Jul 10, 2011 - 64 y, oflndian ethnic origin, and usually resident in the community. Pregnant .... variable, normal was defined as a person with no past history.
Obesity in northern Canadian Indians: determinants, and consequences13 T Kue
Young
patterns,
and GustaafSevenhuysen ABSTRACT that
A survey
a large
aged 45-54 as gauged
alcohol the
use,
sum
age-adjusted
mean
both
and
and
poses
status,
and
thicknesses.
serum
diabetic
Indians
Cree
lipids,
Ojibwa
employment When
blood
hypertensive
was
were
three
pressures,
Obesity
for
future
associated
90%
with
of BMI and
were
found
of females
the
levels
of BMI
compared,
glycosylated
higher
hemoglobin
were
Calorie-adjusted and significant
is a significant chronic
Canada
Almost
was primarily ofthe central type analyses showed that age, sex,
categories glucose,
in northern obese.
energy was decreased. BMI was an independent
status. risks
Indians
groups
26. The obesity ratios. Multivariate
subjects whereas total in the obese subjects. particular
and
in all age-sex
public
diseases.
carbohypredictor
health
Am
J Clin
problem Nutr
1989;
49:786-93.
KEY WORDS
Obesity,
North
American
Indians,
Introduction
diseases,
Westernization
particularly
the
or modernization
demiologic
on
transition since
genetic
susceptibilities
as ischemic
brought
conditions
the end
diseases
of
heart
and diabetes (1), are relaIndians in terms of their burden (2). Such an epi-
has been
socioeconomic
so-called
such
disease, hypertension, stroke, tively new diseases for Canadian overall mortality and morbidity
ticularly
about and
of World
War
perhaps
by rapidly
lifestyles,
par-
II, superimposed peculiar
to
North
American Indians (3). Ethnographic observations, clinical impressions, and limited survey data support the conclusion that obesity is now very widespread in many groups oflndians. Yet, as recently as the 1940s two surveys in subarctic communities in the James Bay region and northeastern Manitoba revealed a native population hovering on the brink of starvation, with deficient energy and nutrient intakes (4, 5). By the l970s a nation-wide nutrition and anthropometric survey showed that Canadian Indians generally had much higher weight ans nationally (6).
This paper
reports
for age
on a study
compared
with
Canadi-
diseases
The study took place in six remote communities in northwestern Ontario and northeastern Manitoba, deep in the boreal forest of the central subarctic, inhabited by the culturally and linguistically closely related Cree and Ojibwa Indians.
To be eligible
on the prevalence
of obe-
Am J C/in Nuir
l989;49:786-93.
for inclusion
in the study,
subjects
must have
been between the ages of 20 and 64 y, oflndian ethnic origin, and usually resident in the community. Pregnant women were excluded from the study. The degree oflndian ancestry was not determined for the participants. In four of the communities all eligible adults were invited to participate, with an overall response rate of65%. In the two larger communities a 1 5% random sample was selected from the eligible list. People on the list but not in the random sample who wished to participate were also accepted; there were 1 27 such volunteers. The final cohort consisted of704 individuals (309 men and 395 women).
1 From the Department ofCommunity Health Sciences, Faculty of Medicine, and the Department ofFoods and Nutrition, Faculty of Human Ecology, University ofManitoba, Winnipeg, Manitoba, Canada. 2 Supported by the National Health Research Development Program of Health and Welfare Canada (6607-1383-53), the Manitoba Medical Services Foundation Inc (704-68), and the National Health
Research
sity, its determinants or risk factors, and its impact on the health ofthe population. It is part ofa larger, comprehensive study ofchronic diseases in Indians in the central subarctic region of Canada.
786
chronic
Methods
Chronic
changing
epidemiology,
Scholar
career
ada (6607-1377-48). 3 Address reprint nity
Health
Sciences,
award
requests
Printed
Health
and
Welfare
Department
ofManitoba,
750
Can-
of Commu-
Bannatyne
Avenue,
for Clinical
Nutrition
R3E 0W3. 1988.
for publication
in USA.
from
to TK Young,
University
Winnipeg, MB, Canada, Received March 25, Accepted
(TKY)
June
© 1989
15, 1988.
American
Society
Downloaded from www.ajcn.org by guest on July 10, 2011
for
adult
index (BMI) and girth
marital
of skinfold
observed in the obese drate intake was higher of
of 704
of individuals
y had a body mass by skinfold thickness
smoking, and/or
proportion
OBESITY
IN CANADIAN
A variety ofdata was collected, including sociodemographic, past health, cardiovascular symptoms, family history, lifestyle factors, anthropometric, physiologic, biochemical, electrocardiographic, and dietary history. (A research monograph detailing the design and methodology of the study is available from the first author.)
From
measurements
ofheight,
weight,
triceps
and subscapu-
lar skinfold thicknesses, and waist and hip circumferences, various indices of overall obesity and fat distribution were constructed. Lange calipers (Cambridge Scientific Industries, Cambridge, MD) were used to measure skinfold thickness. Subjects were requested to remove shoes, disrobe (except for underwear), and put on a disposable hospital gown for the anthropometry part of the examination. Waist girth was mea-. sured in the unclothed abdomen midway between the lower rib margin and the iliac crest and hip girth was measured posteriorly at the maximum protrusion ofthe gluteal area and anteriorly at the level ofthe symphysis pubis in accordance with the guide for anthropometric and nutritional assessment of Canadians published by the Department of National Health and Welfare (12).
For mercury
blood
pressure two sphygmomanometer
mc, Copiaque, computed
WHO
and
categories
NY) used
measurements were taken with a (Baumanometer#{174}, WA Baum, after 10 mm of rest. The means were as continuous variables in the analysis. The
of normal,
borderline,
and hypertensive
sure (DBP) 90 mm Hg. All others were considered hypertensive. Fasting blood samples were collected from all subjects and tested for glucose (FPG), glycosylated hemoglobin (HGB), total cholesterol (TCH), high-density lipoprotein (HDL), low-density lipoprotein (LDL), and triglycerides (TRG). A small subsamples of 143 subjects underwent a 2-h oral glucose-tolerance test. In the analysis diabetic status as a dichotomous variable was defined as a positive past history of physician-diagnosed diabetes (DM) or a current level of FPG 7.8 mmol/L in accordance with the WHO criteria(l4). A Hitachi 705 enzymatic analyzer (Hitachi, Tokyo, Japan) was used for the determination ofFPG, TRG, and TCH. HDL cholesterol was determined by a Chemetrics#{174} analyzer (Milliport Corp, Freehold, NJ) with the phototungstic acid precipitation method (15) and LDL cholesterol values were computed from TCH, HDL, and TRG values according to Friedewald’s formula (16). All data entry, management, and analyses were performed on a microcomputer with the software package STA TPAC (Walonick Associates, Minneapolis, MN). For certain statistical procedures, the SAS package ( 1 7) was used with a mainframe computer. This study was approved by the Faculty Committee on the Use ofHuman Subjects in Research ofthe University of Manitoba. Signed informed consent was obtained from all individual participants in the study. The Indian band councils in the communities surveyed also approved the study.
Results Obesity
Tables 1 and 2 show the correlation matrices of various anthropometric indices in this study for males and females, respectively. All the measures correlate poorly with height whereas body mass index (BMI) and the
girths in particular correlate strongly with weight and thus possess the desirable criteria for indices of overall obesity (1 8-20). BMI, also known as Quetelet index (wt/ ht2), is the preferred measure because existing standards for classifying individuals into obesity categories as well as Canadian
national
reference
data
are
available
from
the 198 1 Canada Fitness Survey and 1978 Canada Health Survey (2 1 22). BMI was shown in other studies to correlate with more direct measures ofbody fat such as skinfold thickness (23) and body densitometry (24, 25). However, the correlation between BMI and skinfold thickness was not strong in our sample. It was, however, ,
stronger thickness.
triceps sure may with
were
also used (1 3). When the category hypertensive was analyzed as a dichotomous variable, normal was defined as a person with no past history of physician-diagnosed hypertension or use of antihypertensive drugs, and with current systolic blood pressure (SBP) 140 mm Hg and fifth-phase diastolic blood pres-
indices
tio
with subscapular In the analysis
skinfold
the
thicknesses
than sum
with ofthe
is used as an alternative
of overall obesity to BMI well represent different different health risks.
because
fold thicknesses (STR) have centrality of fat distribution other.
measures As indices
were
only
offat
the
two
physiological
In the epidemiological literature both (WHR) and the ratio of subscapular
two
triceps skinfold subscapular and
meaindices
constructs
the waist-hip rato triceps skin-
been used as measures (26-28). In our sample
weakly distribution
correlated they
with should
one correlate
of the
an-
Downloaded from www.ajcn.org by guest on July 10, 2011
Procedures and questionnaires described in the World Health Organization (WHO) manual on cardiovascular survey methods were adopted whenever possible (7). A semiquantitative food frequency questionnaire was designed to assess usual diet (8). The food types and usual amounts consumed by each subject were recorded during the food frequency interview conducted by a dietitian. These data were transposed into estimates ofnutrient consumption per day based on the 3200-item Canadian Nutrient File prepared by the Department of National Health and Welfare, updated in 1986 (9). Calculations of nutrient content were carried out by use of the Nutrient Analysis Program (NAP) available in the University of Manitoba mainframe computer. To assess physical activity the Lipid Research Clinic instrument (two questions, three categories) was used ( 10). Those who did not engage in strenuous exercise or hard physical labor were classified as inactive. The active group could be further subdivided into very and moderately active depending on whether individuals exercised three times or more per week or less than three times per week, respectively. Caloric intake adjusted for body weight (1 1) could be used when physical activity was analyzed as a continuous variable. To reflect the episodic drinking pattern of the residents, the frequency of consumption over a l-y period rather than precise quantification was assessed. Three categories were recognized: not at all in the past year, a few times in the past year, and at least once a week. In analyses requiring a dichotomous variable, the last two were combined into a drinker category. The category bush living characterized subjects who spent a substantial portion of their time ( 1 mo continuously every year during the past 5 y) in traditional Native pursuits such as hunting, trapping, and fishing.
787
INDIANS
YOUNG
788 TABLE
AND
SEVENHUYSEN
1
Correlation
matrix
ofanthropometric
Ht Ht Wt BMI
indices
Wt
for males*
BMI
Tri
Sub
SUM
STR
Wai
0.93
Tri
0.05
0.43
0.44
Sub
0.04
0.63
0.65
SUM
0.05
0.58
0.59
0.92
0.89
STR
-0.03
0.05
0.06
-0.61
0.06
-0.33
Wai Hip
0.11 0.14
0.92 0.83
0.93 0.83
0.41 0.41
0.62 0.54
0.56 0.52
0.08 0.06
0.86
WHR
0.03
0.68
0.70
0.27
0.50
0.42
0.07
0.79
Ht, height; Wt, weight;
subscapular WHR, ratio
0.37
-
skinfold ofwaist
thicknesses; STR, to hip circumferences.
ratio
Tn,
triceps
of subscapular
skinfold to triceps
thickness; skinfold
thicknesses;
Wai,
skinfold waist
thickness;
SUM,
circumference;
Hip,
sum
oftriceps
and
hip circumference;
decreased from just over 3 in age group 25-34 y to 1.5 in the group aged 55-64 y. Similarly, if skinfold thicknesses are used for comparison with national standards from the Nutrition Canada Survey conducted in the early l970s (M Jette, unpublished observations, 1983), values for Indians generally exceeded those for Canadians, with the exception oftriceps skinfold thicknesses of older women (Fig 2). No comparable Canadian national data for centrality measures are available. The obesity of this population was predominantly of the central type. Table 3 shows that as BMI increased, the proportion of subjects with WHR and STR values above the top quartile ofthe total sample (males and females combined) also increased, with the exception of STR in men. Although the overall level of obesity was much higher in Indian women, a higher proportion of obese men were ofthe central type.
This feature is well in males, whereas BMI and weight in
‘-
ofobesity
The excessive level ofobesity in Indians in our sample compared with the Canadian national population according to the Canada Fitness Survey of 198 1 (2 1) is evident from Figure 1 In some age-sex groups, eg, women aged 45-54 y, almost 90% ofthe Indians had BMIs in the overweight or obese range. In both men and women and in all age groups the proportion ofthe population with a BMI 26 was higher than it was in Canadians. The ratio of BMI values for Indians to values for Canadians was ‘--2 for males in the youngest (20-24 y) age group and 5 for females. This ratio declined with increasing age. Above the age of 25 y the proportion of men with BMI 26 was only 20-30% higher for Indians when compared with Canadians nationally. For women the ratio .
TABLE
Sub, subscapular
Determinants of obesity Within the study sample sity
indices
between
various
there
were
differences
subgroups
in obe-
of subjects
cate-
2
Correlation
matrix
ofanthropometric
Ht
Ht
-
Wt
0.23
indices
for females*
BMI
Wt
Tri
BMI
-0.09
0.95
Tn Sub SUM ST! Wai Hip
0.01 0.00 0.01 -0.03 0.08 0.10
0.19
0.20
0.55
0.56
0.63
0.39 0.21 0.85 0.86
0.40 0.22 0.84 0.84
0.93 -0.61 0.06 0.23
WHR
-0.01
0.47
0.47
-0.21
Ht, height;
subscapular WHR, ratio
Wt, weight;
skinfold ofwaist
body
mass
thicknesses; STR, to hip circumferences.
BMI,
ratio
index;
Tri, triceps
of subscapular
Sub
SUM
STR
Wai
0.88 -0.01 0.50 0.56
-0.38 0.28 0.42
0.32 0.21
0.89
-0.03
0.32
0.73
0.20
skinfold to triceps
thickness; skinfold
Sub,
subscapular
thicknesses;
Wai,
skinfold waist
thickness;
circumference;
SUM, Hip,
Hip
WHR
0.34
-
sum oftriceps hip circumference;
and
Downloaded from www.ajcn.org by guest on July 10, 2011
Prevalence
0.64
BMI, body mass index;
only minimally with overall obesity. demonstrated by STR, particularly WHR correlated moderately with both males and females.
I
WHR
-
0.32 -0.04
I
Hip
OBESITY
IN CANADIAN
789
INDIANS
100
Subscapular
Skinfold
90
Female
Male
80 25
70 60 20
50
40
E E
30
Cl)
,
Indians
/
, /
15
-
-
Canada
-
-
Canada
(1)
20
a) C
10#{149}
10 C)
a)
-C
0
0.
Triceps
I-
a)
100
0 C
Indians
0
90
a)
Skinfold
V
C
>
80
a)
70
20 Cl) C
Indians
a)
15 Canada 10
2029
3039
4049
5059
6069
Age
2029
30-
40-
50-
60-
39
49
59
69
Group
FIG 2. Mean subscapular and triceps skinfold thicknesses and sex: comparison with Canadian national population.
Age
by age (y)
Group Indians
Canada
Dietary
correlates
::::::.:::
BMI
26-30
BMI>30
BMI
26-30
Total
BMI>30
energy
similarly FIG 1. Categories ison with Canadian
ofbody national
index(BMI)by population.
mass
age and sex: compar-
was decreased
lowered
observed
intake
in the obese
intercorrelation and with each
of other
To determine gorized by sex, age group, and other sociodemographic and lifestyle factors. Stepwise multiple-regression analyses were performed separately for males and females to determine significant predictors ofBMI and sum of skinfold
thicknesses
An increase BMI in both employed
men
(Tables
4 and
5).
in age was associated males and females. also
had
significantly
with an increase in Married women and higher
BMI
signifiNonthick-
nesses.
do not
inactive
men
and
women
who
subjects
(Table
most (29).
dietary
intake,
nutrient
justed
nutrient
scores
proposed
TABLE
3 ofbody
mass
index
STR
>
(BMI)
subjects.
nutrients
A
was also
6), reflecting
variables
the role ofindividual
of caloric
obese
with
nutrients
density
or the
by Willett
to centrality
and
the energy
indepencalorie-ad-
Stampfer
indices*
1.64
WHR
>
0.99
when
other variables were controlled for. Age was not cantly associated with sum ofskinfold thicknesses. smokers in both sexes had lower sums of skinfold Physically
major
dent
Relation
in the
ofmost
use alcohol regularly or are unemployed tended to have significantly higher sums of skinfold thicknesses. The discrepancy between predictors ofBMI and sum of skinfold thicknesses suggest that the two measures of obesity are not interchangeable. However, the r2 values for mdividual predictors were generally small and the overall variance explained in each of the regression models was only 10%.
BMI
Male
Female
Male
Female
%
30
28.0
35.2
75.0
18.2
Total
29.8
22.4
36.0
11.3
1.01 0.604
19.6 0.000
77.6 0.000
18.7 0.000
of subscapular
to triceps
x2 p I
ratio
STR, ratio ofwaist
to hip circumferences.
9.6
skinfold
1.6
thicknesses;
WHR,
Downloaded from www.ajcn.org by guest on July 10, 2011
0
790
YOUNG
TABLE
4
Predictors
ofobesity
indices:
body
mass
Significant
Fe male
r2
predictors
/3
0.0776
0.3041
r2
3
0.0735
0.2485
status
(married
=
0, single
Employment (no=0,yes=
=
1)
-
1)
0.0 133
-
0.0243
0.1581
-
Totalr2
0.1020
0.0868
r SEM F
0.3193
0.2946
*
Other variables
hol use, and physical
tested
4.39
5.66
17.32
18.39
were education,
bush
-0. 1 174
-
living,
SEVENHUYSEN Because few coexisting ischemic heart disease cases have been discovered in the sample (based on medical history, Rose questionnaire [7], or electrocardiography), their association with obesity was not investigated. The survey also inquired into past history of clinical gallbladder disease (as detected by radiography and/or
index
M ale
Age Marital
AND
surgery)
tively) smoking,
alco-
although
those
without
past
disease
were
not
screened for current silent gallstones. Overall, 1 8.5% of women and 2.3% of men had previously detected gallbladder disease (GBD). Women with a past history of GBD had higher sum ofskinfold thicknesses, STR, BMI, and WHR. Of these, the differences in BMI and WHR were statistically significant (.p = 0.048 and 0.01 3, respecon one-way
analysis
of variance (p
WHR was statistically significant ANOVA, adjusting for the effect
(ANOVA). =
ofage
Only
0.04) on two-way groups.
activity.
Discussion
calorie-adjusted
measures
observed when absolute body weight were used. Metabolic
removed
intakes
the
sexual
or intake
difference
consequences
risks
BMI has an independent and significant effect on diabetic status, a dichotomous variable based on both past history and current FPG levels. When multiple logistic regression analysis is used to control for other sociodemographic, lifestyle, and metabolic factors, the odds ratio for diabetes was 1 .59 (95% confidence interval, 1.142.22) when the 75th percentile of BMI was compared with the 25th. Note that in a cross-sectional study the odds ratios (relative risks) refer to the risk of having the disease not ofdeveboping the disease. For hypertensive status (known past history or current SBP > 140 or DBP> 90), the odds ratio was 1.44(1.121 .83),
25th.
comparing
the
75th
percentile
of BMI
with
the
high
limited
level
of obesity
national
in this
survey
data
population
available
confirms on
Canadian
Indians. Considered to be relatively unacculturated until relatively recent times, ie, after the World War II, this group oflndians ofAlgonkian stock in the subarctic cubture area shares the same anthropometric characteristics as many other more southerly located tribes. The sociodemographic and lifestyle determinants of obesity
per kilogram
BMI categories also differed in terms of various metabolic and physiologic measures on univariate comparisons or when sex-specific, age-adjusted means are compared (Table 7). Obese subjects had unfavorable lipid profiles: decreased HDL and HDL:total cholesterol and increased TRG and LDL. Both SBP and DBP were increased in the high-BMI group (with the exception of SBP in men). Both FPG and HGB, the latter a more naturab measure ofusual glucose status (3 1), were higher for the obese subjects. Disease
The the
identified
in this
study
are compatible
with
find-
ings from surveys in other populations. In the Canada Health Survey (22), age, marital status, smoking, family income, and education were significant predictors of BMI in both sexes. Although there is no direct measure ofincome in this study, employment was positively associated with BMI for males but negatively associated with
TABLES Predictors
ofobesity
indices:
sum ofskinfold
thicknesses
M ale Significant
Fe male
r2
predictork
Alcohol use (no = 0, yes = 1) Physical activity (inactive = 0, active= 1) Employment (no = 0, yes = I) Smoking (smoker = 0, nonsmoker = 1)
fi
-
-
0.0261
-0.1618
-
-
0.0658
Totalr2
0.0919
r
0.3031
0.2475
r2
0.0762
0.022
0.0170
F
14.62
16.16
bush
living.
tested
were
1
-0.1430
0.1397
0.3396
17.20
variables
-
0.1153
12.74
Other
-0.1926
-
SEM
*
fi
age,
education,
marital
status,
and
Downloaded from www.ajcn.org by guest on July 10, 2011
(30) can be used. These measures generally change the direction ofthe association between obesity and absolute intake, or nutrient per kilogram. Subjects with a high BMI consumed significantly more carbohydrates independent oftotal caloric intake than did those with a low BMI. Fat intake was significantly decreased in obese subjects when nutrient density was used but not when cabrie-adjusted scores were used as the measure. The use of
OBESITY TABLE Mean
IN
CANADIAN
INDIANS
791
6 daily
nutrient
intake
by obesity
category
and sex Main
effect:
BMI
Main
effect:
sex Interaction
Variable
30
cholesterol
LDL cholesterol
31
p
31
p
(mmol/L)
1 .42
1 .26
1 . 16
0.000
1 .42
1 . 18
0.98
0.000
(mmol/L)
2.98
3.30
3.30
1.52 0.25
1.85 0.23
3.30 1.06 0.28
3.67 2.07 0.2 1
3.67 2.20 0. 18
0.003
I . 18 0.29
0.00 1 0.0 12 0.000
Triglycerides (mmol/L) HDL:total cholesterol Blood Pressure Systolic(mmHg) Diastolic(mmHg)
121.67 72.51
127.19 78.50
0.000 0.000
128.73 80.41
0.000 0.000
132.46 80.74
134.10 83.49
5.53 7.87
0.004
4.89
5.42
5.55
0.094
0.000
6.83
7.45
7.74
0.0 12
133.68 87.15
0.398 0.000
Glucose
Fasting glucose (mmol/L) Glycosylated hemoglobin
(%)
4.68 6.65
5. 13 7.31
Downloaded from www.ajcn.org by guest on July 10, 2011
Fiber(g/kg)
Calorie-adjusted Proteins Fats Carbohydrates Cholesterol
26-30
792
YOUNG
AND
SEVENHUYSEN
Table 6, energy per kilogram body weight is lowest for people with a BMI > 30. However, physical activity as a dichotomous variable was a significant determinant of sum of skinfobd thicknesses in men but not of BMI in either sex (Tables 4 and 5). This study also provides cross-sectional evidence of the adverse health impact of obesity. Of particular concern
is that
the
predominant
pattern
of obesity
is of the
ger that
resulted
in the current
are
undergoing
ous.
There
promotion
is thus
some
studies
an
epidemiologic
strategies
urgency
and in this
transition
to conduct
intervention
and L Wilsack conducted
N Ling, and B Roche conducted all physical examinations.
base-line
trials
epi-
of health-
7. 8. 9. 10.
ity,
and assistance ofthe Chiefs and the staff and manageunits. Ms L Wadsworth, J
the dietary the remainder
treadmill
interviewsand
MsJ
of the interviews
Hovi,
and
13.
World
Health
WHO expert
test
performance
and
plasma
HDL
choles-
series
Organization.
Arterial
committee.
Geneva:
hypertension:
WHO,
report
of
1978. (Technical
a
report
628.)
14.
World
15.
study group. Geneva: WHO, 1985. (Technical report series 727). Calvert GD, Yeates RA, Mannik T. A comparison ofthe phosphotungstate magnesium precipitation and heparin-manganese pre-
Health
16.
Organization.
procedures
lesterol.
Diabetes
for estimating
Ann Clin Biochem
Friedewald
WT,
Levy
mellitus:
report
high-density
of a WHO
lipoprotein
cho-
1980; 17:105-8.
RI,
Fredrikson
centration ofbow-density out use of the preparative
DS.
Estimation
of the
con-
lipoprotein cholesterol in plasma, withultracentrifuge. Clin Chem 1972; 18:
499-502.
17. Statistical Analysis System Institute. SAS user’s guide: statistics. 5th ed. Cary, NC: SAS Institute, mc, 1985. 18. Keys A, Fidanza F, Karvonen MJ, et al. Indices ofrelative weight and obesity. J Chron Dis l972;25:329-43. 19. Khosla T, Lowe CR. Indices of obesity derived from body weight and height. Br J Prey Soc Med 1967;2 1:122-8. 20. Fborey C duV. The use and interpretation of ponderal index and other weight-height ratios in epidemiobogical studies. J Chron Dis 1970;23:93-103.
21.
Millar in
WJ,
Britain,
Stephen
T. The
Canada
and
prevalence
the
United
l987;77:38-4l. 22. Evers S. Economic and social factors adult Canadians. Nutr Res 1987;7:3-l3. 23.
Micozzi body
References 24. 1 . Trowell HC, Burkitt DP, eds. Western diseases: their emergence and prevention. Cambridge, MA: Harvard University Press, 1981. 2. Young TK. Are subarctic Indians undergoing an epidemiologic transition? Soc Sci Med l988;26:659-7 1. 3. Weiss KM, Ferrell RE, Hanis CL. A New World Syndrome of met-
exercise
terol: the Lipid Research Clinic’s Program Prevalence Study. Circulation l980;62(suppl 4):53-61. 1 1. Sopko G, Jacobs DR. Taylor HL. Dietary measures of physical activity. Am J Epidemiol 1984; 120:900-1 1. 12. Department of National Health and Welfare Canada. Guide for anthropometric classification of Canadian adults for use in nutritional assessment. Ottawa: Department of National Health and Welfare 1983.
#{163}3
population.
We wish to acknowledge the cooperation and Council of the six Indian communities ment of the local Indian Health Services
Britten,
6.
cipitation
and the increasing prominence ofchronic diseases in the near future can be expected (2). The conception that these diseases are diseases of affluence and thus would spare the economically disadvantaged Indians is erronedemiobogicab
5.
“epidemic.”
In a cross-sectional design where temporal sequence cannot be ascertained, the designation ofcertain factors as determinants and others as effects is often subject to debate. Long-term prospective studies are needed to elucidate cause-and-effect relationships. Indians in subarctic Canada
4.
MS. mass
Albanes
indices
D, with
Jones
weight,
of overweight States.
associated
V, Chumlea stature,
Am
WC.
and
J Public
with
obesity Health
obesity
in
Correlation
of
composition in II. Am J Clin Nutr l986;44: and
body
men and women in NHANES I and 725-31. Revicki DA, Israel RG. Relationship between body mass indices and measures of body adiposity. Am J Public Health l986;76: 992-4.
25.
Roche AF, Siervogel RM, Chumlea fatness from limited anthropometric 198 l;34:283 1-8.
WC,
Webb P. Grading data. Am J Clin
body
Nutr
Downloaded from www.ajcn.org by guest on July 10, 2011
central, or centripetal, type. Increasingly, more epidemiobogic studies of a variety of populations have demonstrated that central fat distribution is an independent risk factor ofsuch chronic diseases as diabetes, hypertension, gallbladder disease, ischemic heart disease, and stroke (26, 27, 35-37). These observations are supported by metabolic studies ofregional adipose tissues (38, 39). There may well be a common underlying derangement in lipid and carbohydrate metabolism that beads to the development of these obesity-associated diseases. Insulin resistance has been suggested as the underlying derangement from the Pima, Israel glucose intolerance, obesity, and hypertension (GOH), and Texas Mexican studies (40-42). The high prevalence of obesity and its association with diabetes and gallbladder disease bed some population geneticists to postulate the existence of a New World Syndrome among Amerindians and the natural selection ofa genetic trait that conferred survival advantages during the early settlement of humans in North America (3). Rapid social change in the past half century, perhaps through decreased physical activity and/or dietary change, provided the environmental trig-
diseases with a genetic and evolutionary basis. Yearbk Phys Anthropol l984;27: 153-78. Vivian RP, McMillan C, Moore PE, et al. The nutrition and health oftheJames Bay Indian. Can Med AssocJ l948;59:505-l8. Moore PE, Kruse HD, Tisdall FF, Corrigan RSC. Medical survey ofnutrition among the northern Manitoba Indian. Can Med Assoc J l946;54:223-33. Department of National Health and Welfare. Nutrition Canada anthropometry report: height, weight and body dimensions. Ottawa: Department ofNational Health and Welfare, 1980. Rose GA, Blackburn H, Gillum FR, et al. Cardiovascular survey methods. 2nd ed. Geneva: World Health Organization, 1982. Sampson L. Food frequency questionnaires as a research instrument. Clin Nutr 1985;4:171-8. Department of National Health and Welfare. Canadian nutrient file. Ottawa: Department ofNational Health and Welfare, 1986. Haskell WL, Taylor HL, Wood PD, et al. Strenuous physical activabolic
OBESITY 26.
Hartz iU, surements 1984;
27.
Rupley DC, Rimm AA. The association with disease in 32,856 women. Am
IN
CANADIAN
of girth meaJ Epidemiol
119:71-80.
Lapidus
L, Bengtsson C, Larsson B, et al. Distribution of adipose and risk of cardiovascular disease and death: a 12-year foblow-up ofparticipants in the population study ofwomen in Gothenburg, Sweden. Br Med J l984;289: 127-61. 28. Haffner SM, Stern MP, Hazuda HP, et al. The role of behavioral variables and fat patterning in explaining ethnic differences in serum lipids and lipoproteins. Am J Epidemiol 1986; 123:830-9. 29. Gordon T, Fisher M, Rifkind BM. Some difficulties inherent in the interpretation of dietary data from free-living populations. Am J Clin Nutr 1984;39: 12-6. tissue
30.
Willett
WC,
Stampfer
M. Total
energy
intake:
implications
Baecke
JAH,
van
habitual physical Am J Clin Nutr 34.
Gordon
T, Kagan
Staveren
WA,
activity, and body 1983; 37:278-86. A, Garcia-Palmieri,
Burema
fatness
J. Food
consumption,
in young
Dutch
adults.
to coronary
adipose
39.
and
its relation-
disease
and
death
in three
populations.
Cir-
40.
tissue
distribution.
insulin
resistance
Caucasians.
41.
42.
J Clin
Invest
1983;72:
1 150-62.
Kissebah AH, Vydelingum N, Murray R, et al. Relation of body fat distribution to metabolic complications ofobesity. J Clin Endocrinol Metab 1982; 54:254-60. Nagulesparan M, Savage PJ, Knowler WC, et al. Increased in vivo in
Diabetes
non-diabetic
Pima
Indians
compared
with
1982;3l:952-6.
Modan M, Halkin H, Almog S. et al. Hyperinsulinemia-a link between hypertension, obesity and glucose intolerance. J Clin Invest l985;75:809-17. Haffner SM, Stern MP, Hazuda HP, et al. Hyperinsulinemia in a population
et al. Diet
heart
N EnglJ
at high
Med
risk
ofnon-insulin-dependent
1986;3l5:220-4.
diabetes
mellitus.
Downloaded from www.ajcn.org by guest on July 10, 2011
33.
ship
culation 1981;63:500-lS. 35. Szathmary EJE, Holt N. Hyperglycemia in Dogrib Indians of the NWT, Canada: association with age and a centripetal distribution ofbody fat. Hum Biol 1983;SS:493-SlS. 36. Welin L, Svardsudd K, Wilhelmsen L, et al. Analysis ofrisk factors forstrokeinacohortofmenbornin 1913. NEnglJ Med 1987;317: 52 1-6. 37. Blair D, Habicht J-P, Sims EAH, et al. Evidence for an increased risk for hypertension with centrally located body fat and the effect ofrace and sex on this risk. Am J Epidemiol 1984; 1 19:526-40. 38. Krotkiewski M, Bjorntorp P. Sjostrom L, Smith U. Impact of obesity on metabolism in men and women: importance of regional
for epi-
demiobogic analysis. Am J Epidemiol 1986; 124:17-27. 31. Duncan BB, Heiss G. Nonenzymatic glycosylation of proteins-a new tool for assessment ofcumulative hyperglycemia in epidemiologic studies, past and future. Am J Epidemiol 1984; 120:169-89. 32. Folsom AR, Caspersen CJ, Taylor HL, et al. Leisure time physical activity and its relationship to coronary risk factors in a population-based sample: the Minnesota Heart Survey. Am J Epidemiol 1985; 12 1:570-9.
793
INDIANS