Associations between food patterns, socioeconomic position ... - Nature

European Journal of Clinical Nutrition (2010) 64, 1150–1157

& 2010 Macmillan Publishers Limited All rights reserved 0954-3007/10 www.nature.com/ejcn

ORIGINAL ARTICLE

Associations between food patterns, socioeconomic position and working situation among adult, working women and men in Oslo MK Ra˚berg Kjøllesdal1, G Holmboe-Ottesen2 and M Wandel1 1 Department of Nutrition, Institute for Basic Medical Sciences, University of Oslo, Oslo, Norway and 2Department of General Practice and Community Medicine, Institute of Health and Society, University of Oslo, Oslo, Norway

Background/Objectives: Socioeconomic disparities in diet are well documented, but the relative importance of different indicators of socioeconomic position (SEP) is not well known. The aim of this study was to explore relationships between food patterns, SEP (occupation, education and income) and degree of work control. Subjects/Methods: A cross-sectional population-based study 2000–2001, using three self-administered questionnaires including food frequency questions (FFQs). Factor analysis was used to explore food patterns. Participants include 9762 working Oslo citizens, 30–60 years of age, having answered the questionnaires with o20% of the FFQ missing. Results: Four food patterns were found: Western, prudent, traditional and sweet. In multivariate analyses, the likelihood of having a high intake of the Western pattern was lowest in the two highest educational groups (women: odds ratio (OR) ¼ 0.54/ OR ¼ 0.75; men: OR ¼ 0.51/OR ¼ 0.76), and in the two highest occupational groups for men (OR ¼ 0.73/OR ¼ 0.78). The odds of having a high intake of the prudent pattern was highest in the two highest educational groups (women: OR ¼ 2.50/ OR ¼ 1.84; men: OR ¼ 2.23/OR ¼ 1.37), and among the self-employed (women OR ¼ 1.61, men OR ¼ 1.68), as well as in the highest occupational group for men (OR ¼ 1.33). Women always having work control were least likely to have high intake of the Western pattern (OR ¼ 0.78) and most likely to have high intake of the prudent pattern (OR ¼ 1.39). Conclusions: The SEP indicators were in different ways related to the food patterns, but the effect of occupation and income was partly explained by education, especially among women. Women’s work control and men’s occupation were important for their eating habits.

European Journal of Clinical Nutrition (2010) 64, 1150–1157; doi:10.1038/ejcn.2010.116; published online 11 August 2010 Keywords: food patterns; SEP; work control; gender

Introduction It is well documented that lower socioeconomic position (SEP) is associated with poorer health (The World Bank, 1993; Wilkinson and Marmot, 2006), even in highly developed welfare societies (Mackenbach et al., 1997; Kunst et al., 1999; Silventoinen and Lahelma, 2002; Mackenbach et al., 2003; Strand and Tverdal, 2004; Roos et al., 2005; Rognerud and Zahl, 2006). Much of the social inequalities in health are due to lifestyle diseases (Kunst et al., 1999; Mackenbach et al., 2003; Strand and Tverdal, 2004;

Correspondence: MK Ra˚berg Kjøllesdal, Department of Nutrition, Institute for Basic Medical Sciences, University of Oslo, PB 1046 Blindern, 0316 Oslo, Norway. E-mail: [email protected] Received 13 September 2009; revised 7 June 2010; accepted 8 June 2010; published online 11 August 2010

Avendano et al., 2006). Previous studies indicate a more healthy diet among those in higher social classes (Perrin et al., 2002; Giskes et al., 2006). Commonly used indicators of SEP are education, income and occupation. Although related, these indicators represent different aspects of the relationship between SEP and health, and can therefore not be used interchangeably (Galobardes et al., 2007). Education may influence diet through the comprehension of nutritional information and food labelling. Income determines the potential to afford preferred food. Occupation may affect eating habits during work time, social environment and the lifestyle of colleagues. Furthermore, physical and mental work strain and degree of work control have been reported to influence eating habits (Wickrama et al., 1995; Lallukka et al., 2004; Raulio et al., 2008). Shift/night workers may in addition encounter special challenges with regard to where and what to eat.

Food patterns and socioeconomic position MK Ra˚berg Kjøllesdal et al

1151 All three indicators of SEP have shown to be associated with dietary habits (Johansson et al., 1999; Mishra et al., 2002; Perrin et al., 2002; Robinson et al., 2004; Metcalf et al., 2006; Lallukka et al., 2007; Petkeviciene et al., 2007; Arkkola et al., 2008). Previous studies most often use only one or two of the indicators. It is interesting to explore the relative importance of the different SEP indicators, to see which one is most strongly associated with food habits, and if the impact of one is explained by that of another. Food patterns may be explored with the help of factor analysis. This makes it possible to study dietary disparities in a holistic way, not only for single food items or nutrients, as often previously carried out (Perrin et al., 2002; Giskes et al., 2006). Research in this area is scant. The aim of this paper is to explore associations between food patterns and different indicators of SEP, as well as of the working situation in a welfare society. All three indicators of SEP (education, income and occupation) are used in the analyses. Degree of control over own work and shift/night work are chosen to reflect aspects of the working situation that may affect eating habits.

Methods Design The Oslo Health Study was conducted in 2000–2001 by the Norwegian Institute of Public Health, in collaboration with the Oslo City Council and the University of Oslo. All inhabitants born in seven different age groups were invited to participate. The study involved the use of three questionnaires and a health examination. The participants received a letter of invitation and the main questionnaire in the mail, and two additional questionnaires at the examination, which they were asked to complete and return in a prepaid envelope. The study design and questionnaires are described in detail at http://www.fhi.no/hubro-en. The study was approved by the Norwegian Data Inspectorate and the Regional Committee for Medical Research Ethics.

Sample This article is based on analysis of data from persons born in 1940, 1941, 1955, 1960 and 1970. Of 34 151 invited persons, 15 186 came to the health examination and/or answered at least one questionnaire. The overall attendance rate was 44.5%. For this analysis, 19% were excluded because they did not return both questionnaires containing food frequency questions (FFQs), and a further 8% were excluded due to X20% missing responses to the food frequency items. The excluded participants were less likely than those included to be female (Po0.001), born in 1940–41 (Po0.001), and from the highest educational group (Po0.001). However, the two groups were similar in income distribution. We also excluded those with no reported work (1314 persons). The total number of persons included in the analyses was 9762.

FFQs The questionnaires contained questions about 82 foodrelated items (68 foods, 13 drinks and 2 supplements), covering intake of bread, bread spreads, dinner dishes, sauces/dressings, cakes/sweets, fats, fruit, vegetables, milk, fruit juice, soft drinks and alcohol. The FFQs have earlier been validated against intake of the matching food/food group based on a 14-day diet diary (Mosdøl, 2004). Missing values (2.4% of values) for the food groupings were replaced with the lowest value (‘seldom/never’). In all, 67 nonoverlapping food groupings from the FFQs were included in the factor analyses.

Socioeconomic and demographic factors Years of schooling were recoded into groups according to the Norwegian education system: ‘p high school (p 12 years), ‘college/university lower’ (13–16 years) and ‘college/ university higher’ (X17 years). Personal income per year was measured in eight groups and recoded into ‘0–200 000 NOK per year’ (0–25 000 h), ‘200 000–300 000 NOK per year’ (25 000–38 000 h) and ‘4300 000 NOK per year’ (438 000 h). The occupational groups were constructed after the Erikson– Goldthorpe’s scheme, with seven categories (Erikson, 1992): i Higher-grade professionals, administrators and officials; managers in large industrial establishments; large proprietors. ii Lower-grade professionals, administrators and officials; higher-grade technicians; managers in small industrial establishments; supervisors of non-manual employees. iii Routine non-manual employees, higher and lower grade. iv Small proprietors, artisans, farmers and smallholders; other self-employed workers in primary production. v Lower-grade technicians, supervisors of manual workers. vi Skilled manual workers. vii Semi- and unskilled manual workers. These occupational categories were further divided into four groups for regression analyses; higher- or lower-grade professionals (group I and II), routine non-manual employees (group III), artisans and self-employed workers in primary production (group IV) and manual workers (group V–VII). The variables related to working situation were shift/night work (0 ¼ ‘no’, 1 ¼ ‘yes’) and control over own working situation, assessed through a question about being able to make decisions about own work. This was recoded from four categories to 1 ¼ ‘never/seldom’, 2 ¼ ‘most often’ and 3 ¼ ‘always’. In addition, the following variables were used. Age was divided into ‘30 years’, ‘40–45 years’ and ‘59–60 years’. Region of origin was assessed through a question about mother’s country of birth, and recoded into: 0 ¼ ‘Norway’, 1 ¼ ‘other Western countries’ (Western Europe, North America and Australia), 2 ¼ ‘non-Western countries’ (East Europe, Middle East, Africa, Asia, the Pacific and Latin America). European Journal of Clinical Nutrition


1152 Household composition was coded 0 ¼ ‘living alone’, 1 ¼ ‘living with only adults’, 2 ¼ ‘living with children (o18 years)’.

Analyses Data were analysed in SPSS 14.0 (SPSS Inc., Chicago, IL, USA). Different food patterns were identified using factor analysis with varimax rotation. Varimax rotation was chosen to maximize the variance of factor loadings, and thereby make the factors more interpretable. Each food grouping used to characterize a pattern had factor loadings of X0.35. Four food patterns were identified with eigenvalues 42. A scree plot also supported a four-factor solution. Separate analyses for men and women gave the same patterns with close to the same food groups loading above 0.35. Thus, we chose to report the results from factor analysis for the collected sample (also including those with no reported work, n ¼ 1314). Labelling of the factors was based on our interpretation of the factor structures. Factor scores were divided into tertiles. Cross tables and w2-tests were used to analyse differences between proportions of low SEP groups in each tertile of the Western and the prudent food patterns. Multiple logistic regressions were used to analyse likelihoods of being in the highest tertile of the Western and prudent patterns as functions of SEP and work situation. The analyses were carried out in several steps. The independent variable in model 1 was one SEP- or work-related variable, adjusted for demography. The other SEP variables were added in steps, in the order education, occupation and income, and then work control and shift work (simultaneously), resulting in model 2. Thus, model 2 included all SEP- and work-related variables, adjusted for each other and demography. Only models 1 and 2 are shown in tables. The independent variables were checked for collinearity, and there was no need for exclusion. Significance level was set to Po0.05. There were significant (Po0.001) interactions between SEP and gender, thus we analysed women and men separately.

Table 1 Proportion of men and women in different groups of age, ethnic background, socioeconomic position and work situation. % Women (n ¼ 5355)

% Men (n ¼ 4916)

Age (years) 30 40–45 59–60

26.6 43.7 29.6

26.9 40.4 32.7

Region of origin Norway Other Western country Non-Western country

88.0 7.4 4.6

86.3 6.1 6.5

Household composition Living alone With adults, but no children With children

24.3 35.5 40.1

22.7 30.2 47.1

33.3 33.8

31.8 33.7

32.9

34.5

Income 0–200 000 NOK 200 000–300 000 NOK X300 000 NOK

29.7 45.5 24.7

12.9 27.5 59.6

Occupational group I II III IV V VI VII

15.1 10.0 62.3 5.8 0.6 1.8 4.4

29.5 13.2 26.5 11.5 3.9 7.0 8.3

Control over own work Seldom/never Most often Always

31.5 55.0 13.5

19.9 57.9 22.2

Shift/night work

15.7

15.2

Education pHigh school (12 years) Lower college/university education (13–16 years) Higher college/university education (X17 years)

Difference between men and women: all variables Po0.05, except: 30 years, Norwegian origin and shift work.

Results Characterization of the sample The distribution of the sample according to demographic and socioeconomic factors is described in Table 1. Almost one-fourth lived alone, and less than half of the sample lived with children o18 years of age. Around two-thirds of the sample had education from university or college. More than half of the men and about one quarter of the women were in the highest income group. Close to two-thirds of the women were employed in routine non-manual work, and almost one-third of the men were employed in the highest occupational group. More than half of the sample reported to most often have control over own working situation. A higher proportion of women than men reported low work European Journal of Clinical Nutrition

control. In all, 15% of both women and men were shift or night workers.

Characterization of food patterns We identified four food patterns through factor analysis. The ‘Western’ food pattern was characterized by high factor loadings for French fried potatoes, hot dogs, hamburgers, be´arnaise sauce, coleslaw, pizza, potato salad/mashed potatoes, crisps, mayonnaise and soft drinks with sugar (Table 2). The ‘prudent’ food pattern was based on fruit, vegetables (cooked and raw), dishes with fish (other than fish fingers), beans/lentils, shellfish, oil, oil-based dressings and sour


1153 Table 2 Results from factor analysis Interpreted food pattern

Food name

Loading coefficient

Cumulative variance explained

Western

French fries Hot dog, hamburger Be`rnaise sauce Coleslaw Pizza Potato salad/mashed potato Crisps Mayonnaise Soft drink with sugar

0.552 0.512 0.489 0.443 0.439 0.432 0.437 0.412 0.353

6

Prudent

Oil-based dressing Dishes with fish Fresh vegetables Oil Fruit Cooked vegetables Sour cream Shellfish Beans, lentils

0.552 0.536 0.535 0.468 0.436 0.435 0.427 0.398 0.360

12

Traditional

Boiled potato Sauce Melted butter Rice Oil Spaghetti/pasta/macaroni

0.741 0.520 0.466 0.359 0.372 0.424

17

Sweet

Cake, sweet biscuit Dessert Bun Chocolate, sweets Icecream Jam Danish pastry Cheese Waffle

0.602 0.521 0.488 0.443 0.422 0.419 0.390 0.370 0.360

20

Factor loadings of X0.35 are presented. The factor loadings are rotated.

cream. The ‘traditional’ Norwegian diet was characterized by boiled potatoes, sauce and melted butter and less rice, spaghetti/macaroni/pasta and oil. The ‘sweet’ pattern was high in sugar, with high factor loadings for cakes/sweet biscuits, desserts, buns, chocolate/sweets, ice cream, jam, Danish pastry, cheese and waffles. These four factors explained 20% of the total variance.

Crude associations between food patterns and SEP Table 3 shows how women and men in the lowest SEP groups were distributed in the different tertiles of factor scores for the Western and the prudent food patterns. The largest proportion of both women and men who were in the lowest educational group (o12 years) or had manual work fell into the lowest tertile for the factor scores for the prudent pattern. Among men, the largest proportion of the manual workers was in the highest tertile of the Western food pattern. The distribution pattern according to income was less consistent. Western food pattern Multivariate logistic regressions were carried out in several steps to study the relationships between food patterns, SEP and working condition more thoroughly. The likelihood of being in the highest tertile of the Western food pattern was lower among the higher educational groups for both genders, adjusted for demographic (Table 4, model 1) and other SEP and work-related variables (model 2). Women and men in the two highest occupational groups and selfemployed women were less likely than the manual workers to be in the highest tertile of the Western food pattern (model 1). For men, these differences remained significant after adjustments (model 2). For women, the difference was no longer significant for the routine non-manual workers and the self-employed when adjusting for education. The odds ratio (OR) in the highest occupation group

Table 3 Proportions of participants in low income and low educational groups and of manual workers in the tertiles of Western and prudent food pattern Women Low income (o200 000 NOK) Western Low tertile (%) Middle tertile (%) High tertile (%) P-valuea

36.4 29.8 33.8 0.001

Prudent Low tertile (%) Middle Tertile (%) High tertile (%) P-valuea

33.7 33.0 33.3 NS

Low education (p12 years)

35.2 22.6 32.2 NS

38.9 33.7 27.4 o0.001

Men Manual workers

Low income (o200 000 NOK)

33.2 28.8 38.1 NS

35.5 32.8 31.7 NS

41.1 31.5 27.4 0.004

32.1 31.0 36.9 NS

Low education (p12 years)

32.9 32.6 34.5 NS

38.9 33.7 27.4 o0.001

Manual workers

28.6 33.8 37.6 0.003

42.4 31.1 26.5 o0.001

Abbreviation: NS, non-significant. a P-value for difference between low income or education group or manual workers and the others, from w2-tests.

European Journal of Clinical Nutrition


1154 Table 4 Being the highest tertile of the Western food pattern as a function of socioeconomic factors and work control Women

Education (ref.: p12 years) 13–16 years X17 years

Occupation (ref.: V–VII) I þ II III IV

Income (ref.: 0–200 000 NOK) 200 000–300 000 NOK 4300 000 NOK

Work control (ref.: never/seldom) Most often Always

Shift work (ref.: no)

Men

Model 1 OR (95 % CI)




0.726*** (0.616–0.857) 0.506*** (0.427–0.600)

0.753** (0.636–0.891) 0.537*** (0.449–0.643)

0.714*** (0.598–0.581) 0.475*** (0.396–0.569)

0.755** (0.625–0.911) 0.511*** (0.418–0.624)

0.564*** (0.422–0.754) 0.697* (0.530–0.916) 0.595** (0.408–0.867)

0.749 (0.549–1.022) 0.810 (0.612–1.071) 0.820 (0.551–1.221)

0.583*** (0.479–0.711) 0.649*** (0.525–0.802) 0.861 (0.661–1.120)

0.728** (0.577–0.918) 0.780* (0.622–0.977) 1.089 (0.816–1.452)

0.900 (0.770–1.052) 0.780** (0.653–0.933)

0.941 (0.802–1.105) 0.948 (0.780–1.151)

1.059 (0.824–1.362) 0.933 (0.735–1.186)

1.057 (0.817–1.366) 1.157 (0.899–1.488)

0.935 (0.812–1.077) 0.684** (0.552–0.849)

1.038 (0.892–1.208) 0.775* (0.612–0.981)

0.980 (0.818–1.175) 0.845 (0.679–1.051)

1.060 (0.871–1.289) 0.869 (0.683–1.107)

1.081 (0.915–1.278)

0.991 (0.829–1.184)

1.077 (0.894–1.298)

0.908 (0.740–1.113)

Abbreviations: CI, confidence interval; OR, odds ratio; ref., reference. Model 1: Each SEP- or work-related variable adjusted for age, region of origin and household composition. Model 2: All SEP- and working-condition-related variables adjusted for each other and for age, region of origin and household composition. *Po0.05, **Po0.010 and ***Po0.001 for difference from reference category for each variable.

increased with 29%, but was still significantly lower than among the manual workers, however, not so after additional adjustment for income. The likelihood of being in the highest tertile of the Western food pattern was lowest in the highest income group among women (model 1); however, after adjustment for education, the difference between groups was not significant. Women always having control over own working situation were least likely to be in the highest tertile of the Western food pattern, when adjusted for demographic and SEP variables.

Prudent food pattern The likelihood of being in the highest tertile of the prudent food pattern was higher among the higher educational groups among both genders when adjusted for demographic (Table 5, model 1), and other SEP and work-related variables (model 2). Women and men in the highest occupational group, and men in the next highest group, as well as all the self-employed were significantly more likely than the manual workers to be in the highest tertile of the prudent European Journal of Clinical Nutrition

food pattern (model 1). Among women, adjustment for education resulted in non-significant differences between the highest group and the manual workers. Among the self-employed, adjustment for education reduced the OR with 29%, income with additionally 2% and work variables 13%, but the OR was significantly higher, also in model 2. Among men, the adjustment for education resulted in nonsignificant difference between the routine non-manual and manual workers. In the highest group and the self-employed, further adjustment for the other SEP/working variables reduced the ORs with 33 and 28%, respectively, and the likelihood of being in the highest tertile was still higher than among the manual workers. In model 1, women in the highest income group, and men in the middle group, had higher odds of being in the highest tertile of the prudent pattern than those in the lowest group. Adjusted for education, the difference was no longer significant for women, and OR was reduced with 2% among men. After additional adjustment for occupation, the difference between the groups was no longer significant. Women who most often or always had control over own work had higher OR than


1155 Table 5 Being the highest tertile of the prudent food pattern as a function of socioeconomic factors and work control Women

Education (ref.: p12 years) 13–16 years X17 years

Occupation (ref.: V–VII) I þ II III IV

Income (ref.: 0–200 000 NOK) 200 000–300 000 NOK 4300 000 NOK

Work control (ref.: never/seldom) Most often Always

Shift work (ref.: no)

Men





1.916*** (1.619–2.267) 2.793*** (2.354–3.314)

1.840*** (1.547–2.187) 2.495*** (2.081–2.991)

1.517*** (1.267–1.817) 2.530*** (2.121–3.019)

1.370** (1.130–1.663) 2.227*** (1.828–2.713)

1.991*** (1.475–2.687) 1.311 (0.985–1.745) 2.692*** (1.868–3.879)

1.219 (0.883–1.682) 1.027 (0.765–1.380) 1.613* (1.094–2.377)

1.992*** (1.622–2.445) 1.539*** (1.232–1.922) 2.346*** (1.807–3.047)

1.327* (1.041–1.691) 1.158 (0.913–1.469) 1.678*** (1.259–2.237)

0.947 (0.812–1.106) 1.448*** (1.219–1.721)

0.881 (0.750–1.035) 1.075 (0.889–1.299)

0.759* (0.593–0.971) 1.052 (0.835–1.324)

0.841 (0.652–1.083) 0.880 (0.689–1.125)

1.266** (1.098–1.461) 1.854*** (1.521–2.259)

1.070 (0.917–1.248) 1.387** (1.112–1.730)

1.222* (1.015–1.471) 1.439** (1.162–1.783)

0.986 (0.807–1.205) 1.059 (0.835–1.342)

0.837* (0.704–0.995)

0.984 (0.817–1.185)

0.680 (0.557–0.831)

0.854 (0.687–1.063)

Abbreviations: CI, confidence interval; OR, odds ratio; ref., reference. Model 1: Each SEP- or work-related variable adjusted for age, region of origin and household composition. Model 2: All SEP- and working-condition-related variables adjusted for each other and for age, region of origin and household composition. *Po0.05, **Po0.010 and ***Po0.001 for difference from reference category for each variable.

others of being in the highest tertile of the prudent pattern (model 1). Adjustment for education made the difference between ‘most often’ and ‘never/seldom’ no longer significant. The OR for the category ‘always’ was reduced by 12% adjusting for education and another 14% by adjusting for occupation, and was still significantly different from the reference group in model 2. Among men, those most often and always having control over work had significantly higher odds of being in the highest tertile of the prudent pattern (model 1). When adjusted for education, the difference was no longer significant for those most often having control, and the OR was reduced with 11% among those who always had control. Further adjustment for occupation made the difference non-significant.

Discussion The Oslo Health Study provides an opportunity to contribute to the knowledge about social disparities in diet, because of its large number of participants from different age groups

and the combination of FFQs and questions about SEP and working conditions. However, the attendance rate at 44.5% may cause some challenges. An analysis of the nonattendants found a somewhat higher attendance rate among females (OR 1.32) and persons with higher age (OR 2.20 for 59–60 years compared with 30 years), education (OR 1.46 for education from college or university compared with p9 years) and annual income (OR 1.52 for X400 000 NOK compared with o100 000 NOK), but the results were concluded to be robust (Sogaard et al., 2004). Furthermore, as the focus of this study is on associations, and not prevalence, this is not likely to have weakened the results substantially.

Dietary factors The exploratory a posteriori approach involved in factor analysis implies that the results reflect the actual situation, rather than any optimal food patterns. However, the ‘Western pattern’ and the ‘prudent pattern’ are similar to the ‘Western pattern’ and the ‘prudent pattern’ found in other European Journal of Clinical Nutrition


1156 studies (Fung et al., 2001; Osler et al., 2001; Heidemann et al., 2008). The Western food pattern has been characterized by red meat, processed meat, pizza, French fries and sweets, and the prudent food pattern by high loadings for vegetables, fruit, legumes, whole grain, fish and poultry. These patterns are found to be associated with chronic diseases (Slattery et al., 1998; Hu et al., 2000; Fung et al., 2001; Fung et al., 2004; Heidemann et al., 2008); the ‘Western pattern’ being positively associated with coronary heart disease (CHD), diabetes and colon cancer, whereas the ‘prudent’ pattern is inversely associated with CHD and colon cancer.

SEP and work control Three indicators of SEP were used in this study. Years of education and current occupation are likely to be accurately reported. In the case of immigrants, number of years of education may not represent the same as in the general population, and may be differently related to current occupation. However, this would have affected only a small number of participants. Self-reported personal income is more vulnerable to bias, and its effect is likely to be modulated by household composition and household income. Thus, we adjusted for household composition in the multivariate analyses. Possible inaccuracy in income, as well as the lower attendance rate among low SEP groups, may have led to an underestimation of socioeconomic differences in dietary habits. All three indicators of SEP were related to the Western and the prudent food pattern, either crude or adjusted. This confirms previous research (Johansson et al., 1999; Irala-Estevez et al., 2000; Mishra et al., 2002; Perrin et al., 2002; Engeset et al., 2005; Lallukka et al., 2007), indicating that people in higher educational and occupational groups have an overall healthier diet, with education being the strongest predictor of a healthy diet. That there was no significant association between the Western pattern and education in crude analyses is likely to be due to age- and gender variations in diet (Hjartaker and Lund, 1998; Wandel and Fagerli, 1999). The effect of occupation was partly explained by educational level, however, more strongly among women than men. In the full multivariate model among women, only the self-employed had higher odds than manual workers of being in the high tertile of the prudent food pattern. Thus, occupation seemed to be more important for eating habits among men than women. The self-employed do not fit into an occupational hierarchy, but may rather represent more freedom in the work situation. Having control over own work situation increases the opportunity to choose place and time to eat and thereby also the types of food. A perception of having control over work may spill over to other areas of life, giving people a general feeling of more control and opportunities to make healthy food choices. A prudent food pattern, characterized by higher intakes of vegetables, fish and fruit, can be more time consuming to prepare than other meals, European Journal of Clinical Nutrition

making a more flexible time schedule a predictor of opportunity for frequent consumption of these items. This spill over effect from work to private life has been studied qualitatively in the United States (Devine et al., 2003; Devine et al., 2006). The investigators found that many employees, both women and men, felt that lack of time was a barrier to provide adequate and healthy food for themselves and their families because of long or inconvenient working hours and inflexible schedules. In this study, work control was especially important for dietary habits among women. This supports previous findings of a study carried out in Finland of a significant association between high work control and a healthy diet among women but not men (Lallukka et al., 2004). Stallone et al. (1997) analysed data from the Whitehall II Study and found that low energy reporting (here o1.2 times calculated BMR) was associated with lower employment grade, and the proportions underreporting ranged from almost 20% among those with higher employment status to close to 50% among those with low employment status. Underreporting of energy most probably concerns foods rich in fat, which would correspond to the Western food pattern in this study. If the proportion of low energy reporting follows the same socioeconomic patterns in our data as in the Whitehall II study, the true proportions with high intake of a Western food pattern are likely to be higher, especially in lower SEP groups, making the social inequalities found in diet even wider.

Conclusion Four distinct food patterns were discerned in this study among adult, working women and men in Oslo, and these were called Western, prudent, traditional and sweet. The three SEP indicators were in different ways related to the food patterns, but the effect of occupation and income was partly explained by years of education, especially among women. Occupation seemed to be more important for eating habits among men than women. Women with high degree of work control were less likely to have Western, and more likely to have a prudent food pattern, than those who had low control. This study highlights the importance of the work situation for eating habits, especially among women.

Conflict of interest The authors declare no conflict of interest.

Acknowledgements The data collection was conducted as part of the Oslo Health Study carried out in 2000–2001 as a collaboration between the Norwegian Institute of Public Health, the Oslo City Council and the University of Oslo. No external funding


1157 was used. All authors contributed in the conception and the writing of the article. Statistical analyses were executed by MKRK.

References Arkkola T, Uusitalo U, Kronberg-Kippila C, Mannisto S, Virtanen M, Kenward MG et al. (2008). Seven distinct dietary patterns identified among pregnant Finnish women—associations with nutrient intake and sociodemographic factors. Public Health Nutr 11, 176–182. Avendano M, Kunst AE, Huisman M, Lenthe FV, Bopp M, Regidor E et al. (2006). Socioeconomic status and ischaemic heart disease mortality in 10 western European populations during the 1990s. Heart 92, 461–467. Devine CM, Connors MM, Sobal J, Bisogni CA (2003). Sandwiching it in: spillover of work onto food choices and family roles in lowand moderate-income urban households. Soc Sci Med 56, 617–630. Devine CM, Jastran M, Jabs J, Wethington E, Farell TJ, Bisogni CA (2006). ‘A lot of sacrifices:’work-family spillover and the food choice coping strategies of low-wage employed parents. Soc Sci Med 63, 2591–2603. Engeset D, Alsaker E, Ciampi A, Lund E (2005). Dietary patterns and lifestyle factors in the Norwegian EPIC cohort: the Norwegian Women and Cancer (NOWAC) study. Eur J Clin Nutr 59, 675–684. Erikson R (1992). The Constant Flux: A Study of Class Mobility in Industrial Societies. Clarendon Press: Oxford. Fung TT, Schulze M, Manson JE, Willett WC, Hu FB (2004). Dietary patterns, meat intake, and the risk of type 2 diabetes in women. Arch Intern Med 164, 2235–2240. Fung TT, Willett WC, Stampfer MJ, Manson JE, Hu FB (2001). Dietary patterns and the risk of coronary heart disease in women. Arch Intern Med 161, 1857–1862. Galobardes B, Lynch J, Smith GD (2007). Measuring socioeconomic position in health research. Br Med Bull 81-82, 21–37. Giskes K, Turrell G, Van Lenthe FJ, Brug J, Mackenbach JP (2006). A multilevel study of socio-economic inequalities in food choice behaviour and dietary intake among the Dutch population: the GLOBE study. Public Health Nut 9, 75–83. Heidemann C, Schulze MB, Franco OH, van Dam RM, Mantzoros CS, Hu FB (2008). Dietary patterns and risk of mortality from cardiovascular disease, cancer, and all causes in a prospective cohort of women. Circulation 118, 230–237. Hjartaker A, Lund E (1998). Relationship between dietary habits, age, lifestyle, and socio-economic status among adult Norwegian women. The Norwegian Women and Cancer Study. Eur J Clin Nutr 52, 565–572. Hu FB, Rimm EB, Stampfer MJ, Ascherio A, Spiegelman D, Willett WC (2000). Prospective study of major dietary patterns and risk of coronary heart disease in men. Am J Clin Nutr 72, 912–921. Irala-Estevez JD, Groth M, Johansson L, Oltersdorf U, Prattala R, Martinez-Gonzalez MA (2000). A systematic review of socioeconomic differences in food habits in Europe: consumption of fruit and vegetables. Eur J Clin Nutr 54, 706–714. Johansson L, Thelle DS, Solvoll K, Bjorneboe GE, Drevon CA (1999). Healthy dietary habits in relation to social determinants and lifestyle factors. Br J Nutr 81, 211–220. Kunst AE, Groenhof F, Andersen O, Borgan JK, Costa G, Desplanques G et al. (1999). Occupational class and ischemic heart disease mortality in the United States and 11 European countries. American J Public Health 89, 47–53. Lallukka T, Laaksonen M, Rahkonen O, Roos E, Lahelma E (2007). Multiple socio-economic circumstances and healthy food habits. Eur J Clin Nutr 61, 701–710. Lallukka T, Sarlio-Lahteenkorva S, Roos E, Laaksonen M, Rahkonen O, Lahelma E (2004). Working conditions and health behaviours

among employed women and men: the Helsinki Health Study. Prev Med 38, 48–56. Mackenbach JP, Bos V, Andersen O, Cardano M, Costa G, Harding S et al. (2003). Widening socioeconomic inequalities in mortality in six Western European countries. Int J Epidemiol 32, 830–837. Mackenbach JP, Kunst AE, Cavelaars AE, Groenhof F, Geurts JJ (1997). Socioeconomic inequalities in morbidity and mortality in western Europe. The EU Working Group on Socioeconomic Inequalities in Health. Lancet 349, 1655–1659. Metcalf P, Scragg R, Davis P (2006). Dietary intakes by different markers of socioeconomic status: results of a New Zealand workforce survey. N Z Med J 119, U2127. Mishra G, Ball K, Arbuckle J, Crawford D (2002). Dietary patterns of Australian adults and their association with socioeconomic status: results from the 1995 National Nutrition Survey. Eur J Clin Nutr 56, 687–693. Mosdøl A (2004). Dietary assessment—the weakest link?: a dissertation exploring the limitations to questionnaire based methods of dietary assessment. Doctoral Thesis, Faculty of Medicine, University of Oslo. Osler M, Heitmann BL, Gerdes LU, Jorgensen LM, Schroll M (2001). Dietary patterns and mortality in Danish men and women: a prospective observational study. Br J Nutr 85, 219–225. Perrin AE, Simon C, Hedelin G, Arveiler D, Schaffer P, Schlienger JL (2002). Ten-year trends of dietary intake in a middle-aged French population: relationship with educational level. Eur J Clin Nutr 56, 393–401. Petkeviciene J, Klumbiene J, Prattala R, Paalanen L, Pudule I, Kasmel A (2007). Educational variations in the consumption of foods containing fat in Finland and the Baltic countries. Public Health Nutr 10, 518–523. Raulio S, Roos E, Mukala K, Prattala R (2008). Can working conditions explain differences in eating patterns during working hours? Public Health Nutr 11, 258–270. Robinson SM, Crozier SR, Borland SE, Hammond J, Barker DJ, Inskip HM (2004). Impact of educational attainment on the quality of young women’s diets. Eur J Clin Nutr 58, 1174–1180. Rognerud MA, Zahl PH (2006). Social inequalities in mortality: changes in the relative importance of income, education and household size over a 27-year period. Eur J Public Health 16, 62–68. Roos E, Lahelma E, Saastamoinen P, Elstad JI (2005). The association of employment status and family status with health among women and men in four Nordic countries. Scand J Public Health 33, 250–260. Silventoinen K, Lahelma E (2002). Health inequalities by education and age in four Nordic countries, 1986 and 1994. J Epidemiol Community Health 56, 253–258. Slattery ML, Boucher KM, Caan BJ, Potter JD, Ma KN (1998). Eating patterns and risk of colon cancer. Am J Epidemiol 148, 4–16. Sogaard AJ, Selmer R, Bjertness E, Thelle D (2004). The Oslo Health Study: the impact of self-selection in a large, population-based survey. Int J Equity Health 3, 3. Stallone DD, Brunner EJ, Bingham SA, Marmot MG (1997). Dietary assessment in Whitehall II: the influence of reporting bias on apparent socioeconomic variation in nutrient intakes. Eur J Clin Nutr 51, 815–825. Strand BH, Tverdal A (2004). Can cardiovascular risk factors and lifestyle explain the educational inequalities in mortality from ischaemic heart disease and from other heart diseases? 26 year follow up of 50 000 Norwegian men and women. J Epidemiol Community Health 58, 705–709. The World Bank (1993). Investing in health. World Development Report. Oxford: Oxford University Press. Wandel M, Fagerli RA (1999). Norwegians’ opinions of a healthy diet in different stages of life. J Nutr Educ 31, 339. Wickrama K, Conger RD, Lorenz FO (1995). Work, marriage, lifestyle, and changes in men’s physical health. J Behav Med 18, 97–111. Wilkinson RG, Marmot MG (2006). Social Determinants of Health. Oxford University Press: Oxford.

European Journal of Clinical Nutrition