© International Epidemiological Association 2001
International Journal of Epidemiology 2001;30:334–340
Printed in Great Britain
Diet and socioeconomic position: does the use
of different indicators matter?
Bruna Galobardes, Alfredo Morabia and Martine S Bernstein
Objectives
To describe the association of diet and socioeconomic position and to assess
whether two different indicators, education and occupation, independently contribute in determining diet.
Methods
A community-based random sample of men and women residents of Geneva
canton, aged 35 to 74, participated in a survey of cardiovascular risk factors conducted annually since 1993. Lifetime occupational and educational history and a
semi-quantitative food frequency questionnaire were obtained from 2929 men
and 2767 women.
Results
Subjects from lower education and/or occupation consumed less fish and vegetables
but more fried foods, pasta and potatoes, table sugar and beer. Iron, calcium,
vitamin A and vitamin D intake were lower in the lower educational and occupational groups. Both indicators significantly contributed to determining a less
healthy dietary pattern for those from low social class. The effects of education
and occupation on dietary habits were usually additive and synergistic for some
food groups.
Conclusion
Assessing both education and occupation, improves the description of social class
inequalities in dietary habits, as they act, most of the time, as independent factors.
Keywords
Socioeconomic position, social class, education, occupation, diet, food intake,
nutrient intake
Accepted
21 August 2000
It is well established that there are socioeconomic inequalities in
health.1,2 Some of these inequalities are mediated by different
exposures to risk factors such as poor diet. Indeed, in economically developed countries, most3–7 but not all8,9 studies have reported healthier diet among subjects with higher socioeconomic
status. However, differences in the amount of food or in nutrient
intake among social classes are often small and can hardly explain the major inequalities observed in morbidity and mortality
in these countries.
The measures of socioeconomic position classify individuals
in groups of similar status or prestige, power, knowledge and
resources.10 Often, educational, occupational and income level
are used to characterize socioeconomic groups. The availability
of these indicators is sometimes limited to those routinely collected by national statistics. For historical and cultural reasons,
Europe traditionally uses education and occupation (information
on occupation is available in Britain from death certificates)
while the US mainly relies on income and education (since
1991, years of educational attainment is recorded in the death
certificate of almost all states).11,12
Division of Clinical Epidemiology, University Hospital of Geneva, 24 rue
Micheli du Crest, 1211 Geneva 14, Switzerland. E-mail: bruna.galobardes@
hcuge.ch
The correlation between these three indices as measures
of socioeconomic position in developed countries is relatively
weak (0.3–0.6).10,13–15 These results suggest that each index
explains a different component of social class variability which
contributes differently to health inequalities. In addition, the
interpretation of a given indicator might differ among subgroups of the population, such as women, older people or
different ethnicities.16
The objective of this study is to describe the association of diet
with socioeconomic position in a large population-based survey,
and to assess whether the results differ depending on which
indicator, education or occupation, is used.
Materials and Methods
The Bus Santé 2000 Survey is an ongoing, community-based
survey of cardiovascular risk factors conducted annually since
1993.17 Geneva (city and surroundings) has a population of
395 609 distributed over 242 km2 of land.18 Data reported
here comprise subjects randomly selected throughout 1993 to
1998, to represent the 89 000 men and 98 000 women noninstitutionalized residents aged 35–74 years.
Subjects were randomly identified from the residents’ register
published each year.18 Random sampling in age-sex-specific
334
INDICATORS OF DIET AND SOCIOECONOMIC POSITION
strata was proportional to the corresponding frequencies in
the population. In the first letter mailed to a potential subject,
the selected individual was asked to indicate a convenient day
and time to visit a mobile unit. In the case of non-response, up
to seven attempts were made to reach the person by phone, at
different times of the day and various days of the week, including Saturday and Sunday. Two more mailings were sent when
a selected individual could not be reached by phone. A person
who could not be contacted after three mailings and seven phone
calls was replaced using the same selection protocol. A systematic
check in the following year of the population register showed
that over 90% of ‘unreachable’ subjects no longer resided in
Geneva. Subjects who were contacted but refused to participate
were not replaced. Overall participation rate was 63%.
Participants completed a self-administered, semi-quantitative
food frequency questionnaire (FFQ) at home. This FFQ has
been developed in the target population. Mean values obtained
from a 24-hour recall diary were very similar between the
two questionnaires.19 It asked about 100 food items and their
serving sizes, and could be converted into daily energy, nutrient
and alcohol intakes.19 On the day of the visit to the mobile unit,
participants brought back the completed FFQ as well as a selfadministered questionnaire covering lifestyle factors, education,
reproductive history and classic cardiovascular disease risk
factors. Occupational history consisted of current and the two
longest past occupations, their duration and workplace characteristics. Trained interviewers checked these questionnaires for
completion.
For this analysis we excluded subjects with missing information on education or occupation (18 men and 88 women).
Subjects who reported having never worked for pay were also
excluded (2 men and 58 women). Women were eight times
more often excluded than men. A total of 2929 men and 2767
women were finally included in the analysis.
335
Germany and England, and finally, an amalgam of different
countries was grouped as ‘Other countries’.
Statistical analyses
Food consumption and nutrient intake were log transformed in
order to improve the normality of their distribution. Results were
back-transformed and reported as geometric means. Adjusted
mean of food consumption and nutrient intake was estimated
using multiple linear regression. The effect of education was adjusted for occupation and the effect of occupation for education.
All models were additionally adjusted for age, country of birth
and total energy intake. P-values for trend, obtained from linear
regression with education and occupation categorized as scores,
tested the null hypothesis that the slope equals 0. P-values
of each individual level of education and occupation were
obtained coding education and occupation as dummy variables.
Fitting a product term of the two variables categorized as score
assessed interaction between education and occupation. For those
food groups and nutrient intake where education and occupation had a significant interaction the results are presented
in Figures stratifying for each level of education within each
occupational level.
Results
Table 1 shows descriptive characteristics of the participants.
Almost half the men were from the high occupational group
Table 1 Descriptive characteristics of the study participants. Geneva,
Switzerland 1993–1998
Total (n)
Men
%
Women
%
2929
2767
Age (years)
Variable definition
35–44
31.6
34.5
According to the type and level of schooling, education was
categorized as: low (<8 years of schooling), medium (9–12 years
of schooling) and high (>13 years and including people who
obtained the Swiss baccalaureate).
Occupational level was measured using the respondent’s own
occupation: current occupation at the time of the survey or the
longest occupation ever held for those not currently working.
We grouped them in three occupational levels based on the
British Registrar General’s Scale:20 high (I and II from the original
British classification : professional and intermediate professions),
medium (III-N: non-manual occupations) and low (III-M, IV
and V: manual or lower occupations).
Groups of food items from the FFQ assessed were: regular dairy
products (excluding cheese), low-fat dairy products (excluding
cheese), cheese, bread and cereals, pasta and potatoes,
vegetables, meat (including lamb, beef, chicken and pork), fish,
fruits, pastries and desserts, table sugar, fat and oil for cooking
or added into the dishes, fried food (included French fries and
fried fish), juices, wine and champagne, beer and hard alcohol.
The food items and their serving sizes were converted into daily
energy, nutrient and alcohol intakes. Micronutrients included
iron, calcium, vitamin D and vitamin A.
Country of birth was classified as Switzerland, a group comprising Spain, Italy and Portugal, another group with France,
45–54
32.6
32.8
55–64
22.2
21.0
>65
13.6
11.7
Country of birth
Switzerland
57.0
55.8
France, Germany and England
13.0
16.6
Spain, Italy and Portugal
16.3
13.4
Other countriesa
13.7
14.2
High
35.5
34.9
Medium
56.4
57.1
8.1
8.0
High
48.7
30.9
Medium
23.7
53.8
Low
27.6
15.3
Educationb
Low
Occupationc
a ‘Other countries’ included an amalgam of different countries.
b Education: high (>13 years of schooling, including people having obtained
the Swiss baccalaureate), medium (9–12 years of schooling) and, low
(<8 years of schooling).
c Current occupation or longest occupation ever held for those not currently
working: high (I and II from the British Registrar Classification: professional
and intermediate professions), medium (III-N: non-manual occupations)
and low (III-M, IV and V: manual or lower occupations).
336
INTERNATIONAL JOURNAL OF EPIDEMIOLOGY
Table 2 Adjusted food intake (geometric means in g/week unless specified) in the group with high education or with high occupation, and the
difference in food intake between the medium and the low versus the high groups, by gender. Geneva, Switzerland 1993–1998
Educationa
Occupationb
High
Medium
minus high
Low
minus high
trend Pc
High
Medium
minus high
Low
minus high
trend Pc
Fish
156.1
–16.0**
–35.6***
Vegetables
849.0
–57.8**
–75.3*
0.0002
151.3
–15.7**
–22.8***
0.0002
0.01
842.4
–48.5*
–65.4**
Meat
569.8
5.9
0.003
26.4
0.4
592.4
–20.6
–15.1
Fried foods
128.8
9.8*
0.3
18.7*
0.02
131.1
5.7
15.8**
0.01
Fruits
848.8
–37.1
82.6
Beer (cc/week)
618.3
164.7***
174.3*
0.7
891.4
–34.4
–51.5
0.2
0.001
632.7
63.5
238.4***
,0.0001
Fish
178.3
–26.8***
–8.4
0.01
177.1
–10.9
–21.2*
0.03
Vegetables
991.6
–52.3**
–109.6**
0.01
993.4
–74.8**
–93.2**
0.003
Meat
412.5
–0.8
–9.0
0.7
385.5
26.0*
46.3**
0.01
96.7
9.0*
16.6*
0.01
95.9
5.4
23.4***
0.001
1183.0
–75.5
–41.5
0.3
1182.5
–73.4
–42.2
0.3
305.0
–15.6
–21.4
0.6
292.3
–2.6
3.3
0.9
Men
Women
Fried foods
Fruits
Beer (cc/week)
a Adjusted for age, occupation, country of birth and total energy intake.
b Adjusted for age, education, country of birth and total energy intake.
c P-values for trends were obtained from linear regression with education and occupation categorized as a score.
* P < 0.05; ** P < 0.01; *** P < 0.001: P-values associated with each individual level of education and occupation were obtained coding education and
occupation as dummy variables.
while more than half of women were from the medium
occupational group. Almost 28% of men versus 15% of women
were classified in the low occupational group.
Table 2 shows the absolute food intake in men and women
with high education or high occupation and the difference
in intake with the medium and low groups. In men, lower
education and lower occupation was associated with a lower
consumption of fish and vegetables but a higher consumption
of fried foods and beer. There were no differences in meat
and fruits consumption. In women, lower education and lower
occupation were associated with lower consumption of fish and
vegetables and higher consumption of fried foods. Women with
lower occupation consumed more meat (trend P = 0.01). No
differences by educational or occupational level were found for
fruit and beer consumption among women.
Figure 1 shows the consumption of pasta and potatoes in men
and women for each level of education and occupation (P-value
for interaction = 0.001). Only among men of medium and low
occupation was decreasing education associated with greater
pasta and potatoes consumption. Women consumed, at each
level, smaller amounts of pasta and potatoes than men. None of
the tests for trend were statistically significant in women.
Table sugar consumption increased with decreasing education
among men of high (trend P = 0.002) and middle class (trend
P = 0.01) occupations while no association was found among
those in low occupation (Figure 2). Similar results, although not
statistically significant, were observed for women.
Consumption of regular and low-fat dairy products, cheese,
bread and cereals, juices, coffee and tea and alcoholic beverages
(except for wine consumption in men which increased with
decreasing education) did not differ among different educational
or occupational groups (results not shown).
There were no differences in nutrient intake by educational
or occupational level with few exceptions (Table 3). Fibre intake
decreased with decreasing level of occupation in men and
women (men: trend P = 0.03; women: trend P = 0.01). Total
protein intake decreased with decreasing education in women
(trend P = 0.03). Saturated fat was lowest in the low educational groups in both, men and women and, finally, monounsaturated fat decreased with decreasing occupation in
women (trend P = 0.01).
Calcium intake decreased with decreasing occupational level
in men (trend P = 0.004) and with decreasing education in
women (trend P = 0.04) (Table 4). Decreasing educational and
occupational level, in men and women, was strongly associated
with decreasing vitamin D intake. Vitamin A intake decreased
with decreasing education only among men (trend P = 0.001).
Figure 1 Pasta and potatoes consumption by educational and
occupational level in men and women. Geneva, Switzerland,
1993–1998. Adjusted for age, country of birth and total energy intake.
INDICATORS OF DIET AND SOCIOECONOMIC POSITION
337
Finally, 6% of participants reported taking vitamin E supplements and less than 14% using multivitamin products (results
not shown).
Discussion
The present results show that lower education and lower
occupation independently contribute to determining differences
in dietary habits and that the effect of the two indicators is
cumulative. Men from lower socioeconomic position consumed
less fish and vegetables but more pasta and potatoes, fried foods,
table sugar and beer. Women of lower socioeconomic position
also consumed less fish and vegetables but more meat, fried
foods, table sugar, pasta and potatoes. Lower intake of iron,
calcium, vitamin A and vitamin D was present among lower
socioeconomic groups. All results were adjusted for total energy
intake and therefore the differences in food and nutrient intake
among social classes were independent of the actual amount
of food consumed. The differences in food intake due to education
and occupation, for most food groups, were simply additive and
sometimes more than additive.
Our results are consistent with previous reports on diet and
social class. Differences are often of small magnitude and for
some, but not all, key components of the diet.3–9,21–23 The
present results suggest that these reports may underestimate
true differences because, when using a single measure of social
class, they only assessed one component of the socioeconomic
position. Despite differences in food consumption, nutrient intake
was similar among socioeconomic groups, as these differences
Figure 2 Table sugar consumption by educational and occupational
level in men and women. Geneva, Switzerland, 1993–1998. Adjusted
for age, country of birth and total energy intake.
There was a significant interaction between education and
occupation with iron intake, in men (P-value for interaction =
0.01) but not in women, although absolute intake differences
were small (Figure 3). In high levels of occupation there was a
decrease in iron intake with decreasing educational levels. An
opposite trend, although not statistically significant, was found
for medium and low occupational levels. Among women a
similar pattern of decreasing iron intake with decreasing educational level, was found at every occupational level.
Table 3 Adjusted macronutrient intake (geometric means in g/day unless specified) in the group with high education or with high occupation,
and the difference in macronutrient intake between the medium and low versus the high groups, by gender. Geneva, Switzerland 1993–1998
Educationa
Occupationb
High
Medium
minus high
Low
minus high
trend Pc
High
Medium
minus high
Low
minus high
1998.3
–37.3
–54.7
0.2
1937.8
47.6
41.9
0.2
14.6
–0.3
0.6
0.7
15.0
–0.4
–0.5**
0.03
trend Pc
Men
Total energy (kcal/day)
Fibre
Carbohydrates
224.7
–0.5
3.8
0.5
223.8
2.8
3.3
0.1
Total protein
78.7
0.3
–0.4
0.9
79.3
–0.7
–1.3
0.1
Total fat
79.6
–0.1
–2.0
0.2
79.5
–0.8
–1.0
0.2
Saturated fat
30.4
0.2
–2.0***
0.1
30.0
–0.1
–0.5
0.4
Monounsaturated fat
30.1
–0.4
–0.5
0.3
30.1
–0.4
–0.5
0.1
Polyunsaturated fat
11.7
0.2
0.7*
0.02
12.0
–0.2
0.1
0.9
1622.4
–29.8
–24.9
0.4
1607.9
–26.1
14.9
0.9
14.3
–0.4**
–0.1
0.3
14.6
–0.7**
–0.7*
0.01
Women
Total energy (kcal/day)
Fibre
Carbohydrates
182.4
–0.3
5.8
0.2
183.3
0.6
2.1
0.5
Total protein
64.9
–1.0
–2.0
0.03
63.5
0.1
1.1
0.3
Total fat
63.9
0.6
–1.8
0.5
64.2
–0.6
–1.5
0.1
Saturated fat
22.7
0.5
–1.1*
0.5
22.5
0.1
–0.2
0.8
Monounsaturated fat
24.8
0.0
–0.8
0.4
25.1
–0.6*
–1.1**
0.01
9.8
0.2
0.2
0.2
9.8
0.0
0.2
0.5
Polyunsaturated fat
a Adjusted for age, occupation, country of birth and total energy intake.
b Adjusted for age, education, country of birth and total energy intake.
c P-values for trends were obtained from linear regression with education and occupation categorized as a score.
* P < 0.05; ** P < 0.01; *** P < 0.001: P-values associated with each individual level of education and occupation were obtained coding education and
occupation as dummy variables.
338
INTERNATIONAL JOURNAL OF EPIDEMIOLOGY
Table 4 Adjusted nutrient intake (geometric means in mg/day unless specified) in the group with high education or with high occupation, and
the difference in nutrient intake between the medium and the low versus the high groups, by gender. Geneva, Switzerland 1993–1998
Educationa
High
Occupationb
Medium
minus high
Low
minus high
trend Pc
High
Medium
minus high
Low
minus high
trend Pc
Men
Calcium
1021.5
25.8
–84.3**
0.2
1033.0
–32.2
–62.9**
0.004
Vitamin D (µg/day)
2.6
–0.3***
–0.8***
,0.0001
2.3
–0.1
–0.3***
,0.0001
Vitamin A
2.6
–0.2***
–0.2**
0.001
2.5
–0.1
0.0
0.8
Women
Calcium
920.0
–5.2
–79.9**
0.04
905.3
–22.2
–21.1
0.3
Vitamin D (µg/day)
2.3
–0.2***
–0.4***
,0.0001
2.3
–0.2**
–0.4***
,0.0001
Vitamin A
2.6
–0.1
–0.1
0.1
2.5
0.0
–0.1
0.5
a Adjusted for age, occupation, country of birth and total energy intake.
b Adjusted for age, education, country of birth and total energy intake.
c P-values for trends were obtained from linear regression with education and occupation categorized as a score.
* P < 0.05; ** P < 0.01; *** P < 0.001: P-values associated with each individual level of education and occupation were obtained coding education and
occupation as dummy variables.
Figure 3 Iron intake by educational and occupational level in men
and women. Geneva, Switzerland, 1993–1998. Adjusted for age,
country of birth and total energy intake.
may not be substantial enough to translate into differences in
nutrient intake. Vitamin and supplement intake was uncommon
in this population compared to the US where about 50% of the
population report multivitamin use.24
Some studies use a battery of different indicators in order to
describe inequalities and for most situations the results are fairly
similar. An occupation-based measure of social class remained
significant in predicting diet after adjusting for education and
house tenure.4 Race, education, and income were significantly
associated with some nutrient intake among 9 and 10 year old
girls.25 Education and income remained independently associated
with diet in a population of Ontario seniors.26 Other studies
have used different indicators of social class but they did not
report whether they had an independent effect.21,23,27 Finally,
other studies have used composite indices, which preclude
differentiating the independent effect of each component of the
index.5,28
We measured the effect of education adjusting for occupation
and the effect of occupation adjusting for education. Both indicators were consistently associated with similar dietary patterns.
Clearly, both indices measure aspects of the same concept, socioeconomic position. Indeed, the educational level determines
the occupation and jointly with occupation they determine the
income level. On the other hand, there are plausible pathways
through which education and occupation could have an
independent role in predicting health-related behaviours. The
amount of education and knowledge individuals acquire can
influence their lifestyle, problem-solving capacity and values,10,14
the importance given to preventive health measures and the
capacity to generate behaviours that will bring benefits on a
long term basis.29 The occupational level is a measure of social
prestige. Occupation is related to differential exposure to
environmental risk factors and to psychological stress.30,31 It
determines income and therefore, access to certain food products.
At the same time, it generates social networks that can greatly
influence behavioural health habits. Thus, both indicators
measure different pathways through which socioeconomic
position can have an independent effect on diet. It is reasonable
to conceive that, for example, poor dietary habits acquired in
youth can be added to poor dietary choices in the restaurant of
an industrial complex where healthy diet may not be promoted.
On the contrary, someone with high education may have broader
knowledge about diet and health and will probably choose
healthier meals at restaurants with colleagues who might also
be more predisposed towards healthier habits.
Education and occupation were similarly associated with diet
in men and in women although the significance and meaning
of these indicators may vary by gender.12,16 Occupational level
was obtained measuring the participant’s own current or longest
occupation held if not currently working. Whether women are
better categorized by their own or by their husband/partner’s
occupation has been long debated. Household socioeconomic
class was a better predictor of health than the woman’s own
position.32 As more women enter the paid work force similar
results may apply to men and household position might be
more relevant than individual measures, independent of gender.
INDICATORS OF DIET AND SOCIOECONOMIC POSITION
Another important issue is the limitation that occupation-based
classifications may have in measuring women’s socioeconomic
position. Most of the occupational scales were created using men’s
occupational categories as reference. Whether these scales
equally apply to women has not yet been demonstrated.
The differences we found among socioeconomic groups were
often small, thus bringing into question their clinical relevance.
Nevertheless, it is important to describe and follow behavioural
differences over time33,34 as inequalities are likely to increase.1
Reporting bias among higher social classes could explain some
of our results if people from this socioeconomic group tended to
report food intakes closer to dietary guidelines without reflecting their true consumption. In this instance we would expect
to find high consumption of fruits among men and women of
high socioeconomic status, along with the observed higher
consumption of fish and vegetables, but this is not the case.
Participation in the study was similar in all age groups and by
gender with the exception of women older than 65 who had
higher non-response rate. Information on smoking status was
obtained from all selected subjects and there were no differences between participants and non-participants. Bias due to
non-participants cannot be excluded based on this information,
but at least non-respondents did not differ on a variable known
to influence diet.35
The main strengths of our study were: (1) a food frequency
questionnaire developed and tested in the target population,
339
and (2) a detailed occupational history with current and past
occupations. A limitation was that other measures of the socioeconomic position such as income, wealth or supra-individual
indices, were not available.
In summary, we found an association between education and
occupation and diet. Both indices contributed to explaining a
dietary pattern that tended to be worse for the low social class
groups. The effects of education and occupation for some foods
and nutrients were additive or even synergistic. Our results
suggest that both indicators should be assessed in order to
provide a full description of the social inequalities in dietary
habits.
Acknowledgements
This study was funded by the Swiss National Fund for
Scientific Research (Grants No 32.31.326.91, 32–37986.93 and
32–49847.96). We are indebted to Moyses Szklo, Ann Sorenson
and Mike Constanza for their advice and comments.
Contributions: B Galobardes planned the study, analysed the
data and wrote the paper; A Morabia planned the study, supervised data analysis and participated writing the paper; MS
Bernstein coordinated and supervised the data collection and
participated in writing the paper.
KEY MESSAGES
• Diet differs across socioeconomic groups.
• The effects of low education and low occupation on food or nutrient intake are additive and sometimes synergistic.
• Education and occupation measure different aspects of the socioeconomic position. Several indicators are needed
to fully capture someone’s social status.
References
1 Smith GD, Bartley M, Blane D. The Black report on socioeconomic
inequalities in health 10 years on. BMJ 1990;301:373–77.
2 Marmot MG, Smith GD, Stansfeld S et al. Health inequalities among
British civil servants: the Whitehall II study. Lancet 1991;337:1387–93.
9 Windham CT, Wyse BW, Hansen RG, Hurst RL. Nutrient density of
diets in the USDA Nationwide Food Consumption Survey, 1977–1978:
I. Impact of socioeconomic status on dietary density. J Am Diet Assoc
1983;82:28–34.
10 Liberatos P, Link BG, Kelsey JL. The measurement of social class in
epidemiology. Epidemiol Rev 1988;10:87–121.
3 Pietinen P, Uusitalo U, Vartiainen E, Tuomilehto J. Dietary survey of
11 Fein O. The influence of social class on health status: American and
the FINMONICA project in 1982. Acta Med Scand Suppl 1988;728:
169–77.
12 Berkman LF, Macintyre S. The measurement of social class in health
4 Bolton-Smith C, Smith WC, Woodward M, Tunstall-Pedoe H.
Nutrient intakes of different social-class groups: results from the
Scottish Heart Health Study (SHHS). Br J Nutr 1991;65:321–35.
5 Hulshof KF, Lowik MR, Kok FJ et al. Diet and other life-style factors
in high and low socio-economic groups (Dutch Nutrition Surveillance
System). Eur J Clin Nutr 1991;45:441–50.
6 Shimakawa T, Sorlie P, Carpenter MA et al. Dietary intake patterns
and sociodemographic factors in the Atherosclerosis Risk in Communities Study. Prev Med 1994;23:769–80.
7 Diez-Roux AV, Nieto FJ, Caulfield L, Tyroler HA, Watson RL, Szklo M.
Neighbourhood differences in diet: the Atherosclerosis Risk in communities (ARIC) Study. J Epidemiol Community Health 1998;53:55–63.
8 Cronin FJ, Krebs-Smith SM, Wyse BW, Light L. Characterizing food
usage by demographic variables. J Am Diet Assoc 1982;81:661–73.
British research on health inequalities. J Gen Intern Med 1995;10:577–86.
studies: old measures and new formulations. IARC Sci Publ 1997,
pp.51–64.
13 Abramson JH, Gofin R, Habib J, Pridan H, Gofin J. Indicators of social
class. A comparative appraisal of measures for use in epidemiological
studies. Soc Sci Med 1982;16:1739–46.
14 Winkleby MA, Fortmann SP, Barrett DC. Social class disparities in
risk factors for disease: eight-year prevalence patterns by level of
education. Prev Med 1990;19:1–12.
15 Geronimus AT, Bound J. Use of census-based aggregate variables
to proxy for socioeconomic group: evidence from national samples.
Am J Epidemiol 1998;148:475–86.
16 Krieger N, Williams DR, Moss NE. Measuring social class in US public
health research: concepts, methodologies, and guidelines. Annu Rev
Public Health 1997;18:341–78.
340
INTERNATIONAL JOURNAL OF EPIDEMIOLOGY
17 Morabia A, Bernstein M, Heritier S, Ylli A. Community-based
26 Keller HH, Ostbye T, Bright-See E. Predictors of dietary intake in
surveillance of cardiovascular risk factors in Geneva: methods,
resulting distributions, and comparisons with other populations.
Prev Med 1997;26:311–19.
27 Smith AM, Baghurst KI. Public health implications of dietary
18 Office Cantonal de la Statistique. Annuaire statistique du Canton de
Genève, OCSTAT.
19 Morabia A, Bernstein M, Kumanyika S et al. Development and
validation of a semi-quantitative food questionnaire based on a
population survey. Soz Praventivmed 1994;39:345–69.
20 Leete R, Fox J. Registrar General’s social classes: origins and uses.
Popul Trends 1977;8:1–7.
21 Roos E, Prattala R, Lahelma E, Kleemola P, Pietinen P. Modern and
healthy?: socioeconomic differences in the quality of diet. Eur J Clin
Nutr 1996;50:753–60.
22 La Vecchia C, Negri E, Franceschi S, Parazzini F, Decarli A. Differences
in dietary intake with smoking, alcohol, and education. Nutr Cancer
1992;17:297–304.
23 Hjartaker A, Lund E. Relationship between dietary habits, age,
Ontario seniors. Can J Public Health 1997;88:305–09.
differences between social status and occupational category groups.
J Epidemiol Community Health 1992;46:409–16.
28 Popkin BM, Siega-Riz AM, Haines PS. A comparison of dietary trends
among racial and socioeconomic groups in the United States. N Engl J
Med 1996;335:716–20.
29 Fuchs VR. Economics, health, and post-industrial society. Milbank
Mem Fund Q Health Soc 1979;57:153–82.
30 Marmot MG, Bosma H, Hemingway H, Brunner E, Stansfeld S.
Contribution of job control and other risk factors to social variations
in coronary heart disease incidence. Lancet 1997;350:235–39.
31 Marmot M, Feeney A. General explanations for social inequalities in
health. IARC Sci Publ 1997;138:207–28.
32 Krieger N, Chen JT, Selby JV. Comparing individual-based and
household-based measures of social class to assess class inequalities in
women’s health: a methodological study of 684 US women. J Epidemiol
Community Health 1999;53:612–23.
lifestyle, and socio-economic status among adult Norwegian women.
The Norwegian Women and Cancer Study. Eur J Clin Nutr 1998;52:
565–72.
33 Morabia A. From disease surveillance to the surveillance of risk
24 Slesinski MJ, Subar AF, Kahle LL. Trends in use of vitamin and
34 Morabia A. Worldwide surveillance of risk factors to promote global
mineral supplements in the United States: the 1987 and 1992 National
Health Interview Surveys. J Am Diet Assoc 1995;95:921–23.
25 Crawford PB, Obarzanek E, Schreiber GB et al. The effects of race,
household income, and parental education on nutrient intakes of
9- and 10-year-old girls. NHLBI Growth and Health Study. Ann Epidemiol
1995;5:360–68.
factors. Am J Public Health 1996;86:625–27.
health. Am J Public Health 2000;90:22–24.
35 Morabia A, Curtin F, Bernstein MS. Effects of smoking and smoking
cessation on dietary habits of a Swiss urban population. Eur J Clin Nutr
1999;53:239–43.