One aspect investigated related to the participant’s self-perception about the practice of a balanced diet. Unbalanced diets lead to malnutrition, referring either to under-nutrition or to over nutrition. Nevertheless, very often malnutrition is associated with excessive calories in relation to requirements, coupled with low intake of micronutrient-rich foods [54
]. Food security entails aspects such as availability, utilization and access, but these may not necessarily cover nutritional aspects. In fact, easily accessed and available food products may not necessarily provide a healthy and balanced diet. The Family Nutrition Guide of the FAO (Food and Agriculture Organization of the United Nations) [55
] states that a balanced diet should be able to provide the right amount of food, energy and nutrients needed during the day to cover the dietary requirements of a particular person. Additionally, a balanced diet must include a variety of foods from different food groups to ensure the supply of all the nutrients needed for adequate body functioning, thus significantly contributing to a healthy life, while at the same time lowering mortality incidences related to malnutrition. Because a healthy diet needs to be balanced, the participants were asked about the frequency with which they practiced a balanced diet.
shows the results obtained for this question, considering the possible effect of some sociodemographic variables, assessed through chi-square tests. When considering the global sample, it was found that nearly half of the participants (46.9%) believed they frequently practice a balanced diet, while 16.5% responded that they always have a balanced diet. A considerable fraction (30.0%) responded to eating a balanced diet only sometimes.
Additionally, Table 4
also shows that no significant differences were found between sexes (p
= 0.060), but considering the two countries involved significant differences were encountered (p
< 0.0005), with participants from Portugal reporting a higher frequency of balanced diet as compared with Brazilians, although the association between these two variables (balanced diet and country) was weak (V = 0.198).The results in Table 4
also show that high percentages of underweight and obese participants admitted to practicing a balanced diet only sometimes (45.7% and 37.7%, respectively), and people who are overweight believed they frequently consume a balanced diet (50.7%). These differences were significant (p
= 0.001), but with a low association (V = 0.116). Considering age classes, an significant number of young adults state that they have a balanced diet only sometimes (41.6%), but the care for a balanced diet increases for the elderly, with 58.3% saying that they always have a balanced diet. These differences were significant (p
< 0.0005). Older people tend to have more health problems, sometimes suffering from non-communicable and age related diseases, thus making it more important to have a healthier food intake, either because they want to remain healthy or to diminish the morbidity associated with their diseases. Results from countless studies have revealed that healthy dietary patterns are associated with a reduced risk of developing several chronic diseases or having a better control of these, such as obesity, dyslipidemia, hypertension and diabetes mellitus. Hence, for the elderly, nutrition assumes a particular role in the maintenance of acceptable health standards and functional capacity [56
]. As for marital status, the widowed showed a higher frequency of balanced diet as compared with the other groups (single, married or divorced), with 61.5% having a balanced diet frequently and 15.4% always. These results are in line with those of age class, since most often the widowed consisted of older people, who had already lost their spouses. The differences are significant (p
= 0.009), although the association between balanced diet and marital status is weak (V = 0.102). Finally, the results showed that higher education levels are associated with higher frequency of practicing a balanced diet, with the percentages of participants who report this frequently increasing systematically from 20.0% for people with the lowest level of education completed, up to 47.8% for participants with a university degree. Again, these differences were significant (p
< 0.0005). These results clearly indicate that more educated people are better aware of the importance given to a proper diet to benefit from a globally healthier status [58
Another aspect investigated in this research was the satisfaction with body weight, and the results are shown in Table 5
for the global sample and also according to BMI classes and to the practice of a balanced diet. Most participants (40.1%) believed they are of normal weight but would still like to lose 2 to 5 kg.
Considering the underweight participants, 42.9% are aware of their condition, and stated they would like to gain a few kilos (Table 5
). Of the participants with normal weight, an important part (39.5%) is satisfied with their body weight and wish to maintain it, while 46.4% would like to lose 2 to 5 kg, even though they know their weight is normal. For those participants who are overweight, a high percentage (48.9%) wrongly think they have normal weight but would still like to lose 2 to 5 kg, and 32.1% are conscientious about being overweight. From the obese participants, 75.4% know about their condition and would like to lose a few kilos, but say they have tried and not succeeded. These results are interesting, because they reveal that, in general, the participants were aware of their bodyweight status.
The results in Table 6
refer to the practice of physical exercise and also the level of activity related to active daily traveling, by walking or riding a bike, for example. Most participants (38.8%) practiced exercise occasionally, i.e., once a week, but a considerable part (33.9%) does 2 to 3 times a week, which is the frequency recommended for most cases. Nevertheless, 17.4% of participants never do physical exercise, which can have a very negative impact on health at so many levels [59
]. Regarding active traveling, for example to and from work, most participants (52.2%) do not or do less than half an hour per day, but 33.8% spend between 30 min and 1 h daily in these activities.
presents the results obtained for the hours spent daily on sedentary activities. The results indicated that a high percentage (32.4%) spend between 1 and 2 h watching TV, and 13.8% spend up to 5 h watching TV. The participants do not engage much in activities like playing with computers or mobile phones or on social networks, i.e., screen entertainment, with 39.9% spending less than 30 min per day in those activities. On the other hand, the use of computers for professional purposes is a reality for the great majority of the participants, with 30.4% spending between 2 to 5 h, and 39.6% spending more than 5 h daily working on the computer. Finally, the daily hours of inactive traveling, for example in a car, motorbike, bus or train, are low, less than 30 min for 54.2% of participants, or between 30 min and 1 h for 30.4% of participants.
3.3.1. Factor Analysis for Lifestyle
Some variables related to lifestyle were subjected to exploratory FA, starting with Principal Component Analysis (PCA), to identify a possible grouping structure between the variables used to evaluate aspects related to physical activity or sedentary lifestyle of the participants.
The correlation matrix showed some correlations between the variables, although they were relatively weak. The value of KMO was low (0.496), but the results of the Bartlett’s test of sphericity indicated adequacy for applying FA (p
< 0.0005), thus leading to the rejection of the null hypothesis that the correlation matrix was equal to the identity matrix. Analysis of the anti-image matrix revealed that values of MSA (Measure of Sampling Adequacy) were close or above 0.5, meaning that in general the variables could be included in the analysis. The solution obtained by rotation of FA with PCA originated three components, explaining 59.7% of total variance, distributed by the three factors as: F1—22.4%, F2—19.3% and F3—18.0%. All variables had communalities higher than 0.4: the variable practice of physical exercise had the highest value (0.827, indicating that this variable had 82.7% of its variance explained by the solution), while the variable with lowest communality was daily hours working on computer (0.509). Rotation converged in five iterations and extracted three factors, grouping the variables as shown in Table 8
Considering the components in Table 8
, F1 was clearly linked to activities related to screens, where watching TV and playing on screen devices have positive high loads, while the use of computer for work has a considerable load, but negative, indicating that these variables contribute strongly to the definition of the factor. Factor F2 is strongly linked with variables related to daily hours of traveling, either active or inactive, and finally F3 is very strongly related with only one variable, which is physical exercise.
3.3.2. Cluster Analysis for Lifestyle
The factors identified though FA were subject to CA in order to perceive if there was a cluster structure among the people surveyed. Cluster analysis was based on the three factors resulting from FA and started with four hierarchical methods that indicated that four clusters was the most suitable grouping structure for this set of data. Following that, the k-means method was applied using as initial solutions those obtained with the hierarchical methods. In all cases, the k-means cluster analysis produced clustering variables with means that differ significantly, as indicated by ANOVA since p
-value < 0.0005 for the three input variables, i.e., factors F1 to f3. The results obtained for the cluster centers and number of members are presented in Table 9
and show that the four initial solutions tested converged to a similar final solution. Because the final solutions obtained from WARD and AL-BW methods converged to the exact same one, this was then considered as the final solution, which is characterized by:
Cluster 1: individuals with strong focus on F1 and negative input for F2 and F3, i.e., those whose lifestyle is very much dominated by screens. These were named screeners;
Custer 2: individuals with strong focus on F3 and negative input for F1 and F2, i.e., those whose lifestyle is very influenced by physical exercise. There were named exercisers;
Cluster 3: individuals with strong focus on F2 and negative input for F1 and F3, i.e., those whose lifestyle is very strongly dominated by daily travelling hours. There were named travelers;
Cluster 4: individuals with negative input for all three factors, i.e., those whose lifestyle is inversely associated with screens, travelling and exercise. Because these participants did not present a specific feature, this group was named others.
3.3.3. Cluster Characterization
To better understand the type of people who fall into each of the four categories that cluster analysis indicated, cross tabulation between cluster membership and the sociodemographic and behavioral variables was undertaken (Table 10
The results in Table 10
show that Portuguese participants are essentially exercisers (40.2%) while Brazilians were travelers (32.8%). While a major part of the women surveyed were exercisers (30.4%), men were mostly travelers (35.0%). Regarding the age class, young adults and the elderly were mostly screeners (44.2% and 33.3%, respectively), while middle aged adults were exercisers (36.5%), and senior adults were travelers (38.4%). While participants with a university degree were exercisers and travelers (32.8 and 33.1%, respectively) those with lower levels of education were mostly screeners (60.0% for primary school, 42.9% for basic school and 36.6% for secondary school). As for marital status, most single participants were screeners (31.7%), while those married, divorced and widowed were exercisers and travelers, in relatively similar percentages.
shows the cross tabulation between cluster membership and some anthropometric and lifestyle variables, specifically BMI and frequency of practicing a balanced diet. Results showed that underweight people are essentially screeners (37.1%) and travelers (34.3%), those with normal weight are exercisers (31.7%) and travelers (32.4%) and so are the overweight (34.4% exercisers and 32.6% travelers). The obese are equally distributed by clusters 1 (screeners: 28.1%), 2 (exercisers: 27.2%) and 3 (travelers: 28.9%). Regarding the practice of a balanced diet, those who practice it frequently or always are exercisers and travelers, while those who never or seldom have a balanced diet are screeners or exercisers.
3.4. Consumption Habits Regarding Dairy Products
shows the consumption habits of dairy products in Portugal and Brazil. Concerning milk consumption, for both countries the percentage of participants who never consume milk products is high, ranging between 46.7% (Brazil: never consume semi skimmed milk) to 94.8% (Portugal: never consume enriched milk). For those who consume milk, it was observed that 52.6% of Portuguese and 53.3% of Brazilians consume semi skimmed milk revealing a similar trend in both countries, although with variability according to the frequency of consumption. In Portugal, a higher frequency was observed for consumption of semi skimmed milk once a day (20.2%) and for Brazil for once a week (24.2%). On the other hand, the consumption of skimmed milk presents bigger differences between countries: 49.1% of Brazilians and 35.2% of Portuguese consume it, and in both countries a higher percentage of participants consume it rarely (13.8% and 27.9%, respectively, for Portugal and Brazil). This survey shows that the consumption of chocolate flavored and enriched milks is very low in these two countries, particularly in Portugal. Brazilians consume more chocolate flavored milk (34.4%) than the Portuguese (17.4%), but in both countries this is consumed with a very low frequency (seldom). In Portugal there is a slightly higher percentage of people who never consume enriched milk (94.8%) as compared with Brazil (85.3%). However, because portion and serving sizes were not specified in the questionnaires, the obtained responses may not be fully indicative, given that different participants may have interpreted the questions differently.
Milk and dairy products are considered by the FAO as important in the human diet, given their high quality protein and micronutrients in an easily absorbed form [63
]. However, it has been reported in several sources that milk intake has gradually declined over the past decades [64
], maybe due to the aforementioned gastrointestinal problems that can appear after lactose ingestion [14
]. There are considerable studies related to the health benefits of milk consumption e.g., for bone strength [33
], risk factor of osteoporosis [70
] and protective effects against asthma, current wheeze, hay fever or allergic rhinitis, and atopic sensitization [71
]. Concerning this point of view, some controversial studies can be found, like that of Wang et al. [72
], which concluded that the risk associated with the consumption of milk depends on the quantities, so that moderate milk consumption diminished the risk of mortality associated with cardiovascular diseases and a high milk consumption showed an increased risk of cancer mortality. Specifically, in the two countries under study, some studies have addressed the problem of osteoporosis in Portugal and in Brazil [36
], a recent investigation evaluated the sleep patterns in Brazilian children and the consumption of dairy products [73
] and one study shows arterial hypertension management strategies according to some factors, such as dietary management, including milk and dairy product recommendations [74
More recently a variety of enriched milks have appeared in the market, mainly with calcium and vitamin D. Calcium is critical for children’s development and is necessary for skeletal consolidation and preventing fractures and osteoporosis in old age [75
]. According to the Dietary Guidelines Advisory Committee [77
], lower calcium consumptions are linked with adverse health outcomes. The importance of vitamin D is well known owing to its importance in helping with the fixation of calcium in the bones, among other roles in the human body, namely regulating the brain, liver, lungs, heart, kidneys, skeletal, immune and reproductive systems. This vitamin also has significant anti-inflammatory, anti-aging, anti-stress, anti-arthritic, anti-osteoporosis, anti-apoptotic, wound healing, anti-cancer, anti-psychotic and anti-fibrotic actions [78
As observed previously for milk consumption, a high number of participants in both countries never eat cheese (Table 12
). However, this percentage is higher for Brazilians compared with the Portuguese for most types of cheese, except only for whey cheese. In fact, Portugal has a long history of eating traditional cheeses that are presently recognized with PDO (Protected Designation of Origin) [84
]. The highest percentage of Portuguese who never eat a certain type of cheese is verified for imported cheeses (67.9%) and the lowest for fresh cheeses (28.8%), while for Brazilians, those who never eat cheese are, also, predominantly for imported cheeses (77.0%) and with least expression for whey cheeses (37.2%). For all categories of cheese and in both countries, the more usual consumption frequency is once a week. The differences observed in the cheese consumption patterns in both countries could be attributed to the different habits and production modes. The Brazilian cheese market has been reported to vary according to place of origin, type of milk (cow, buffalo, goat), manufacturing procedures, texture, and maturation time, among other factors [85
]. For Portugal, with good pastures and a tradition in pastoralism, there are several types of cheese, made with cow, goat, sheep or mixtures of different milks, which have different tastes and consistencies. According to Guiné et al. [84
], the traditional Portuguese cheeses can be classified according to the type of milk used for the cheese production, the fat content, ripening and paste consistency.
Whey is a dilute liquid resulting from cheese manufacture that contains lactose, proteins, minerals, such as calcium, and traces of fat and organic acids [84
]. An important difference was observed in the consumption of whey cheese. Brazilians eat whey cheese in higher percentages (62.8%) compared to the Portuguese (54.1%), but in both cases the frequency of consumption is mostly once/week (42.9% of Portuguese and 39.1% of Brazilians).
According to Ferrão & Guiné [84
], cheese is a good source of calcium, fat, protein, and some vitamins (A, B2
), as well as other dietary minerals such as zinc or phosphorus. Cheese is not only consumed in its original form, and during the last decade, it has become one of the most widely used food ingredients, leading to the development of several types of low-fat cheeses that have health-promoting benefits beyond their nutritional value [86
Butter is one of the most ancient and popular dairy products. This dairy product contains valuable fatty acids, as well as fat-soluble vitamins (A, D, E, K), tocopherols and carotenoids, among its important nutrients. However, the consumption of butter must be moderate because it has been linked to high cholesterol, atherosclerosis, and heart disease [88
]. Consumer acceptance of butter is influenced by its sensory properties, which are dependent on milk raw material quality that influence the final flavor, aroma, appearance, and rheological properties [90
]. Table 12
also presents the results for butter consumption in Portugal and Brazil. Around 50% of Brazilians never consume butter and the observed percentage is 43.8% for the Portuguese who also never consume butter. These percentages increase for the consumption of skimmed butter (Brazil: 82.6% never consume it; Portugal: 66.0% never consume it) and unsalted butter (Brazil: 69.1%; Portugal: 70.2%). For those who consume butter, the highest percentage of people consume it seldom, only once/week.
Yogurts are obtained from milk fermented by lactic acid bacteria such as Lactobacillus
, and Leuconostoc
which allows for an extension of the product shelf life and improves its taste compared to milk, giving way to differentiated products in the market. Fermented dairy products’ consumption has been increasing widely around the world and different companies encourage new product development to satisfy the consumers with new tastes and flavors. Some of these products have demonstrated nutritional value and health benefits. For example, it has been shown that intestinal bacterial microbiota contributes to a healthy life and increases life expectancy [91
]. This kind of product contains a high amount of live bacteria, which has benefits for human health, contributing to the maintenance and balance of the intestinal flora, facilitating digestion and preventing constipation and other gastrointestinal disorders [94
]. Several studies state that yogurt presents antimutagenic and anticarcinogenic effects and provides protection against colorectal adenomas [95
The results regarding yogurt consumption (Table 12
) reveal that a high percentage of participants never consume yogurt and, on average, this value is higher for Brazilians (52.3% to 74.2% depending on the type of yogurt) when compared with the Portuguese (33.1% to 73.3%). Regarding the results in Table 12
, it is observed that the classes of yogurts which are consumed by a lower percentage of participants are the ones with separated flavors (only by 26.7% of Portuguese and by 25.8% of Brazilians). Natural yogurt is consumed more frequently by the Portuguese (59.8%): 29.3% once a week and 19.3% 2–3 times/week, while the Brazilians consume 42.3% of this kind of yogurt: 30.9% once a week and 7.9% 2–3 times/week. A relatively similar trend is also observed for the aromatized yogurt. For the other classes of yogurt, the differences between Portuguese consumers and Brazilian are small.
From the collected data it was further possible to determine those participants who never consumed certain classes of dairy products or those who never consumed any of the investigated dairy products at all, and these results are shown in Table 13
for the global sample and separated by country. The results indicate that butter is the class which a highest percentage of participants never consume, 30.0%, with a higher expression in Brazil as compared to Portugal. The second class corresponds to milk products, which are never consumed by 22.8% of people, but in this case it is in Portugal that the percentage in higher. Following comes the yogurt category, never consumed by 21.1% of Brazilians and by 7.6% of Portuguese. Last appears the cheese category, with the lowest percentage of people who never consume this type of dairy product. Finally, one can see that a residual number of participants identified as never consuming any of the dairy products considered, and this result might not correspond to the reality in both countries, since this was a questionnaire survey in which the volunteers participated knowing from the start that it was about the consumption of dairy products (this information was provided before the participants gave their informed consent). Therefore, it is possible that people who never consume dairy products did not even respond to the questionnaire, by considering that their participation was not useful, or because they did not want to spend time with a subject that was not important for them.
3.5. Variables Influencing Dairy Product Consumption
As explained in the section Materials and Methods, the variables accounting for the average level of consumption of the four categories of dairy products (milk, cheese, butter and yogurt) were submitted to a tree classification analysis for evaluation of the relative importance of each of the possible influential variables considered: country, sex, age class, education, BMI class, satisfaction with body weight, balanced diet and lifestyle clusters. Figure 1
, Figure 2
, Figure 3
and Figure 4
show the obtained classification trees, and they reveal that some of the variables considered in the analysis were not influential, for example, variables such as country and balanced diet never appeared in any of the diagrams, meaning that they do not determine the consumption of any of the dairy products evaluated.
The tree in Figure 1
, for consumption of milk products, contains 4 levels and 17 nodes, of which 9 are terminal. The risk estimate for re-substitution was 0.062 with standard error 0.008 and the risk estimate for cross-validation was 0.061 with standard error 0.008. These values indicate goodness of fit to the model. The results in Figure 1
reveal that, for the whole sample (node 0), a huge majority of participants have a low milk consumption (93.8%) and that the first discriminant variable was age, so that younger people (from 18 to 30 years) tend to have a slightly higher consumption of milk than older people (moderate consumption: 11.2% and 4.2%, respectively, for people up to 30 years and older). For young people, the second discriminating factor was sex, with men showing higher consumption of milk as compared with women. For young men, the next discriminant variable was body weight satisfaction and the final differentiating factor was lifestyle cluster. Regarding older people, the next discriminating factor after age was education, with lower milk consumption for people with university degrees or higher levels of education. Following in the order of appearance the discriminating variables were sex, lifestyle cluster and age class again, differentiating in the last level middle aged adults (93.5% low and 6.5% moderate consumption) from senior adults and elderly (100% low consumption).
The tree in Figure 2
for the consumption of cheese has 5 levels with 17 nodes, including 9 terminals. The risk estimates for re-substitution and for cross-validation were in both cases 0.062 with standard error 0.008. The results for the whole set of participants (node 0) indicate that, similarly to milk consumption, cheese consumption is also very low (93.8% low, and only 5.9% moderate). The discriminant variable in the first level was BMI class, separating the overweight and obese participants as having slightly higher consumption of cheese (8.6% moderate and 91.1% low). For these, the next discriminant variable was education and in this case those with higher levels of education tend to have a higher consumption of cheese (8.6% moderate against 0% for those with lower education). For the branch of underweight and normal weight, BMI class was again the differentiating factor, with lower consumption for normal weight participants (3.5% moderate against 11.4% moderate for the underweight). The following discriminant variables were body weight satisfaction (level 3), sex and age class (level 4) and lifestyle cluster (level 5).
presents the results of the tree obtained for butter consumption. This has only 3 levels, representing 7 nodes, of which 4 were terminal. The risk parameters were equal for re-substitution and cross-validation: risk estimate = 0.076, standard error = 0.009. In the case of butter, the values for the whole sample are low (92.4% low, 6.6% for moderate and 1.1% for high consumption), being in line with the trends previously observed for milk and for cheese. The first discriminant was sex, differentiating men as having lower consumption than women (94.9% and 91.0%, respectively). For women the node was terminal, while for the men, the following discriminant was lifestyle cluster, separating the screeners and exercisers as showing a slightly higher butter consumption (6.3% moderate as compared with 1.9% moderate for travelers and others). Finally, at level 3 the discriminating variable was age class, separating people over 50 years, which presented a higher butter consumption (4.7% moderate).
shows the tree for yogurt consumption, with 4 levels and 11 nodes (six of which are terminal). The risk estimate for re-substitution was 0.085 with standard error 0.010 and equal values were obtained for cross-validation. The consumption of yogurts is again low (91.5% at node 0). The first discriminant was BMI, as was observed for cheese, and the discriminant at level 2 was body weight satisfaction, regardless of the BMI class (i.e., on both branches). For the overweigh and obese (i.e., BMI of 25 or over) who are satisfied with their body weight, 25% have a moderate yogurt consumption. For these, the next level was separated according to their age class, and for the participants aged up to 50 years the last discriminant was lifestyle cluster.
Overall, these results seem to indicate that the most relevant discriminant factors for dairy consumption were age, BMI, sex, education, satisfaction with body weight and lifestyle cluster, following more or less the same order regardless of the type of dairy product. On the contrary, factors such as country or balanced diet were found to have no discriminant capacity for the variables under study, i.e., the consumption of dairy products is not influenced significantly by these two variables.
According to Wolf et al. [99
] age is a factor determining milk consumption in the United States, with people born in the 1990s consuming milk less often than earlier generations, and this trend is expected to continue with the replacement of older generations by younger ones. Nevertheless, the consumption of other dairy products seems to be increasing in the US, in the case of cheese mostly because it is widely used in pizzas and in the case of butter because there has been a setback regarding health views of butterfat [101
]. The study by Xu et al. [4
] highlighted also differences in dairy consumption according to sex and BMI. On the other hand, lifestyle behaviors have been proven to influence dairy consumption according to the recent study by Santaliestra-Pasías et al. [102
]. Their results suggest that European children with healthier lifestyles, specifically regarding aspects such as physical activity and sedentary behaviors, tend to consume higher quantities of milk and yogurt.