Council at the Academy of Finland had no role in the design, analysis or writing of this article.
N. E. K., S. M. and M. M. contributed to the design and implementation of the FinHealth 2017 Study and to acquiring funding for the project. N. E. K., T. K., A. K. and A. G. K. planned the content and analysis for this study. A. G. K. performed the statistical analyses under supervision of A. K. A. G. K. wrote the manuscript under supervision of T. K. and N. E. K. All authors critically appraised the paper and accepted the final version of the manuscript.
There are no conflicts of interest.
