The data were collected by the National Institute for Health and Welfare in spring 2012. The survey is a part of the annual Health Behaviour and Health among the Finnish Adult Population (AVTK) monitoring survey. For the survey, each year a random sample of 5,000 Finnish citizens residing permanently in Finland and aged 15–64 has been drawn from the National Population Register. The questionnaire is mailed between April and June, with three reminders. The survey instrument and the method have been approved by the Research Ethics Board of the National Institute for Health and Welfare.

The 12-page questionnaire covered questions on socioeconomic background and various health behaviours, such as eating, physical activity, smoking, and alcohol consumption. Food frequency questions (FFQ) and questions on fat choices, which are also reported in the present study, have been repeated in the annual survey since 1978 (20–22) but have not been validated. For the purposes of the present study, new questions on the following areas were added to the questionnaire in 2012: relation to LC diets, perceptions of healthiness of fats and carbohydrates, and eating motives. The new questions were piloted among the consumer panel of the National Consumer Research Centre (n=411). After the pilot, the questions were further developed into their final form.

In the final questionnaire, the respondents’ relation to LC diet was investigated with the question: ‘What is your relation to the LC diet?’, with the answering options ‘I have not heard about the LC diet’, ‘I have heard about the LC diet, but do not practice it’, and ‘I follow the LC diet’. The question contained a term that is commonly used in Finnish with reference to Atkins-type diets (karppaus), hence typically referring to limited consumption of carbohydrates, increased consumption of meats and other animal proteins, and the view that saturated fats are not harmful to health. At the time of the survey, the concept ‘karppaus’ had been established in Finnish and was frequently used in the media (5, 6), although it did not always refer to a clearly defined set of dietary practices and it probably could be interpreted in varying ways. In common parlance, it was a very typical way of referring to LC diets.

Weight reduction was measured with the question ‘Are you slimming down at the moment?’, with the response options of ‘no’ and ‘yes’. BMI was calculated by dividing the respondent's weight (kg) by height squared (m²).

Consumption of foods and beverages containing carbohydrates was measured with a FFQ enquiring about the number of days during the past week on which the respondent had eaten ‘fresh vegetables’, ‘cooked vegetables’, ‘fruits and berries’, ‘cooked potatoes’, ‘French fries’, ‘rice/pasta’, ‘cereals’, ‘sweets and chocolate’, ‘juices’, and ‘sweetened soft-drinks’. The response options for each item were ‘not at all’, ‘on 1–2 days’, ‘on 3–5 days’, and ‘on 6–7 days’. Consumption of bread was enquired with a question ‘How many slices of bread do you usually eat daily’ for ‘dark bread (rye bread or rye crisp, etc.)’, ‘mixed, graham, wholegrain, or oat bread’ and ‘bread from refined wheat’, and the respondents were asked to fill in the number of slices for each bread type. In addition, the respondents were asked if their diet was ‘low in all carbohydrates’ and ‘low in sugar and refined grain products’, both with the response options of ‘yes’ and ‘no’.

Meat, fish, and egg consumption was measured with a FFQ enquiring about the number of days during the past week on which the respondent had eaten ‘meat’, ‘chicken’, ‘fish’, and ‘egg’, with the response options ‘not at all’, ‘1–2 days’, ‘3–5 days’, and ‘6–7 days’.

Fat choices were measured with two questions: the type of fat the respondents used most often for cooking (with the options of vegetable oil, liquid margarine, low-fat spread with ca. 60% fat, soft margarine, hard margarine, mixture of butter and oil, butter, and none), and the type of bread spread used most often (with the options of not at all, spread with max. 40% fat, spread with max. 60% fat, plant stanol/sterol spread, and soft margarine with 70–80% fat, mixture of butter and oil, and butter). In the analysis, products processed from vegetable oils, that is, margarines, low-fat spreads, and plant stanol/sterol spreads, were combined into one category.

The perceptions about the healthiness of fats and foods containing carbohydrates were investigated with a question that was further developed from a question on knowledge of the relationship between dietary fat type and heart disease introduced by Eckel et al. (23). The question ‘Foods may have an impact on health. What is your perception of the effects of the following foods on your health?’ was followed by a list of foods, including animal fats, vegetable oils, vegetable bread spreads, whole grain products, refined wheat products, vegetables and root vegetables, and fruits and berries. The options were ‘are harmful for my health’, ‘are beneficial for my health’, ‘have no effect on my health’, and ‘I don't know’.

In order to investigate the respondents’ food choice motives, we applied the Food Choice Questionnaire (24). However, because of limited space in the questionnaire, it was only possible to include a short list of items that were expected to cover several motives. The motives were asked with the phrase: ‘In my opinion …’, followed by 18 motives and 5 answering options ‘is not at all important’, ‘is not very important’, ‘is somewhat important’, ‘is important’, and ‘is very important’.

Principal component analysis (PCA) was conducted separately for motives related to the contents of diet and other motives. Based on the high loadings of items, five factors emerged in total. The reliability of the factors was evaluated with Cronbach's α values. The three factors related to the contents of the diet were named ‘healthy carbohydrates’ (including items ‘to eat a lot of vegetables, fruits, and berries’, ‘to eat fibre-containing foods’, ‘not to eat sugary foods’, ‘not to eat bread and pasta from refined grains’; Cronbach's α=0.770, mean=3.47, SD=0.78), ‘vegetable/low fat’ (including items ‘to use vegetable oil’, ‘to use vegetable bread spread’, ‘to eat low-fat foods’; Cronbach's α=0.746, mean=3.09, SD=0.88), and ‘meat and animal fats’ (including items ‘to eat foods containing butter and cream’, ‘to eat a lot of beef, pork, and chicken’; Cronbach's α=0.45, mean=2.37, SD=0.78). The two factors related to potential other motives were ‘pleasure and sociability’ (including items ‘that food tastes good’, ‘that I eat together with my family or friends’, ‘that I can eat what I please’, ‘that it puts me in a good mood’; Cronbach's α=0.607, mean=3.66, SD=0.63), and ‘health and weight’ (including items ‘that food is good for my health’, ‘that I don't gain weight from the food I eat’; Cronbach's α=0.615, mean=3.67, SD=0.80). Three items were omitted (‘to eat a lot of fish’, ‘that I gain fast energy/feeling of fullness from food’, ‘that I do not worry about the healthiness of food’) due to their low contributions in the PCA.

A total of 2,601 respondents were included in the data, with a 52% response rate. The response rate of the yearly survey has been steadily decreasing, falling under less than 60% during the recent years (20). Of the respondents, 57% were women, 24% were 15–30 years old, 35% were 30–49 years old, and 41% were 50–64 years old. Compared to the general population (25), men (43% in the data vs. 51% in the population) and the youngest age group (22% in the data vs. 28% in the population) were underrepresented in the data. In total, 12% had completed less than 10 years of education, 30% had 10–12 years of education, and 58% had more than 12 years of education. In 2010, the mean of completed years of education among Finns was 12.3 (26), compared to 13.8 in the present data. Although the level of education might have slightly increased in the population in the 2 years between 2010 and 2012, persons with a higher educational level were probably somewhat overrepresented and those with a low level of education were underrepresented in our data.

In the analysis, SPSS version 21 was applied. The differences between the followers of the LC diet and the other respondents with regard to food choices and perceptions of health risks are explored with cross tabulations (Chi-square test) and with ANOVA. The associations between the socioeconomic factors, perceptions of health risks, and eating motives were investigated using Pearson correlation coefficients, and in the final analysis, the associations between these variables and following an LC diet were examined with logistic regression analysis.

In total, 7% of the respondents reported that they were following the LC diet in 2012. Only 3% of the respondents reported that they had not heard of the LC diet.

Of those who were following the LC diet, 57% also reported that they were currently trying to lose weight, compared to 21% among the other respondents (p<0.001). The mean BMI of the LC diet group was 27.21 and among other respondents it was 25.59 (p<0.001).

With regard to foods containing carbohydrates, frequent consumption of fresh vegetables was more common among the followers of the LC diet than among the other respondents (Table 1). The association was similar in cooked vegetables (data not shown), whereas there were no differences between the groups when it came to eating fruits and berries. Eating cooked potatoes, rice, and pasta was less frequent among the followers of the LC diet than among the other respondents. The association was similar in cereals, but with French fries there was no difference between the two groups (data not shown). Drinking juices and sugar-sweetened soft-drinks was less frequent among followers of the LC diet than among the other respondents (Table 1), whereas there were no differences between the groups when it came to eating candies and chocolate (data not shown). Seventy four per cent of the LC group and 80% of the other respondents ate candies or chocolate at least once a week (ns.). The LC diet group ate less rye bread compared to the other respondents (2.1 slices/day vs. 2.6 slices/day, p=0.001); less mixed, graham, wholegrain, or oat bread (0.5 slices/day vs. 1.2 slices/day, p<0.001); and less breads from refined wheat (0.04 slices/day vs. 0.2 slices/day, p=0.01). Moreover, the followers of the LC diet reported more frequently than the other respondents that they were avoiding all carbohydrates (83% vs. 22%, p<0.001) and that they were avoiding sugars and refined wheat products in their diet (87% vs. 47%, p<0.001).

The followers of the LC diet ate chicken and eggs more frequently than the other respondents. With regard to eating fish (data not shown) and meat, there were no differences between the two groups.

The followers of the LC diet differed from the other respondents in that they used butter more often, and margarines and vegetable spread products less often, in cooking and on bread (Table 2). However, even in the LC diet group, butter was not the most common cooking fat or bread spread. Instead, in both groups, vegetable oil was the most common cooking fat, and vegetable bread spread was used on bread by a third of the followers of LC diets and by nearly a half of the other respondents. A mixture of oil and butter was used on bread by one third in both groups.

Although a majority in both groups considered products from refined wheat products as harmful to health, this perception was more prevalent among the followers of the LC diet (87% vs. 69%, p<0.001). A majority in both the LC diet group (80%) and the other respondents (90%) regarded whole grain products as good for health. Vegetables and root vegetables were considered as good for health by nearly all the respondents in both groups (followers of LC diet 97%, other respondents 95%). Likewise, fruits and berries were almost unanimously considered as good for health (97 and 96%, respectively).

Concerning the healthiness of fats, the followers of the LC diet differed from the other respondents in that they regarded animal fats as good for health more commonly (26% vs. 10%), and as harmful to the health less commonly than the other respondents (44% vs. 57%, p<0.001); and they considered vegetable bread spreads as harmful to health more commonly (26% vs. 8%), and as good for health (40% vs. 52%), or as having no effect (19% vs. 23%, p<0.001), less commonly. In total, 84% of the respondents in both groups regarded vegetable oils as good for health, and 10% as having no effect on health.

Finally, in order to examine how socioeconomic background, perceptions of health risks related to foods and eating motives are related to following an LC diet, logistic regression analysis was conducted. Of the risk perceptions, we included those which in the previous analysis had been shown to be associated with an LC diet (white wheat, animal fats, vegetable bread spreads). The associations between the independent variables were moderate (Pearson correlation coefficients). Gender and education did not predict the LC diet either in the unadjusted or the adjusted models. In the final model, the older respondents had, higher the probability of following an LC diet (Table 3). The odds of following the LC diet increased by 4% per one unit increase in BMI. Although having significant unadjusted effects, in the final model, perceptions of the health effects of animal fats, vegetable bread spreads, and refined wheat products were not associated with an LC diet. With regard to eating motives, preference for vegetable/low-fat products decreased the probability of choosing an LC diet, whereas preference for healthy carbohydrates and meat and animal fat products were positively associated with an LC diet. Those respondents who considered health and weight management as an important aspect in their food choices were more likely to follow an LC diet, whereas placing value on the pleasure and social aspects of food choices had the opposite effect.

The LC diet was a widely known diet among Finns in 2012. However, only 7% reported that they were following the diet. This proportion resembles earlier results from Finland reporting that 6% (18) and 10% (17) of Finns had followed the LC diet in 2011.

We found that gender and education were not associated with the tendency to follow an LC diet. Population surveys in Finland during the past years show that women are more inclined to try to lose weight than men (20). Other studies have reported similar results (27). The LC diet studied here, in contrast, attracted men and women equally. This might be partly due to the centrality of meat and animal fats in LC diets and the ‘masculinised’ character of the diet (9), in addition to other characteristics that differentiate the LC diet from traditional weight reduction diets. With reference to age, our findings resemble those on dieting in general: The youngest are the least likely to diet (20). Young adults are typically less concerned about weight and health, and hence show lesser tendency to try to change their diet.

To what extent did the followers of LC diet of the present study follow the core rules of LC diets, that is, avoidance of carbohydrates, consumption of meat and other proteins, and acceptance of saturated fats in the diet? In many respects, there was a tendency towards favouring these components: The followers of LC diet consumed potatoes, rice, pasta, bread, juices, and sugar-sweetened soft-drinks less frequently than the others and more frequently reported avoiding carbohydrates, sugars, and refined grain products. They ate eggs and chicken slightly more frequently than the other respondents and had a negative perception of refined wheat products more commonly than the other respondents. Furthermore, the followers of LC diet used less margarine and vegetable bread spread and more butter than the other respondents. The perception that meat and animal fats make up an important part of a diet was also associated with the LC diet.

However, in many respects, the lifestyles of the followers of the LC diet appeared as similar or actually healthier compared to those respondents who were not following the diet. There was no difference in the frequency of fruit and berry consumption between the groups. A majority of the followers of the LC diet and the other respondents agreed with certain core claims of current nutrition recommendations: – that whole grain products, vegetable oils and vegetables, fruits and berries are good for health. Moreover, a majority in both groups agreed that refined wheat products are harmful to their health. Indeed, although the LC diet is typically seen to contradict official nutrition recommendations, there are similarities between the two diet types, too. Excess consumption of sugar-containing products has been linked to obesity, not only in LC diet promotion but also in national nutrition recommendations and scientific publications in the 2000s (28).

Moreover, the followers of the LC diet ate vegetables even more often than the other respondents. Similar results have been reported of LC dieters in the United States, who believed that avoiding sugars and starch and eating vegetables plays a more important part in their diet than increasing fat consumption (10). Moreover, only 5% of these LC dieters had increased their consumption of beef, bacon, and butter to any great degree, whereas more than a majority had increased their consumption of salad/lettuce (10).

Concerning eating motives, the followers of the LC diet did not particularly stand out as pleasure-seekers, as has been suggested by previous research (1, 11). Instead, they placed more importance on the weight-related and health aspects of food, and were more commonly trying to lose weight compared to the other respondents. However, it is interesting to note that even if LC diets have been promoted first and foremost as slimming diets, only a little more than a half of the followers of LC diet in the present study were trying to lose weight. This suggests that LC diets are used for other purposes, too, for instance as a means to improve well-being in terms of feeling good and energetic.

The response rate of the survey was 52%. Although it was even lower among some population groups such as young men, the data still covered these groups, too. Others have concluded that likely candidates not to respond are persons whose lifestyles contain several unhealthy components or who are the least interested in health issues at large (29). Hence, it is possible that the proportion of followers of the LC diet would have been slightly different had the response rate been higher. Still, previous market research of the LC diet has shown similar enough proportions, giving support to the present findings (17, 18). However, the aim of the present study was not to estimate eating patterns and food-related perceptions on the population level but, rather, their associations with the LC diet. It is unlikely that one of the main findings of the present study – that the LC diet is known by a majority of the population but practiced only by a minority and in varying ways – would be significantly different with a better response rate. We should also note that over the last couple of decades, response rates in surveys have generally declined in European countries, including Finland (29). On the whole, it is becoming increasingly difficult to achieve high response rates in postal surveys (30). This decrease has been detected in all population groups, and consequently in recent studies response rates of over 50% have been considered satisfactory (31).

Another issue to be considered is self-report measurement of dietary behaviours. Self-report is the only feasible and cost-effective method for collecting large, representative data. Moreover, although the questions of the questionnaire may cover only some aspects of diets, they still characterise sufficiently certain key preferences in the dietary choices and provide useful information related to, for example, the thus far ill-studied LC diets. In addition, the food frequency questionnaire of the present study has been previously used to characterise patterns of and changes in food consumption among Finns in a consistent manner (22, 32) (33).

What is surprising in the results is that some of the respondents who identified themselves as followers of the LC diet also reported that they were not avoiding carbohydrates, sugars, and refined wheat products in their diet. This coincides with studies on vegetarianism which consistently find that self-defined vegetarianism is more common than vegetarianism measured by FFQ (34). Vinnari et al. (34) suggest the following reasons for the inconsistency: lay persons’ definitions of vegetarianism differ from those of the experts, people are not fully knowledgeable about the food products they consume, and some consumers want to identify as vegetarians although they eat meat. A similar explanation might apply to the results presented here, too. We suggest that in the peak of the LC diet boom, some respondents wanted to identify themselves with the phenomenon even though from the nutritional perspective they were not actually following an LC diet. In addition, some of the respondents may have lacked knowledge of nutritional contents of foods and assumed that they were following the LC diet even though they may have been eating foods actually banned in the LC diet.

An analysis of media discourse on LC diets in Finland showed that the LC diet as a concept is unclear both from a nutritional and a cultural perspective (5). The results of the present study corroborate this finding by showing that people may identify themselves as followers of LC diet irrespective of whether their food choices are typical of a strict LC diet or not. We should also bear in mind that dietary regimes in practice are always variations of the original recommendations. They change and evolve through public discourses and in everyday life as their followers adjust the dietary advice to fit in with the food practices and expectations of their social environments at work, in family life, and in leisure time.

It seems that LC diets are currently losing their status as the most highly debated fad diet. Instead, new fads are emerging, such as the gluten-free diet, the ‘5:2 diet’ on which dieters fast 2 days a week, and even combining various diets. Planners of nutrition policies and recommendations should closely monitor new diets as they emerge and explore the food choices and motives of the dieters. It seems that those who identify themselves with particular diets apply the diets in varying ways and intensities. Moreover, it might be that diets with wide media publicity, such as the LC diets, affect the food choices of not only the adherents of the diet but also those who do not identify themselves with the diet. In Finland, there is some indication of the LC diet boom having increased butter and cream consumption within only a few years’ time so much that it cannot be explained only by the changing consumption habits of low-carb dieters (7). This shows that novel diets may diffuse and affect food cultures in more profound ways than could be expected just by examining the food ways of the followers of the diets.