Overweight and obesity are recognised as public health problems worldwide and as major causes of preventable ill health1. Total obesity is the sixth most important risk factor contributing to the overall burden of disease worldwide, being a major risk factor for chronic non-communicable diseases such as hypertension, coronary heart disease, type 2 diabetes, dyslipidaemia, as well as to some hormone-dependent cancersReference Aranceta, Pérez Rodrigo, Serra Majem, Ribas, Quiles-Izquierdo and Vioque2. Abdominal obesity is a strong predictor of coronary heart disease and related risk factorsReference Rexrode, Carey, Hennekens, Walters, Colditz and Stampfer3. Overweight and obesity also have an important health cost associated with themReference Aranceta, Pérez Rodrigo, Serra Majem, Ribas, Quiles-Izquierdo and Vioque2.
The World Health Organisation (WHO) recently reported 1.1 billion overweight individuals and 300 million obese individuals1, 10% of which are overweight or obese childrenReference Haslam and James4. Overweight and obesity prevalence rates are increasing in both developed and developing societies1. The WHO reported that, since 1980, obesity prevalence rates have increased threefold in Northern America (in the United States over 65% of adults are now overweight or obeseReference Townsend5), the United Kingdom, Central and Eastern Europe, Pacific Islands, Australia and China1.
In Spain, adult, adolescent and child obesity prevalence has also increased in the last decadeReference Aranceta-Bartrina, Serra-Majem, Foz-Sala and Moreno-Esteban6, Reference Serra Majem, Ribas Barba, Aranceta Bartrina, Perez Rodrigo, Saavedra Santana and Pena Quintana7. The group of children aged between 6 and 13 years and the group of women aged over-45 years are the groups with the highest risk of obesity; obesity prevalence is higher among males during years of growth and developmentReference Aranceta-Bartrina, Serra-Majem, Foz-Sala and Moreno-Esteban6, Reference Serra Majem, Ribas Barba, Aranceta Bartrina, Perez Rodrigo, Saavedra Santana and Pena Quintana7, while in the over 45-year group it is significantly higher in femalesReference Aranceta-Bartrina, Serra-Majem, Foz-Sala and Moreno-Esteban6, Reference Gutiérrez-Fisac, Regidor and Rodríguez8. In a recent study in Southern Spain, the authors found that a larger proportion of men were overweight compared to women, but the opposite was found for obesityReference Mataix, Lopez-Frias, Martinez-de-Victoria, Lopez-Jurado, Aranda and Llopis9. In 2004, the results from the DORICA StudyReference Aranceta, Pérez Rodrigo, Serra Majem, Vioque, Tur Marí, Mataix Verdú, Aranceta, Foz, Gil, Jover, Mantilla, Millán, Monereo and Moreno10 showed that the obesity prevalence of the North-Eastern region of Spain (which includes Catalonia) was 8.5% for men and 13.8% for women, which were the lowest out of the eight regions included in the studyReference Aranceta-Bartrina, Serra-Majem, Foz-Sala and Moreno-Esteban6.
Numerous studies have shown that obesity is more frequent in the less socially advantaged population groups, regardless of the variable used to classify socio-economic status (SES); these differences in the prevalence of obesity by SES have been observed in both men and women, but are stronger and more consistent in womenReference Sobal and Stunkard11. The WHO’s MONICA Study showed that the prevalence of obesity is higher among adults and children of low SESReference Molarius, Seidell, Sans, Tuomilehto and Kuulasmaa12. In Spain, in 1987, a group of researchers found a higher prevalence of obesity among the population of a lower educational levelReference Gutiérrez-Fisac, Regidor and Rodríguez8; in the period 1987–97, the same researchers found a higher obesity prevalence in individuals with elementary education, and that the obesity prevalence proportion associated with elementary education increased in women and decreased in menReference Gutiérrez-Fisac, Regidor, Banegas Banegas and Rodríguez Artalejo13. Moreover, the SEEDO’97 Study in Spain showed higher obesity rates in men and women with low educational level, and also that older women with low educational level and low income seemed to be the most susceptible group to weight gainReference Aranceta, Pérez Rodrigo, Serra Majem, Ribas, Quiles-Izquierdo and Vioque2. Adding to this evidence, a significant inverse relationship between SES and overweight and obesity was found by the AVENA (Alimentación y Valoración del Estado Nutricional de los Adolescentes Españoles) StudyReference Moreno, Mesana, Fleta, Ruiz, Gonzalez-Gross and Sarria14, although only in male adolescents.
Overweight and obesity also have a sociodemographic component. In this respect, the SEEDO’97 Study in Spain also showed differences in the distribution of the obesity prevalence by area of residence and geographical zonesReference Serra Majem, Ribas Barba, García Closas, Ramon, Salvador and Farran15.
Other well-known factors that influence the development of obesity are physical inactivityReference Jakicic and Otto16, Reference Gutierrez-Fisac, Regidor, Lopez Garcia, Banegas Banegas and Rodriguez Artalejo17, over-consumption of energy-dense diets (which has been shown to be associated with low SES)Reference Rolls, Drewnowski and Ledikwe18 and genetic factors (although some authors do not agree to this)Reference Townsend5.
The objective of the present paper is to evaluate the trends (1992–2003) of overweight and obesity prevalences in the 18–75-year-old population of Catalonia, Spain, and the influence of socio-economic and sociodemographic determinants on these prevalence trends.
Material and Methods
Sample and subjects
The data analysed in this paper belong to the 1992–93 and the 2002–03 cross-sectional Evaluations of the Nutritional Status of the Catalan Population (ENCAT 1992–93 and ENCAT 2002–03)Reference Serra Majem, Ribas Barba, García Closas, Ramon, Salvador and Farran15, Reference Serra Majem, Ribas Barba, Salvador Castell, Castells Abat, Román Viñas and Serra19. ENCAT is a regional survey carried out periodically by the Department of Health of the Catalan Government and co-ordinated by the Centre for Research on Community Nutrition of the University of Barcelona. The theoretical random sample population and sample size have been described elsewhereReference Serra Majem, Ribas Barba, García Closas, Ramon, Salvador and Farran15, Reference Serra Majem, Ribas Barba, Salvador Castell, Castells Abat, Román Viñas and Serra19, comprising the population source of residents in the official census. The samples were stratified according to household and randomised into sub-groupings with municipalities being the primary sample units, and individuals within these municipalities comprising the final sample units. The valid response rate for the first survey was 69% and for the second 65%.
Adults from each representative sample within the age of 18–75 years were included in the analysis of this study (n in ENCAT 1992–93 = 2248 and n in ENCAT 2002–03 = 1715).
Data collection procedures and variables of the study
In both surveys, dietitians were trained on standardisation of criteria and methodology before data collection, in order to reduce inter-observer measurement variability. The data were collected from 1992 to 1993 and from 2002 to 2003 through questionnaires and anthropometric measurements during a home interview.
In order to analyse the influence of the socio-economic determinants on the prevalence of overweight and obesity, the following variables were used and rearranged according to the following categoriesReference Aranceta, Pérez Rodrigo, Serra Majem, Ribas, Quiles-Izquierdo and Vioque2:
1. Socio-economic level (SEL) (occupation of the subject): (a) low: the non-classifiable, army, agricultural sector, service sector and non-qualified labourers; (b) medium: qualified labourers, foremen, rest of administrative, commercial and technical staff and medium-level technicians; (c) high: high-level technicians, directors/managers, self-employed professionals, business owners or self-employed individuals without staff, business owners or self-employed individuals with staff.
2. Education level of the subject and of the family’s head member (ELS and ELH): (a) low: primary school incomplete or illiterate (<6 years at school); (b) medium: primary school completed, secondary school or further education (6–12 years of education); (c) high: high school, college or university degree (>12 years of education).
The sociodemographic determinants included (1) gender, (2) age group (18–24, 25–44, 45–64 and 65–75 years) and (3) population of residence size (<10,000 inhabitants, 10,000–100,000 inhabitants and >100,000 inhabitants).
Anthropometric measurements
Body mass index
Weight and height were measured with a portable spring scale and a metric tape (Kawe© model). The individuals were measured in standardised conditions, wearing underwear and no shoes. Weight was measured in kilograms, scale measurement error ±100 g. Height was measured standing and head in the Frankfurt horizontal position, expressed in centimetres, instrumental measurement error ±0.1 cm. Body mass index (BMI) was calculated using weight and height and categorised according to WHO criteria20 so that overweight was defined as BMI ≥25.0 to BMI <30.0 kg m−2 and obesity as BMI ≥30 kg m−2.
Waist circumference
Waist circumference (WC) was measured with a non-elastic metric tape halfway between the lower border of the ribs and the iliac crest on a horizontal plane. Measurements were recorded to the nearest 0.1 cm and categorised according to WHO criteria, so that men with a WC 94.0–101.9 cm and women with a WC 80.0–87.9 cm were classified as overweight, and men with a WC ≥ 102.0 cm and women with a WC ≥ 88.0 cm were classified as obese20.
Statistical analysis
All analyses were performed with SPSS 12.0. Proportions of overweight and obesity were estimated for each sample separately and stratified by gender and age (to control for its potential confounding effects). The age distribution of the whole Catalan population in 1992–93 was used as a reference. The proportions from the two surveys were compared using the χ 2 statistic test and the means were compared using the t-test, considering P-values <0.05 for significance.
Results
The sample characteristics of the two surveys are presented in Table 1: the total number of subjects by gender, age group and each socio-economic/sociodemographic variable category.
*Body mass index – according to WHO classification (1998).
†Waist circumference – according to WHO classification (1998): normal: <94 cm for males and <80 cm for females; overweight: ≥ 94 to <102 cm for males and ≥80 to <88 cm for females; obese: ≥102 cm for males and ≥88 cm for females.
ENCAT – Evaluation of Nutritional Status in Catalonia.
Table 2 shows the mean, standard deviations and 5th–95th percentiles of BMI and WC by gender and age group. In 2002–03, male mean BMI was higher than in 1992–93, although the observed difference was significant only for individuals aged 25–44 years (from 25.2 to 25.9) and 45–64 years (from 26.7 to 27.4), and male mean WC was significantly higher in all age groups; for females the observed decreasing trends in mean BMI in most age groups (except for the youngest) were not significant, while mean WC was significantly higher only in the youngest (from 70.3 to 72.7) and eldest (from 92.2 to 95.3) individuals. These results are shown in Figs 1 and 2, which also show how mean BMI and WC increase as age progresses in both genders. Percentiles 50, 75 and 95 of BMI showed increases in males from all age groups and decreases in females (except for the youngest group). As for WC, percentiles 50, 75 and 95 showed increases in males and females of all age groups.
*ns – non-significant difference; **s – significant difference (P-value <0.05).
†t-Test used for between-survey comparison of BMI and WC means.
SD = standard deviation; P = percentile; ENCAT – Evaluation of Nutritional Status in Catalonia
Table 3 shows overall by-gender and by-survey BMI and WC overweight and obesity prevalences; it can be observed that the overall prevalence of BMI obesity increased significantly only in males (6.7 percentage points, from 9.9% to 16.6%) in the 10-year period, while that of WC obesity increased in both sexes (11.3 percentage points for males – from 13.1% to 24.4%, and 6.6 for females – from 24.5% to 31.1%). Table 3 also shows BMI and WC overweight and obesity prevalences when age, SEL, education level and population of residence size are considered.
* s: χ 2 test – significant difference (P ≤ 0.05); **ns: χ 2 test – non-significant difference.
†BMI 25.0 to ≤30.0 kg m−2;
‡ BMI ≥ 30.0 kg m−2;
§ WC 94 to <102 cm for males and 80 to <88 cm for females.
¶ WC ≥ 102 cm for males and ≥88 cm for females.
∥ See Table 4 for adjusted prevalences.
ENCAT – Evaluation of Nutritional Status in Catalonia; BMI – body mass index; WC – waist circumference.
When considering the variable ‘age’, Table 3 shows that in ENCAT 1992–93, the highest prevalence of BMI overweight was found in both males and females aged 45–64 years, which was also the case for female but not for male WC overweight; while in ENCAT 2002–03, only an increase in female BMI overweight and male WC overweight were observed with progressing age. Regarding obesity, both surveys showed an increase in the prevalence of BMI and WC obesity with progressing age in both sexes (note the high prevalence of WC obesity among the eldest men and women in 2002–03, 49.6% and 70.9% respectively). The between-survey comparison showed significant changes only in male BMI overweight and male WC obesity rates, demonstrating alarming increases in the latter rates (i.e. from 1.3% to 6.0% in the 18–24-year-old group).
Regarding the variable ‘socio-economic level’ (SEL), the differences observed in BMI and WC overweight and obesity prevalences of the different SEL groups were significant only in females of both surveys, WC obesity being highest in the low SEL group (Table 3). In ENCAT 1992–93, SEL was inversely related to the prevalence of BMI obesity, but only significantly in females; this inverse relationship was not observed among SEL groups in ENCAT 2002–03. WC obesity was only inversely related with SEL in females of both surveys and the differences among SEL groups were significant (Table 3). The between-survey comparison showed significant increases in male BMI and WC obesity (from 8.3% to 16.5% and from 13.3% to 26.3%) and female WC obesity (from 15.3% to 19.3%).
With regard to the variable ‘education level of the subjects’ (ELS), in ENCAT 2002–03, Table 3 shows an inverse relationship with BMI overweight and obesity prevalence, but with differences among ELS groups only significant in females (note a female BMI obesity prevalence of 36.6% in the lowest ELS group); this inverse relationship was observed between ELS and male and female WC obesity but not WC overweight (note the high male and female WC obesity rates in the lowest ELS group, 34.3% and 69.5% respectively). The between-survey comparison revealed significant differences in both BMI and WC overweight and obesity for both genders. It is worth noticing that while male and female WC overweight seem to have increased among the highest ELS group (from 15.5% to 23.9% and from 11.8% to 14.9% respectively), female WC obesity prevalence in the lowest ELS group increased by 20 percentage points (from 49.3% to 69.5%).
The variable ‘education level of the family head member’ (ELH) showed in ENCAT 2002–03 a significant inverse relationship with BMI overweight, BMI obesity and WC obesity in females (Table 3). Females whose family head member had a medium education level presented the highest WC overweight prevalence compared to females whose family head member had a low or high ELH (23.9% in 1992–93 and 19.0% in 2002–03). The between-survey comparison revealed significant differences in both BMI and WC overweight and obesity for both genders. Male and female BMI overweight, male and female BMI obesity, male WC overweight and male and female WC obesity rates increased in the low and high ELH groups, while in the medium ELH group the increase was only observed in male BMI overweight (from 9.6% to 17.5%) and male and female WC obesity (from 12.9% to 24.9% and from 23.3% to 30.2% respectively). The difference of 27 percentage points in female WC obesity prevalence in the low ELH group is worth noticing (from 35.1 in ENCAT 1992–93 to 61.9% in ENCAT 2002–03).
Regarding ‘population of residence size’, only ENCAT 2002–03 differences observed in female WC overweight and obesity were significant (‘within-survey comparison’, Table 3). The between-survey comparison showed significant differences in all prevalences except for males BMI and WC obesity; it is worth mentioning that, in the <10,000 inhabitants group, while female BMI and WC overweight rates decreased (from 32.5 to 27.8% and from 22.5 to 12.9%, respectively), female BMI and WC obesity rates increased (from 15.6 to 18.0 and from 31.1 to 38.6%, respectively).
Table 4 shows how the ENCAT 2002–03 BMI and WC overweight and obesity prevalences change when adjusting by the ENCAT 1992–93 SEL, ELS, ELH and population of residence size distributions. It is apparent that male and female BMI obesity increased to its highest when standardised by the ENCAT 1992–93 ELS and population size (from 16.6% to 17.8% and from 15.2% to 19.6%, respectively), while male and female WC obesity increased to its highest when standardised by the ENCAT 1992–93 ELS (from 24.4% to 26.7% and from 31.1% to 39.2%, respectively). Male and female BMI and WC obesity decreased when adjusted by the ENCAT 1992–93 SEL distribution.
* BMI 25 to <30 kg m−2 and WC 94 to <102 cm for males and 80 to <88 cm for females, according to WHO classification (1998).
† BMI ≥ 30 kg m−2 and WC ≥ 102 cm for males and ≥88 cm for females, according to WHO classification (1998).
‡ SEL – socio-economic level.
§ SR – standardised rate.
¶ ELS – education level of the subject.
∥ ELH – education level of the family head member.
ENCAT – Evaluation of Nutritional Status in Catalonia; BMI – body mass index; WC – waist circumference.
Discussion
The WHO recognises that the main limiting factors when comparing epidemiological studies on the prevalence of overweight and obesity are the following: the different criteria to define the cut-offs, the variation in age groups considered, the time interval for collection of data and study comparisons based on reported weight and height20. This study is based on the 1992–93 and 2002–03 ENCAT surveys, which were carried out on representative random samples of the Catalan population. Both surveys used the same anthropometric measurement procedures (weight, height and WC were measured instead of reported) and socio-economic factors, and allow for comparison of the same age groups (18–75 years).
The WHO has recommended BMI as a good index of total overweight and obesity20, although it gives no information about body fat distribution, while WC reflects abdominal visceral fat distribution. Nevertheless, the two measures are highly correlatedReference Sarlio-Lähteenkorva, Silventoinen, Lahti-Koski, Laatikainen and Jousilahti21. It has been shown that changes in WC accompany changes in cardiovascular risk factors especially in the elderlyReference Turcato, Bosello, Di Francesco, Harris, Zoico and Bissoli22. Research has also shown that WC can also predict morbidity and mortality, considering it a better measure of obesity than BMI, since it is a simple and easy measurementReference Lean, Han and Seidell23; WC is even more strongly associated with metabolic abnormalities and healthcare costs than BMIReference Sarlio-Lähteenkorva, Silventoinen, Lahti-Koski, Laatikainen and Jousilahti21. A single WC measurement has been suggested to be used to identify individuals who should seek and be offered weight managementReference Lean, Han and Morrison24. We have used both BMI and WC measures to define total and central overweight and obesity in order to have a more complete overall picture of the problem in the Catalan population.
This study has shown that in Catalonia, in 2002–03, mean BMI in males was higher than in 1992–93, and that of females was lower (except for the youngest group); on the other hand, overall prevalence of BMI overweight and obesity were 43.7% and 16.6%; respectively in males and 30.1% and 15.2% respectively in females. When comparing these figures with those of the 2002 IOTF report for Spain (1998–2000), we observe that overweight was lower in Catalonia in both genders (in Spain 48% for males and 40% for females), while obesity was higher for Catalan males and females (in Spain 12% and 15%, respectively)25. Therefore, in terms of gender, this study shows that overweight and obesity are more prevalent in men (obesity was more prevalent in women 13 years ago, but male obesity has caught up and overcome the female prevalence). These findings are in agreement with other literature available from developed countries, which suggests that women hold a more negative attitude towards obesity than men and they are also more heavily influenced by the public negative view towards obesity, thus spending more time, effort and money on the ideal thinner shapeReference Manios, Panagiotakos, Pitsavos, Polychronopoulos and Stefanadis26.
Regarding WC overweight and obesity prevalences, this study has shown that in ENCAT 2002–03, mean WC was higher in males and females as compared to ENCAT 1992–93 (except for the female group aged 45–64 years). In men, overall WC overweight increased (from 21.7% in 1992–93 to 23.8% in 2002–03) as well as overall WC obesity (from 13.1% to 24.4%) increased. In women, overall WC overweight decreased (from 21.8% to 17.7%), while overall WC obesity increased (from 24.5% to 31.1%). In other words, our results on BMI and WC overweight and obesity suggest that Catalan men are getting bigger overall and also specifically around the waist, while Catalan women are getting thinner overall but with bigger waistlines. They also show that WC obesity is increasing more rapidly than BMI obesity and, while BMI obesity is more prevalent among men, WC obesity is more prevalent among women. These findings agree with those of several recent studies carried out in Northern EuropeReference Sarlio-Lähteenkorva, Silventoinen, Lahti-Koski, Laatikainen and Jousilahti21, Reference Chen and Tunstall-Pedoe27.
There are few studies that examine the possible relationship of SES and overweight and obesity prevalence, and even fewer for the actual distribution of its prevalence into the SES groupsReference Sarlio-Lähteenkorva, Silventoinen, Lahti-Koski, Laatikainen and Jousilahti21, Reference Manios, Panagiotakos, Pitsavos, Polychronopoulos and Stefanadis26, Reference Choiniere, Lafontaine and Edwards28. Although comparisons are not directly possible, there are three studies that show that obesity rates have been increasing for decades and are in line with our findings in that the prevalence of obesity is higher for the lower SES groups (two of these studies use educationReference Berkman and Breslow29, Reference Lynch, Kaplan and Salonen30 and one uses incomeReference Choiniere, Lafontaine and Edwards28) and for men. Data from the ENCAT 2002–03 survey showed an increasing trend in the prevalence of BMI obesity in all male SEL (using occupation) groups as compared to ENCAT 1992–93, whereas female BMI obesity prevalence only increased in the high SEL group (although not significantly). The analysis showed that SEL had no influence on male BMI overweight or obesity prevalence, and that it only had an influence on BMI overweight and obesity prevalence among the oldest females (45–64 years and 65–75-year-olds), showing an inverse relationship (this further stratification by age group is not shown in the results). Referring to WC, in the 10-year period, only female WC overweight and obesity changed due to SEL, overweight decreased (being highest in the lowest SEL but no inverse relationship was observed) and obesity increased (highest in the lowest SEL, showing an inverse relationship). These findings are in agreement with numerous studies carried out in developed countries by which, overall, the prevalence of obesity is higher in lower SEL groupsReference Aranceta, Pérez Rodrigo, Serra Majem, Ribas, Quiles-Izquierdo and Vioque2, Reference Sarlio-Lähteenkorva, Silventoinen, Lahti-Koski, Laatikainen and Jousilahti21, Reference Stam-Moraga, Kolanowski, Dramaix, De Backer and Kornitzer31. In developing countries the problem has been shown to be more prevalent among the highest SEL groups, some showing the inverse relationship between overweight/obesity and household amenities in both genders and occupational level in menReference Fezeu, Minkoulou, Balkau, Kengne, Awah and Unwin32.
Studies within the Spanish population have shown that the prevalence of obesity is higher among women and increases with age, particularly in the least educated female subgroupsReference Aranceta, Pérez Rodrigo, Serra Majem, Ribas, Quiles-Izquierdo and Vioque2, Reference Gutiérrez-Fisac, Regidor and Rodríguez8, results that agree with the findings of the present study. The further stratification of each education level by age group (not shown in the results) revealed different prevalences from the overall male and female BMI obesity prevalences probably because the least educated people were mostly the older age groups with a higher obesity prevalence, which agrees with findings from the SEEDO’97 StudyReference Aranceta, Pérez Rodrigo, Serra Majem, Ribas, Quiles-Izquierdo and Vioque2. Regarding WC, we have shown that overweight basically increased in the male and female highest ELS groups, while obesity increased in all ELS groups, being highest in the lowest ELS one (inverse relationship) and most prevalent among the females of this group (reaching an alarming prevalence of 69.5%), and affecting more prominently the 45–64-year-olds (again probably because the least educated people were mostly the older age groups with a higher WC obesity prevalence – analysis not shown). These results coincide with those obtained in recent studiesReference Molarius, Seidell, Sans, Tuomilehto and Kuulasmaa12, Reference Chen and Tunstall-Pedoe27, Reference Gutiérrez-Fisac, López, Banegas, Graciani and Rodríguez-Artalejo33, in particular, a study carried out in Spain, which showed an even higher prevalence of WC obesity in non-educated elderly females (80.9%)Reference Gutiérrez-Fisac, López, Banegas, Graciani and Rodríguez-Artalejo33.
With regard to the influence of the area of residence (population size) on excess body weight, in the 10-year period, significant differences were found for BMI overweight in both sexes and for female BMI obesity; the differences were also significant for WC male and female overweight and female WC obesity. Females living in the smallest communities showed a decrease in BMI and WC overweight. However, this decrease was probably at the expense of an increase in female WC obesity. Studies carried out on the Spanish populationReference Aranceta, Pérez Rodrigo, Serra Majem, Ribas, Quiles-Izquierdo and Vioque2, Reference Aranceta, Pérez Rodrigo, Serra Majem, Ribas, Quiles Izquierdo and Vioque34 disagree with our BMI findings by not showing significant differences on overweight and obesity when stratifying by population size, but no comparable results are available for WC prevalences.
Finally, SES has been found to be associated with dietary patterns and physical activityReference Manios, Panagiotakos, Pitsavos, Polychronopoulos and Stefanadis26, Reference Stam-Moraga, Kolanowski, Dramaix, De Backer and Kornitzer31, Reference Proper, Cerin, Brown and Owen35, Reference Aranceta, Pérez, Marzana, Egileor, González de Galdeano and Sáenz de Buruaga36. For example, showing more disadvantaged population groups generally have a poorer-quality diet (e.g. higher fat intake and lower vegetable consumption) than higher SES groups, which may partly explain the inverse association between SES and obesity demonstrated in some studiesReference Manios, Panagiotakos, Pitsavos, Polychronopoulos and Stefanadis26, Reference Proper, Cerin, Brown and Owen35. Other studies have evaluated how money expenditure on food can assist in the achievement of a healthy dietReference Manios, Panagiotakos, Pitsavos, Polychronopoulos and Stefanadis26, Reference Drewnowski and Specter37. The inverse relationship between energy density and energy cost suggests that ‘obesity-promoting’ foods are simply those that offer the most dietary energy at the lowest cost. The relative cost has also been taken into account in the literature which increases even further the cost of the healthy diet for the low-income familiesReference Manios, Panagiotakos, Pitsavos, Polychronopoulos and Stefanadis26. The present study has not considered diet, physical activity, income (at least not directly), expenditure on food or food costs in its analysis (which was merely descriptive and far from suggesting causality due to the cross-sectional nature of the data); therefore, the authors recognise the need for a further and more robust analysis that involves all these lifestyle variables known to affect the relationship between prevalence of excess body weight and SES. In addition, self-reported occupation and education level may be over or under estimated. However, this probably has not significantly modified the classification of the participants into the three SES groups. Moreover, this study has not adjusted WC for BMI, which should be done due to the influence a high BMI can have on a high WCReference Sarlio-Lähteenkorva, Silventoinen, Lahti-Koski, Laatikainen and Jousilahti21. In spite of the mentioned limitations, we believe that our findings contribute to the evidence needed to guide public health policy makers in the design and implementation of preventive campaigns against the increasing trends of overweight and obesity, paying special attention to males and low SEL and education-level groups, and small population of residence size (for male overweight and female obesity).
Acknowledgements
Sources of funding:This work was made possible by financing from the General Division of Public Health of the Generalitat of Catalonia’s Department of Health, through a research agreement with the Fundación para la Investigación Nutricional (Nutrition Research Foundation).
Conflict of interest declaration: None of the authors had any conflicts of interest in connection with this study.
Authorship responsibilites: AGA was responsible for the statistical analysis, data interpretation and writing of the paper; LSM was director of the study and revised the paper providing expert advice on data interpretation and discussion of the paper; LRB was coordinator of the study and responsible of the databases and revised the paper providing expert advice on data interpretation; CC participated in the study concept and design; MF, RU, AP and LS revised the paper providing expert advice in its discussion.
Guarantor: Lluís Serra-Majem.
Acknowledgements: Part of this analysis belongs to Alicia García-Álvarez's MSc thesis at the London School of Hygiene and Tropical Medicine (2004–05) supervised by Ricardo Uauy.
Special acknowledgement is made to all those persons who were interviewed, and whose collaboration made the realisation of these surveys possible.
Research Group on the Evaluation and Monitoring of the Nutritional Status in the Catalan Population: Lluís Serra-Majem, Director (University of Las Palmas de Gran Canaria); Lourdes Ribas-Barba, Coordinator (FIN-Nutrition Research Foundation, Barcelona Science Park); Gemma Salvador (Generalitat of Catalonia); Conxa Castell (Generalitat of Catalonia); Blanca Román- Viñas (FIN, Barcelona Science Park); Jaume Serra (Generalitat of Catalonia); Lluís Jover (University of Barcelona); Ricard Tresserras (Generalitat of Catalonia); Blanca Raidó (FIN, Barcelona Science Park); Andreu Farran (CESNID, University of Barcelona); Joy Ngo (FIN, Barcelona Science Park); Mari Cruz Pastor (Hospital Germans Trias i Pujol, Badalona); Lluís Salleras (University of Barcelona); Ricardo Uauy (London School of Hygiene and Tropical Medicine (England), and Carmen Cabezas, Josep Lluís Taberner, Salvi Juncà, Josep Maria Aragay, Gonçal Lloveras Vallès († 2003), Antoni Plasencia (Generalitat of Catalonia).