You are on page 1of 5

nutrition survey

Level of education, level of income, and


level of fatness in adults1’ 2
Stanley M. Stephen M. Patricia E. and Ian T. T. Higgins4

ABSTRACT Socioeconomic status is systematically related to the level of fatness, and


therefore the incidence of obesity, in a total community survey of nearly 5,000 adults. Among
males with more than 12 years of schooling, the average thickness of 4 fatfolds is 10% greater,

Downloaded from www.ajcn.org by guest on January 24, 2011


amounting to about 2 kg of total fat, than those with 8 years or less of education. In females,
however, the opposite trend is observed, those in the higher educational group averaging 20%
thinner fatfolds, or about 5.5 kg total fat, than females in the lower educational group. These
findings confirm the need for standards of obesity that take socioeconomic status into ac-
count. Am. J. Clin. Nutr. 30: 72 1-725, 1977.

Although socioeconomic status is known for the most part the wives of the males.
to affect the level of fatness and therefore With a large sample size (N 5 ,000) de- -

the incidence of obesity, neither the nature rived from a total population sample , sev-
nor the direction of the relationship is clear. eral socioeconomic indices and four differ-
In some studies, the poor are leaner and the ent fatfolds, we now have definitive infor-
more affluent are more frequently obese mation on the relationship between various
( 1). In most reports, obesity is more com- socioeconomic measures and the level of
mon in those of lower socioeconomic status, fatness in an agricultural-industrial Amen-
and less frequent in the highest socioeco- can community.
nomic class (2-8). In our own analysis of
fatfold data from the Ten State Nutrition Methods and materials
Survey, fatness increases with income level
This study is based upon four fatfold measurements
in the adult male, yet decreases with in- and three socioeconomic measures on 4,936 adult par-
creasing income in the adult female (9-1 1). ticipants in the Tecumseh (Michigan) Project of the
Some of these discrepancies appear to University of Michigan School of Public Health (12).
arise from differing socioeconomic indices, All measurements were taken during examination
round 2, between 1962 and 1965.
different measures of occupation and educa-
The three socioeconomic measures included I ) edu-
tion, and different definitions of “obesity.” cational level of head of household, 2) educational
In other cases, there are obvious sampling level of the subject and 3) total household income at
problems, with different population seg- the time of the examination. Since educational level
ments involved. Most previous investiga-
I This study was supported by Grant HD 09538
tions have made use of selected groups, and
(Socioeconomic and Genetic Determinants of Obesity)
not natural or demographic populations,
from the National Institute of Health.
nor have they included the complete socio- 2 Address reprint requests to: Dr. Stanley M. Garn,
economic range. Center for Human Growth, 1111 E.
St., Catherine
The present study of adult fatness, there- Ann Arbor, 48109.
Michigan,

fore, is based upon a single community, with Center for Human Growth, University of Michi-
gan and the Nutrition Unit, School of Public Health,
representation from the entire socioeco- Ann Arbor, Michigan. Department of Epidemiol-
nomic range. Both sexes are represented in ogy, University of Michigan. School of Public Health,
nearly equal numbers, and the women are Ann Arbor, Michigan.

TheAmerican Journal of Clinical Nutrition 30: MAY 1977, pp.721-725. Printed in U.S.A. 721
722 GARN ET AL.

ranged from no schooling through postgraduate univer- For the 2,310 adult males, the level of
sity level, three educational groupings were used.
fatness increases with educational level such
These groupings comprised 1 ) 8 years or less (16.5 % of
households), 2) 9 to 12 years (60.5% of households)
that those with more than 1 2 years of educa-
and 3) more than 12 years (23% of households). In tion are fatter than those with 8 years of
comparable fashion, the income groupings were 1 ) be- education or less. Taking all four fatfolds
low $5,000 per year, 2) $5,000 to 10,000 and 3) into account (triceps, subscapular, abdomi-
$10,000 and above. Approximately 60% of the house-
nal, and iliac) and all three decade groups
holds were in the second ($5,000 to 10,000) income
category. (4th through 8th), this trend is highly signifi-
Of the four fatfolds, two are well known-the tn- cant by the k-square test. The x2 value is
ceps fatfold and the subscapular fatfold. The less-famil- 1 1 .26, with corrections for continuity.
iar fatfolds included the abdominal fatfold, 2 cm to the
Expressed in millimeters of the double
right of tile umbilicus, and the iliac fatfold, at the
midaxillary line just above the crest of the ilium. All
fatfold, Tecumseh males with college and/or
were taken with the Lange constant pressure calipers, professional education are approximately
as described by the Committee on Nutritional Anthro- 10% fatter than those with less than high-
pometry, N.R.C. (13), and the encoded data were school education Converted
. into estimated
carefully edited before initial data analysis. Fatfold-
weight of fat (FW) and percent fat (%F),
fatfold correlations averaged 0.79 on a decade-by-dec-
ade basis, with correlations as high as 0.91 for some using the method developed by Oarn and

Downloaded from www.ajcn.org by guest on January 24, 2011


fatfold-combinations, and fatfold-weight correlations Harper (14) and elaborated by Broek (15),
averaged 0.71 for males and 0.75 for females. males with a college education have approx-
Comparisons were decade-specific, to correct for
imately 2 kg greater fat weight than those
age trends. In order to eliminate changing educational
levels, the youngest age category was 25 to 35 . The
with 8 years of education or less and a com-
basic question was the relationship between socioeco- parably greater percentage of fat.
nomic level and level of fatness in adult males and For 2,626 adult females (similarly age-
females in a single community. grouped) fatness decreases with increasing
educational level. This is shown for all four
Findings fatfolds in Table 1 , and for five decade
groupings. Again the trend is highly signifi-
As shown in Table 1 , (partially abridged cant, holding for 19 out of 20 paired com-
to save space) the level of education and the panisons, with a stochastic x2 value of 14.46.
level of fatness are systematically related in Women with college and/or professiunal ed-
both sexes, but in diametrically opposite di- ucation are far leaner than those lacking the
rections for adult males and adult females. high school years.

TABLE I
Effect of educational level on fatfold thickness (mm)”

Subscapular Triceps Abdominal Iliac

Age group N
8 Yr or Over 12 8 Yr or Over 12 8 Yr or Over 12 8 Yr or Over 12
less yr less yr less yr less yr

Males
25-35 178 14.8 13.8 13.5 16.8 29.7 30.2 24.9 25.2
35-45 274 15.5 16.8 14.1 16.4 29.5 30.0 21.2 23.5
45-55 172 17.2 17.9 13.9 15.4 29.6 31.3 23.0 25.3
55-65 130 17.3 22.4 15.1 15.1 29.6 30.1 22.4 22.8
65-75 91 13.5 17.2 12.4 14.7 24.7 29.8 16.8 19.7
Mean difference (mm) +2.0 +1.9 +1.7 +1.8
Percent difference (%) +13 +14 +6 +8
Females
25-35 174 17.4 13.0 22.4 23.1 34.5 24.6 25.3 16.2
35-45 254 18.9 14.8 25.4 24.4 35.5 26.9 24.6 16.5
45-55 176 24.0 18.0 29.6 25.0 44.9 33.3 34.6 23.3
55-65 129 22.4 19.2 27.8 27.2 40.9 35.0 31.8 25.2
65-75 96 24.2 20.8 30.0 25.0 44.9 38.2 35.1 30.8
Mean difference (mm) -4.2 -2.1 -8.5 --7.9
Percent difference (%) -2() -8 -21 -26
(, Computer calculated using randomly generated fractional integers to allow decimal values of medians. b N
for the two educational levels shown.
LEVELS OF EDUCATION, INCOME. AND FATNESS IN ADULTS 723

Expressed as an average for all four fat- by education level of the husband, but there
folds, women in the third (highest) educa- is a somewhat greater fatness spread when
tional category average 6 mm leaner than the womans’ own educational level is used
those in the lowest category (8 years and (cf. solid lines in Fig. 1).
less). This is a 20% difference in the aver- The interaction between education and
aged thickness of outer fat. The FW is 5.5 income is best demonstrated by comparing
kg less in the most educated grouping, and fatness levels with each of the socioeco-
%F decreases from 32% to 26% with in- nomic variables held constant, in turn. This
creasing educational level. design includes all three levels of income
Much the same trend appears when and all three levels of education first holding
household income is employed as the socio-
economic indicator, using all four fatfolds
and all five decade groupings (Table 2). As
shown, the higher income males are fatter
E
for the triceps, subscapular, abdominal, and E
iliac fatfolds. Yet the higher income females
-J
are leaner (at every site measured) than

Downloaded from www.ajcn.org by guest on January 24, 2011


those in the lower income category; this U-

reiteration of fatness trends, detailed in Ta- (.3

ble 2, raises the two questions as to whether -I

the respondents educational level is the ma-


jon parameter (on that of the spouse) and
whether educational level attained or house-
hold income are more closely related to the AGE
fatness trends. FIG. 1 . Relationship between educational level
Comparing fatness level both by educa- and iliac fatfold thickness in women between the 4th
tional level of the respondent and educa- and 8th decade. As shown, women with 8 years of
tional bevel of the head of the household, as education or less are systematically fatter than women
with 12 years of education or more. As shown, this
in Figure 1 , the first question is immediately inverse relationship between education and fatness in
answered. The trends are very much the the female is only partially attenuated when the hus-
same, by respondents educational level and band’s education is used instead.

TABLE 2
Effect of income level on fatfold thickness (mm)”

Subscapular Triceps Abdominal Iliac

Age group N
in- High in- Low in- High in- Low in- High in- Low in- High in-
come come come come come come come come

Males
25-35 136 11.0 14.2 15.2 16.2 24.7 27.1 19.7 23.3
35-45 189 17.2 16.9 13.6 15.8 31.6 29.5 22.4 24.9
45-55 149 14.4 18.3 14.3 16.0 29.6 30.6 21.5 26.1
55-65 123 15.3 23.4 13.6 17.3 26.8 33.1 21.7 25.5
65-75 82 14.0 16.7 12.6 13.7 25.0 20.0 16.9 16.7
Mean difference (mm) +3.5 + 1 .9 +0.6 +2.9
Percent difference (qf,) +24 +14 +2 +14
Females
25-35 156 15.1 13.5 22.4 22.9 30.0 26.5 20.9 15.9
35-45 222 18.8 16.8 23.5 23.2 34.8 27.4 24.8 16.9
45-55 169 25.0 17.0 30.0 24.8 43.4 30.7 32.0 24.1
55-65 132 21.7 17.7 27.3 28.4 44.8 37.0 32.6 25.7
65-75 101 23.0 16.3 27.6 25.1 41.0 29.7 35.0 26.2

Meandifference(mm) -4.5 -1.3 -8.5 -7.3


Percent difference (%) -22 -5 -22 -25

a Computer calculated using randomly generated fractional integers to allow decimal values of medians. b N
for the two income levels shown.
724 GARN ET AL.

tion and income) both relate to fatness.


Within families, however, income appears
to be the more important of the two conre-
lated variables. For comparable educational
level, more money makes for leaner females
and - within the income range of the sam-
pie-leaner males.

Discussion

This study confirms the highly systematic


relationship between socioeconomic status
and the level of fatness in adults as well as
the fact that the relationships are in opposite
directions for the two sexes. Among males
studied in a total community sample higher
education or higher income is associated
with greater fatness. Among females in the

Downloaded from www.ajcn.org by guest on January 24, 2011


very same community those with higher ed-
ucational attainment or higher family in-
come are dramatically leaner at four differ-
ent measuring sites.

L6 FEMALES
FIG. 2. Effects of education on fatness (above)
and of income on fatness (below) in adult males and
females, pooling all ages. Restricting this comparison
to the iliac fatfold, it is possible to show for three levels
of education and three levels of income, that both
educational level and income are related to the level of
fatness, but itt consistently opposite directions in males
and females. (These comparisons involve the $5,000 to
$10,000 income grouping and the 9 to 12 year cduca-
flon grouping and so differ from Tables I and 2 and
Fig. 1.)

income level constant and then holding edu-


cational level constant. The results graphed
in Figure 2 for the iliac fatfold show a de-
crease in fatness with increasing educational
level in the female amounting to some 33%
and a slight increase in fatness with increas-
ing educational level for males. Holding ed-
ucational level constant, income alone has a
considerable effect on fatness in both sexes.
MALES FEMALES
The females show a 29% decrease in iliac
fatfolds with increasing income while males FIG. 3. Systematic but contrasting relationship be-
tween educational level and fatness in males and fe-
register a 20% increase over the same three
males respectively. Using pooled-age data, 4,936 par-
income levels. ticipants. and three educational groupings, it is clear
‘l’here is, therefore, a systematic relation- lhat the male increases approximately 15% in iliac
ship between socioeconomic status and fat- fatfold thickness while the female decreases nearly
30% in iliac fatness in the progression from less than
ness in a total population sample, but in
high school education through to college and beyond.
opposite directions for the two sexes as ti- (This comparison includes all 4,956 participants and
dily summarized in Figure 3. Of the two thus differs from the education-restricted comparison
socioeconomic variables considered (educa- shown in the lower part of Fig. 2.)
LEVELS OF EDUCATION, INCOME, AND FATNESS IN ADULTS 725

These findings, based on nearly 5,000 tional attainment, but married into higher
adults, are consistent with what we have income families, are considerably leaner
previously reported from the lower-income than their low income peers indicates the
Ten State Nutrition Survey (9-1 1), but they extent to which social factors are far from
include additional socioeconomic measures, trivial in the development of obesity. fl
additional fatfolds, and attention to the in- The authors wish to thank Kenneth E. Guire for his
teraction between education and income. aid in data reduction and statistical operations. The
These findings help to explain the obsenva- manuscript was completed by Linda Kelly.
tion by Stunkard et al. (Ref. 7, p. 579) that
References
upper class women were systematically thin-
1. HAMMOND, W. H. Measurement and interpreta-
ner whereas “among boys as among men, tion of subcutaneous fat with norms for children
there were no such differences.” and young adult males. Bnit. J. Prey. Soc. Med. 9:
Both of the trends here reported are pat- 201, 1955.
2. BONNET, F., AND H. LOZET. Le contexte medico-
ently inconsistent with the genetic hypothe-
social de l’ob#{233}sit#{233} chez l’enfant.Acta Paediat.
sis, and they are inconsistent with the stratifi- Belg. 22: 211, 1968.
cation hypothesis (i.e., that the fat are con- 3. GOLDBLAi-r, P. B., M. F. MooRE AND A. SniN-
centrated in the lower income group of re- KARD. Social factors in obesity. J. Am. Med. As-

Downloaded from www.ajcn.org by guest on January 24, 2011


cent immigrants). Nor can the relationships soc. 192: 97, 1965.
4. MooRE, M. E., A. STUNKARD AND L. SROLE. Obe-
between socioeconomic status and adult fat-
sity, social class and mental illness. J. Am. Med.
ness be reconciled with the early-induction Assoc. 181: 138, 1962.
hypothesis. One would have to assume that 5 . SILVERSTONE, J . T. Obesity and social class. Psy-
the poor overfeed their daughters and un- chother. Psychosom. 18: 226, 1970.

denfeed their sons in early infancy, or that 6. SILVERSTONE, J. T., R. P. GORDON AND A. SWN-
KARD. Social factors in obesity in London. Practi-
the rich underfeed their infant girls yet over- tioner. 202: 682, 1969.
feed their baby boys. 7. STUNKARD, A., E. D’AQUILI, S. Fox AND R. D. L.
With data such as these , drawn from a FILI0N. Influence of social class on obesity and
single community (and with the women being thinness in children. J. Am. Med. Assoc. 221:
579, 1972.
the wives of the men), there is further rea-
8. WHITELAW, A. G. L. The association of social
son to rethink the conceptual basis of present class and sibling number with skinfold thickness in
fatness norms. If fatness is related to socio- London schoolboys. Human Biol. 43: 414, 1971.
economic status, as it certainly proves to be, 9. GARN, S. M., AND D. C. CLARK. Economics and
fatness. Ecol. Food Nutr. 3: 19, 1974.
then whose measurements should constitute
10. GARN, S. M., AND D. C. CLARK. The effect of
the norm? Should the fatness norm be de- socioeconomic status on growth, size, fatness and
nived from the poor (with fatter women), or proportions. In: Anthropology of the United
those more affluent (but with fatter men)? States, edited by W. C. Sturtevant, Washington,
Should the proper definition of obesity be D. C.: Anthropological Society of Washington,
1977.
group related , or purely statistical and based
1 1 . GARN, S. M., AND D. C. CLARK. Trends in fatness
on a National Probability Sample as we have and the origins of obesity. Pediatrics 57: 443,
previously discussed (16). 1976.
What we have here are major and defini- 12. NAPIER, J. A., B. C. JOHNSON AND F. H. Ep-
tive indications that the level of fatness (and STEIN. The Tecumseh community health study. In:
Case Book of Community Study, edited by I. I.
therefore the incidence of obesity) differs Kessler and M. L. Levin. Baltimore: Johns Hop-
according to educational level and level of kins Press, pp. 25-44, 1970.
income in an American community, and in 13 . Committee on Nutritional Anthropology. Recom-
differing directions in the two sexes. These mcndations concerning body measurements for
the characterization of nutritional status. Human
findings do not fit available animal models
Biol. 28: 115, 1956.
of fatness and obesity and, indeed, they 14. GARN, S. M., AND R. V. HARPER. Fat accumula-
point out the limitations of available animal tion and weight gain in the adult male. Human
models for obesity research. They point to Biol. 27: 39, 1955.
the need to investigate human obesity in 1 5 . BROEK, J . Quantitative description of body com-
position: Physical anthropology’s ‘fourth” dimen-
cultural context, in the context of the com- sion. Current Anthrop. 4: 3, 1963.
munity, in the family, and in behavioral con- 16. GARN, S. M. The measurement of obesity. Ecol.
text. The fact that women of lower educa- Food Nutr. 1: 333, 1972.

You might also like