Although the role of dietary intake, especially the role of fat intake, in the development of obesity has been widely discussed in the literature, uncertainties remain because of conflicting results reported by cross-sectional and prospective studies . This could be partially explained by the phenomenon of underreporting when using self-reported dietary intake. Most studies in nutritional epidemiology mention underreporting as an important bias. If underreporters are not excluded, spurious associations between dietary intake and obesity may be found. However, only few studies identified and excluded underreporters [4, 7, 38].
In the present study, plausible reporters were compared to underreporters for a number of anthropometric, behavioural and dietary characteristics. Underreporting is commonly associated with obesity quantified by body mass index [8, 12–15, 39]. In this study, a significantly larger percentage of overweight, obese, but also abdominally obese individuals were observed among underreporters. Underreporters were also significantly younger, but no significant difference was found for educational level and smoking status between plausible and underreporters. Some studies found underreporting to be more prevalent among older [13, 14], less  or more educated  subjects and smokers . Again other studies reported, similarly to our results, a higher proportion of underreporting among younger subjects [39, 41] and no significant difference for educational level between under- and plausible reporters [13, 38]. In the present study, male underreporters participated significantly less in health related sports and had lower PAL compared to plausible reporters. These results are in agreement with the study of Johansson et al.  indicating lower activity scores among underreporters. In our study, total daily energy intake and intake of all macronutrients were lower among underreporters than among accurate reporters, and underreporters reported a significantly higher percentage of energy from protein, but lower percentage from fat. These results are in accordance with the literature indicating that underreporters and obese persons tend to underreport foods rich in fat and sugar, and overreport protein intake [12, 14, 38]. Nevertheless, carbohydrates expressed as percentage of energy were significantly higher among female underreporters compared to their plausible counterparts in the present study. A number of the aforementioned inconsistent results may be caused by different dietary assessment methods (FFQ vs. diet diary) and different methods used to identify implausible dietary intake (i.e. doubly labelled water, biomarkers, ratio of reported energy intake to basal metabolic rate). Differences in age, culture and health consciousness among the populations studied may also be responsible for these inconsistencies.
The main purpose of the present study was to analyse the associations of dietary intake with BMI and WC after excluding the underreporters. A first important finding was that fat intake (kcal/day), including saturated, mono- and polyunsaturated fat, was significantly higher in overweight and (abdominally) obese subjects compared to their normal weight counterparts in both genders, statistically controlling age and physical activity variables. In addition, percentage of energy intake from fat was significantly higher in overweight, obese and abdominally obese men compared to men with normal weight or WC. On the other hand, this observation was not significant in women. In the literature, the role of dietary fat in the development of obesity and abdominal obesity is equivocal. Some cross-sectional studies found no association [17, 42], while other observed a positive association [16, 18] between higher fat intake and obesity. In the study of Garaulet et al. it is suggested that, even though obesity is a multifactorial phenomenon, dietary intake, especially fat intake, is the most important factor contributing to obesity . Larson et al.  indicated that dietary fat plays a minor role in increasing overall body fat but not specifically influences fat increase in the intra-abdominal region. In a prospective study of Koh-Banerjee et al.  a 2% increment in energy intake from unsaturated (trans) fat resulted in a 0.77 cm waist gain over 9 year, whereas in other prospective studies no association was found between intake of dietary fat and abdominal obesity [24, 25, 45]. A possible explanation for the lack of association in these latter studies could be that implausible reporters of dietary intake were not always excluded. Nevertheless, the results of the present study confirm the relationship between fat intake and both BMI and WC. Several arguments have been proposed for this association. Firstly, fat is the most energy dense macronutrient. Secondly, fat provides a lower satiety feeling and its great flavour and palatability may lead to a greater consumption of fatty foods. Authors also reported that fat has a lower thermogenic effect than carbohydrates and proteins [46, 47], which results in lower energy expenditure and consequently in larger fat stores. Finally, diet currently contains a lot of added sugars and fats as the food industry has made added sugars and vegetable oils accessible at a remarkably low cost .
An important asset of the present study is that besides fat intake, the association of other macronutrients and total energy intake with obesity has also been analysed. According to the BMI and WC classification in both men and women, total energy intake was found to be positively associated with obesity. Slattery et al.  found a positive correlation of total energy intake with waist-hip-ratio, but not with BMI in white women, but they found no significant correlation in men. In contrast, in a study of Trichopoulou et al. , higher energy intake was significantly associated with waist-hip-ratio independently of BMI in men but not in women. In other studies, there was no significant difference in energy consumption among the BMI categories [17, 18]. Following our results controlled for age, Tsports and Ttv, energy percentages from sugars, starch and fibres are negatively related to BMI and WC in men, whereas in women belonging to the overweight or obesity category is associated with higher carbohydrate and fibre intake (kcal/day). In the literature, overweight and obese individuals are reported to consume generally less carbohydrates [16–18, 42] and dietary fibres [4, 18, 26, 42, 49] than their normal weight counterparts. Several physiological mechanisms for the role of fibre on weight regulation are proposed in the literature. The energy content of fibre per unit weight food is low. Consequently, inclusion of fibre in a diet reduces energy density. Dietary fibre tends to reduce dietary intake by slowing digestion and absorption of nutrients, and by increasing the production of gut hormones enhancing satiety feeling. Moreover, some types of fibre reduce the overall absorption of fat and protein . In the present study, regarding carbohydrate, starch, sugar and fibre intake, opposite results were found for men and women. In men, energy percentages from carbohydrates, starch and fibres are significantly lower in overweight and obese men compared to lean men, whereas in women the intake of carbohydrates, starch, sugars and fibres were found to be positively related with overweight and obesity. A possible explanation for this sex difference regarding fibre intake could be that women tend to be more health-conscious and more knowledgeable about food and nutrition than men [50–52]. In addition, one could assume that fibre intake will generally increase with higher total energy intake which is the case in overweight and obese women, although this is not the case in men.
As to protein intake, the positive association with (abdominal) obesity in both genders is in agreement with the findings of Slattery et al. . Conversely, Davis et al.  reported that protein intake did not differ between BMI groups. Relationships found between alcohol consumption and body weight or fat distribution are inconsistent and seem to be sex specific . In the present study, men and women with high risk abdominal obesity reported a significantly higher percentage of energy intake from alcohol. These results support the findings of other investigators that moderate to heavy drinkers of both genders have a larger WC or higher waist-hip-ratio than light drinkers [42, 53, 54]. However, when using the BMI classification only, no difference was found in alcohol intake between the groups. In other studies, BMI was found to be negatively related to alcohol intake in women, whereas a slight positive association was observed in men [53, 55, 56]. Studies on the association of obesity with iron and cholesterol intake are sparse. Our results show a significantly higher consumption of iron and cholesterol among men and women with abdominal obesity. These results might partially be explained by the higher fat and protein intake among obese individuals in our sample.
The findings of the present study revealed that regarding dietary intake some sex differences were observed between obese men and women. Obese women show a higher consumption of all macronutrients, and surprisingly also of fibres, while men show a higher fat and protein intake, but a lower intake of carbohydrates and fibres. In addition, there was a positive association between alcohol intake and abdominal obesity in men, but not in women. On the other hand, women with abdominal obesity show a significantly higher calcium intake compared to lean women, but this trend was not significant in men.
Another purpose of this study was to investigate whether plausible dietary intake varies between subjects with different level of sports participation after adjustment for age and BMI. Subjects participating in health related sports reported higher intake of carbohydrates, but lower intake of fat compared to subjects not participating in sport activities. An explanation of this finding can be that individuals participating in health related sports are more health conscious and are more prone to adopt a low fat – high carbohydrate diet than sedentary individuals less well concerned with health. It is well accepted that carbohydrates are one of the most important components in sports nutrition. In addition, there is evidence that dietary intake can be influenced by physical activity. High intensity exercise induces a suppression of appetite, and long duration, high-intensity exercise has a stronger effect than a short duration exercise period . The higher sugar intake in physically active women might be explained by the assumption that these women try to compensate their higher intake of sweets by doing more sports.
Several limitations should be considered when interpreting results of the present study. The first limitation is the cross-sectional design, which inhibits to establish causal relationships as both the effect (obesity) and the potential causes (dietary intake and other factors) are measured simultaneously. Second, the use of self-reported data for dietary intake and physical activity depends for a large part on the cooperation and honesty of the participants. On the other hand, the 3-day diet diary  and the physical activity questionnaire (FPACQ)  used in our study have been properly validated previously. Although these instruments have limitations, they are easy to administer and low in cost, whereas the use of more objective methods, such as doubly labelled water technique, indirect calorimetry or accelerometers is unfeasible in large scale epidemiological studies. Another limitation is related to the representativeness of the sample. Although the original sample was randomly selected by the National Institute of Statistics, our sample consisted of volunteers, presumably being more health conscious persons. Moreover, the exclusion of the implausible reporters resulted in a reduction of the sample size and a more select sample which could be considered as a potential weakness. However, excluding the implausible reporters increased the validity of the dietary reports of the subjects and permitted to avoid spurious conclusions. When the analyses were performed on the total sample, associations of obesity indicators with dietary intake, among others total energy and fat intake, were not observed. This finding reinforces the importance of identifying and excluding implausible reporters. A major strength of our study is that dietary intake was not restricted to fat intake but included a large number of dietary factors (total energy intake, macronutrients expressed as both kcal/day and percentage from total energy intake, alcohol, calcium, iron and cholesterol) and that the analyses were adjusted for age and physical activity variables. Another strength is that WC, an indicator of abdominal obesity was also used besides BMI. Finally, it is important to mention that the anthropometric data were not self-reported, but measured by trained staff.