Research

Prevalence of overweight and obesity and their association

with hypertension and diabetes mellitus in an Indo-Asian
population
Tazeen H. Jafar, Nish Chaturvedi, Gregory Pappas
@ See related article page 1081

alized countries, those in the developing world appear to be

Abstract at greater risk of the diseases associated with overweight,
and cardiovascular disease has become the leading cause of
Background: The associations of body mass index (BMI) and disability and death in many developing countries.3–5
chronic disease may differ between Indo-Asian and Western Worldwide, Indo-Asian people are among the popula-
populations. We used Indo-Asian-specific definitions of tions at highest risk for cardiovascular disease.6 Evidence
overweight and obesity to determine the prevalence of these also suggests that associations between body mass index
problems in Pakistan and studied the sensitivity and speci- (BMI), percentage of body fat and chronic disease may differ
ficity of BMI cutoff values for an association with hyperten- between Indo-Asian and European populations.7,8 One pro-
sion and diabetes mellitus. posed explanation for these observations is the effect of
Methods: We analyzed data for 8972 people aged 15 years or poverty and resultant malnutrition during intrauterine and
more from the National Health Survey of Pakistan (1990– early childhood years, coupled with relative overnutrition in
1994). People considered overweight or obese were those later years.9 There may, however, be other factors at work.
with a BMI of 23 kg/m2 or greater, and those considered Because of the observed differences between populations,
obese as having a BMI of 27 kg/m2 or greater. We built multi- the International Association for the Study of Obesity and the
variable models and performed logistic regression analysis. International Obesity Task Force have suggested lower BMI
Results: The prevalence of overweight and obesity, weighted cutoff values for the definitions of overweight (23.0–24.9
to the general Pakistani population, was 25.0% (95% confi- kg/m2) and obesity (25.0 kg/m2 or greater) in Asian popula-
dence interval [CI] 21.8%–28.2%). The prevalence of obesity tions.10,11 Estimates of the prevalence of overweight in Indo-
was 10.3% (95% CI 7.0%–13.2%). The factors independently Asian countries (India, Pakistan, Bangladesh and Sri Lanka)
and significantly associated with overweight and obesity in- based on these cutoff values have not been reported. Further,
cluded greater age, being female, urban residence, being lit- it is not known whether the revised definition of overweight
erate, and having a high (v. low) economic status and a high would be valid at a population level in terms of being better
(v. low) intake of meat. With receiver operating characteris- associated with the consequences of obesity.
tic curves, we found that the use of even lower BMI cutoff We therefore sought to determine the prevalence of over-
values (21.2 and 22.1 kg/m2 for men and 21.2 and 22.9 kg/m2 weight and obesity using the Asian-specific BMI definition in
for women) than those recommended for an Indo-Asian a Pakistani population. We also studied factors and condi-
population yielded the optimal areas under the curve for an tions (hypertension, diabetes mellitus and raised serum cho-
association with hypertension and diabetes, respectively. lesterol concentration) associated with being overweight or
Interpretation: A quarter of the population of Pakistan obese in the Pakistani population, as well as the sensitivity
would be classified as overweight or obese with the use of and specificity of various BMI cutoff values for their associa-
Indo-Asian-specific BMI cutoff values. Optimal identification tion with hypertension and diabetes mellitus.
of those at risk of hypertension and diabetes and healthy tar-
gets may require the use of even lower BMI cutoff values Methods
than those already proposed for an Indo-Asian population.
The analysis was based on cross-sectional data collected in
DOI:10.1503/cmaj.060464

CMAJ 2006;175(9):1071-7
the 1990–1994 National Health Survey of Pakistan, conducted
by the Pakistan Medical Research Council, under the techni-

D eveloping countries are increasingly vulnerable to

the worldwide epidemic of obesity, which affects all
segments of the population, including men, women
and now children.1,2 Compared with populations in industri-
cal guidance and support of the US National Center for
Health Statistics. The overall design of the survey was a modi-
fication of the Third National Health and Nutrition Examina-
tion Survey (NHANES III), conducted by the US National Cen-

CMAJ • October 24, 2006 • 175(9) | 1071

© 2006 CMA Media Inc. or its licensors
 Research

vised criteria for Asian populations, 11 was

Table 1: Age-specific prevalence of overweight and obesity* in Pakistan lower than the conventional cutoff value of
Overweight or obese women Overweight or obese men
25 kg/m2 for populations of European origin
and reflects the higher ratio of body fat to mus-
Age, yr No. % (95% CI)† No. % (95% CI)† cle mass in the former. Obesity was defined as
a BMI of 27 kg/m2 or greater.
15–24 195/1373 13.8 (10.5–17.1) 179/1352 12.4 (9.0–15.8)
Ethnicity was reported as “mother tongue,”
25–34 263/983 26.3 (21.6–31.0) 225/904 24.1 (19.3–28.9) which is specific for each of the 5 major ethnic
35–44 340/777 42.8 (37.1–48.5) 231/712 31.4 (26.8–36.0) subgroups of Pakistan: Muhajir, Punjabi, Ba-
45–54 263/623 41.3 (35.4–47.2) 161/559 28.2 (22.6–33.8) luch, Sindhi and Pashtun. Literacy was defined
55–64 140/444 33.0 (27.6–38.4) 114/427 24.9 (20.1–29.7) as whether an individual could read or not.
≥ 65 95/358 25.1 (19.5–30.7) 109/460 21.4 (17.7–25.1) Three levels of economic status were defined
All 1296/4558 28.2 (24.7–31.7) 1019/4414 22.0 (18.9–25.2)
by simply counting owned items with use of a
list of household electrical items and owned
Note: CI = confidence interval. transport vehicles; this measure has been vali-
*The category of overweight and obesity was defined as a body mass index (BMI) ≥ 23 kg/m .
2

†Weighted to the general Pakistani population.

dated.17 High intake of meat, fruit, milk, eggs,
rice and potatoes was defined as consumption
of these items at least every other day. Use of
ter for Health Statistics, tailored to the needs of Pakistan. The tobacco (cigarettes, “beddies” [hand-rolled, often filterless
details of sampling, design, components, survey instruments cigarettes wrapped in temburni leaf or tendu leaf that are
and quality control have previously been reported.12 Ethical available in a variety of candy-like flavours], and chewing
approval for the survey was obtained from the Institutional tobacco or snuff) was dichotomized as current use or not.
Review Board at the Pakistan Medical Research Council.13 Hypertension was defined as a systolic blood pressure of
In brief, the survey was conducted on a nationally repre- 140 mm Hg or greater or a diastolic blood pressure of
sentative sample of 18 135 people aged 6 months to 110 years 90 mm Hg or greater (based on the mean of the 2 readings)
from 2400 urban and rural households who gave informed or current therapy with antihypertensive medication. Dia-
consent. A 2-stage stratified design was used.14 The urban betes was defined as a nonfasting blood glucose concentra-
and rural areas of each of the 4 provinces of Pakistan were tion of 140 mg/dL (7.8 mmol/L) or greater or a history of dia-
taken as strata. There were 80 urban or rural primary sam- betes; this definition, based on old criteria,18 diverges from
pling units; 30 households were drawn into the sample from the recent standard criterion of a fasting blood glucose
each unit, and all residents of the household were included in concentration of more than 126 mg/dL (7.0 mmol/L).19 Pro-
the study. The overall individual response rate was 92.6%. teinuria was defined as a urine protein concentration of
Data on demographic, socioeconomic and health-related 30 mg/dL (0.3 g/L) or greater as measured by a reagent strip
variables were collected with a questionnaire validated in (reading of 1+ or greater).20
local languages. Dietary data were collected with a food- Prevalence and 95% confidence intervals (CIs) were calcu-
frequency questionnaire. All women aged 40 years or under lated with weighting that reflected the oversampling of urban
were asked whether they were currently pregnant. Physicians areas and the 3 smaller provinces. Multivariable models were
at mobile examination centres performed a standardized built for the primary outcome of overweight. The candidate
physical examination that included 2 blood pressure readings predictor list included sociodemographic variables (age, sex,
obtained at least 20 minutes apart from the right arm by urban v. rural dwelling, literacy status, economic status, high
means of a mercury sphygmomanometer with the subject intake of meat, fruit, milk, eggs, rice and potatoes, and cur-
sitting. Trained technicians performed anthropometric exa- rent cigarette use). Variables associated with the primary out-
minations. Weight and height were recorded while the sub- come with p < 0.1 in the stepwise multiple regression analy-
ject was in light clothing and without shoes. BMI was calcu- sis were considered for selection in the multivariable model.
lated as weight (in kilograms) divided by height (in metres)
squared. Blood samples were obtained at least 1 hour after the Table 2: Prevalence of overweight and obesity in the surveyed
subject arrived at the examination centre; fasting was not Pakistani population aged 15 years or more according to various
required. Blood glucose and serum cholesterol concentra- BMI cutoff values
tions were determined with the use of the Reflotron, a multi-
phasic biochemical analyzer.15 Random midstream urine Proportion overweight or obese, % (95% CI)
BMI,
samples were tested with Multistix 10 SG urinalysis strips kg/m 2
All Women Men
(Bayer HealthCare, Diagnostics Division, Tarrytown, NY;
reagent, tetrabromophenol blue) in subjects aged 5 years or ≥ 23 25.0 (21.8–28.2) 27.9 (24.4–31.4) 22.0 (18.5–25.5)
over. Quality-control measures included a field visit by expert ≥ 25 15.7 (12.9–18.6) 18.6 (15.5–21.7) 12.5 (9.6–15.4)
consultants, duplicate examination by field supervisors, cali- ≥ 27 10.3 (7.0–13.2) 12.9 (10.1–15.7) 7.5 (4.7–10.3)
bration protocols and retraining exercises.16 ≥ 30 5.7 (3.4–8.4) 7.3 (4.6–10.0) 4.0 (1.3–5.3)
The category of overweight and obesity was defined as a
BMI of 23 kg/m2 or greater. This definition, based on the re- *Weighted to the general Pakistani population.

CMAJ • October 24, 2006 • 175(9) | 1072

 Research

For the primary outcome, we performed logistic regression Health Statistics for that purpose. The final model included
analysis specific for complex survey designs21 that accounted variables that were associated with the outcomes at p < 0.05.
for the clusters (primary sampling units), strata (provinces) The presence of each of hypertension, diabetes and raised
and data weighted to the general population of Pakistan in serum cholesterol concentration was then entered into the fi-
1990, with weights calculated by the Pakistan Federal Bureau nal model to explore the independent association of BMI with
of Statistics and confirmed by the US National Center for these conditions.

Table 3: Association between overweight and obesity (BMI ≥ 23 kg/m2) and sociodemographic factors in the surveyed
Pakistani population aged 15 years or more

No. (%) overweight Unadjusted Adjusted

Factor Total no.* or obese* OR (95% CI) OR (95% CI)†

Age, yr, mean (SD)‡ 36.8 (17.3) 39.6 (15.3) 1.08 (1.06–1.10) 1.11 (1.09–1.13)
Ethnicity§ NS
Muhajir 1606 554 (34.5) 1.63 (1.17–1.28)
Punjabi 3536 859 (24.3) 1.00
Baluch 282 85 (30.1) 1.05 (0.68–1.64)
Sindhi 1425 293 (20.6) 0.96 (0.61–1.50)
Pashtun 1072 292 (27.2) 1.07 (0.69–1.65)
Other 1051 232 (22.1) 0.90 (0.61–1.34)
Male 4414 1019 (23.1) 0.73 (0.65–0.81) 0.66 (0.58–0.75)
Female 4558 1296 (28.4) 1.00 1.00
Urban residence 3239 1216 (37.5) 2.36 (1.76–3.16) 2.20 (1.62–2.99)
Rural residence 5733 1099 (19.2) 1.00 1.00
Literate (able to read) 2943 845 (28.7) 1.25 (1.06–1.48) 1.25 (1.09–1.44)
Illiterate (unable to read) 6029 1470 (24.4) 1.00 1.00
Economic status¶
High 1607 492 (30.6) 1.60 (1.08–2.39) 1.56 (1.06–2.26)
Middle 4396 1206 (27.4) 1.36 (1.06–1.74) 1.14 (0.88–1.46)
Low 2969 617 (20.8) 1.00 1.00
Food intake**
Meat, high 2719 994 (36.6) 2.13 (1.71–2.65) 1.65 (1.37-1.98)
Meat, low 6253 1321 (21.1) 1.00 1.00
Fruit, high 3224 972 (30.2) 1.34 (1.08–1.65) NS
Fruit, low 5748 1343 (23.4) 1.00
Milk, high 4431 1027 (23.2) 0.78 (0.66–0.92) NS
Milk, low 4541 1288 (28.4) 1.00
Eggs, high 1434 451 (31.5) 1.44 (1.17–1.77) NS
Eggs, low 7538 1864 (24.7) 1.00
Rice, high 2220 620 (27.9) 1.21 (0.95–1.55) NS
Rice, low 6752 1695 (25.1) 1.00
Potatoes, high 4672 1252 (26.8) 1.14 (0.90–1.44) NS
Potatoes, low 4300 1063 (24.7) 1.00
Use of cigarettes or “beddies”
Current 1412 328 (23.2) 0.82 (0.69–0.96) NS
Not current 7560 1987 (26.3) 1.00
Use of chewable tobacco or
snuff
Current 1003 233 (23.2) 0.83 (0.64–1.07) NS
Not current 7969 2082 (26.1) 1.00

Note: OR = odds ratio, CI = confidence interval, SD = standard deviation, NS = not significantly associated with overweight in the multivariable model
and therefore not included in the final model.
*Unless stated otherwise.
†Adjusted for age, sex, urban v. rural dwelling, literacy status, economic status and high v. low intake of meat.
‡Odds ratios are for each 5-year increase.
§Reported as “mother tongue,” which is specific for each of the 5 major ethnic subgroups of Pakistan: Muhajir, Punjabi, Baluch, Sindhi and Pashtun.
¶The 3 levels were defined through a simple count of the number of items owned.
**Defined as high if the particular type of food was consumed at least every other day.

CMAJ • October 24, 2006 • 175(9) | 1073

 Research

The sensitivity and specificity of the BMI for an association sensitivity of 56% and 59%, respectively, and a specificity of
with hypertension or diabetes were expressed by receiver op- 72% in both sexes. A cutoff value of 23 kg/m2 had a sensitivity
erating characteristic (ROC) curves. of 46% and 59% and a specificity of 78% and 73% in men and
women, respectively.
Results
Interpretation
Of the 9442 subjects aged 15 years or more, 470 women re-
ported being pregnant and were excluded from the analysis. This nationally representative survey showed 25.0% of the
Thus, our final sample was 8972 subjects. Pakistani population to be overweight or obese according to
The overall prevalence of overweight and obesity, weight- the Asian-specific BMI cutoff value of 23 kg/m2 and 10.3% to
ed to the general Pakistani population, was 25.0% (95% CI be obese according to the BMI cutoff value of 27 kg/m2. Our
21.8–28.2%). The prevalence was highest, 42.8%, among data confirm a major public health problem in Pakistan.
women aged 35–54 years but was also high among those Discussions around the global epidemic of obesity have
aged 15–24 years, at 12.4% for men and 13.8% for women often used the future tense for the developing world. We
(Table 1). Table 2 shows the weighted prevalence of over- have shown that current rates of overweight and obesity are
weight and obesity according to various BMI cutoff values. already unacceptably high among youths. This is of consid-
The overall weighted prevalence of obesity according to our erable concern for a number of reasons. Obesity tends to
study definition was 10.3% (95% CI 7.0%–13.2%). track within individuals and populations: obese children be-
The factors independently and significantly associated come obese adults. This tendency, combined with the con-
with being overweight or obese included greater age, being tinued trend toward urbanization, will serve to seriously es-
female, urban residence, being literate, and having a high calate adult levels of obesity: we observed a 2.5 times greater
(v. low) economic status and a high (v. low) intake of meat prevalence of obesity among urban residents than among ru-
(Table 3). ral residents. In addition, there are indications that obesity in
Being overweight or obese was independently associated youth coupled with low birth weight is the worst possible
with having hypertension, diabetes and a raised serum cho- combination for adult cardiovascular disease and diabetes,
lesterol concentration (Table 4). conditions to which Indo-Asian populations are already par-
The ROC curves (Fig. 1) illustrate the sensitivity and speci- ticularly susceptible.9
ficity of the BMI for an association with hypertension or dia- Critics have challenged the revised BMI thresholds for
betes. For hypertension, the optimal cutoff value for the BMI defining overweight in subpopulations on the grounds that
was 21.2 kg/m2 in both men and women, yielding a sensitivity these thresholds may not necessarily be associated with ad-
of 61% and 65% and a specificity of 68% and 63%, respec- verse clinical outcomes.22,23 However, data for Chinese sub-
tively. A cutoff value of 23 kg/m2 had a sensitivity of 43% and jects aged 40 years or over revealed a U-shaped association
50% and a specificity of 82% and 76% in men and women, re- of BMI with 10-year all-cause mortality rates, with the least
spectively. For diabetes, the optimal cutoff value for the BMI risk of death at a BMI of less than 21 kg/m2 among men and
was 22.1 kg/m2 in men and 22.9 kg/m2 in women, yielding a less than 22 kg/m2 among women.24 We found an inde-

Table 4: Association of overweight and obesity (BMI ≥ 23 kg/m2) with clinical characteristics in the surveyed
Pakistani population aged 15 years or more

No. (%) Unadjusted OR Adjusted OR

Clinical characteristic Total no.* overweight* (95% CI) (95% CI)†

Hypertension‡ 1757 814 (46.3) 3.20 (2.74–3.74) 2.32 (2.00–2.69)

No hypertension 7215 1501 (20.8) 1.00 1.00
Diabetes mellitus§ 482 245 (50.8) 3.05 ( 2.42–3.86) 2.10 (1.66–2.65)
No diabetes 8490 2070 (24.4) 1.00 1.00
Proteinuria¶ 360 111 (30.8) 1.22 (0.88–1.70) NS
No proteinuria 7388 1946 (26.3) 1.00
Raised serum cholesterol
concentration,** mg/dL, mean (SD) 154.0 (38.1) 168.3 (40.8) 1.06 (1.05–1.08) 1.04 (1.03–1.05)

Note: OR = odds ratio, CI = confidence interval, SD = standard deviation, NS = not significantly associated with overweight in the multivariable
model and therefore not included in the final model.
*Unless stated otherwise.
†Based on the multivariable model. Adjusted for age, sex, urban v. rural dwelling, literacy status, economic status and high v. low intake of meat.
‡Defined as having a systolic blood pressure of 140 mm Hg or greater or a diastolic blood pressure of 90 mm Hg or greater (based on the mean
of the 2 readings) or currently receiving therapy with antihypertensive medication.
§Defined as having a nonfasting blood glucose concentration of 140 mg/dL (7.8 mmol/L) or greater or a history of diabetes mellitus.
¶Defined as a urine protein concentration of 30 mg/dL or greater as measured by a reagent strip (reading of 1+ or greater). Data on urine
protein levels were not available for 1224 people.
**For each increase of 5.0 mg/dL (0.13 mmol/L).

CMAJ • October 24, 2006 • 175(9) | 1074

 Research

A Hypertension B Diabetes
BMI — Men BMI — Men
100 100

80 80
Sensitivity

Sensitivity
60 60

40 40

20 20

0 0
0 20 40 60 80 100 0 20 40 60 80 100
100 — Specificity 100 — Specificity

BMI — Women BMI — Women

100 100

80 80
Sensitivity

Sensitivity

60 60

40 40

20 20

0 0
0 20 40 60 80 100 0 20 40 60 80 100
100 — Specificity 100 — Specificity

Fig 1: Receiver operating characteristic (ROC) curves for the sensitivity and specificity of the association between body mass index
(BMI) and hypertension (A) and diabetes mellitus (B). For hypertension in men, the area under the curve (AUC) was 0.68 (95% confi-
dence interval [CI] 0.66–0.69) and the optimal BMI cutoff value 21.2 kg/m2 (triangle); this BMI value had a sensitivity of 61% and a
specificity of 68%. The sensitivity and specificity were 43% and 82%, respectively, for a BMI cutoff value of 23 kg/m2 and 28% and 90%,
respectively, for one of 25 kg/m2. For hypertension in women, the AUC was 0.66 (95% CI 0.65–0.68) and the optimal BMI cutoff value
21.2 kg/m2 (triangle); this value had a sensitivity of 65% and a specificity of 63%. The sensitivity and specificity were 50% and 76%,
respectively, for a BMI cutoff value of 23 kg/m2 and 37% and 85%, respectively, for one of 25 kg/m2. For diabetes in men, the AUC was
0.64 (95% CI 0.63–0.66) and the optimal BMI cutoff value 22.1 kg/m2 (triangle); this value had a sensitivity of 56% and a specificity of
72%. The sensitivity and specificity were 46% and 78%, respectively, for a BMI cutoff value of 23 kg/m2 and 29% and 88%, respectively,
for one of 25 kg/m2. For diabetes in women, the AUC was 0.66 (95% CI 0.65–0.68) and the optimal BMI cutoff value 22.9 kg/m2 (trian-
gle); this value had a sensitivity of 59% and a specificity of 72%. The sensitivity and specificity were 59% and 73%, respectively, for a
BMI cutoff value of 23 kg/m2 and 42% and 82%, respectively, for one of 25 kg/m2.

CMAJ • October 24, 2006 • 175(9) | 1075

 Research

pendent association between a BMI of 23 kg/m2 or greater ulation. Immediate efforts are needed at a national level to
and the presence of hypertension, diabetes or a raised control this problem in Pakistan and possibly the neighbour-
serum cholesterol concentration. Further, the ROC curves ing developing countries.
showed that even lower cutoff values than the recom-
mended 23 kg/m2 (21.2 and 22.9 kg/m2 in women and 21.2 This article has been peer reviewed.
and 22.1 kg/m2 in men), albeit unadjusted for other factors, From the Clinical Epidemiology Unit, Department of Community Health Sci-
yielded the optimal areas under the curve for the associa- ences (Jafar, Pappas), and the Section of Nephrology, Department of Medi-
tions with hypertension and diabetes, respectively. Using a cine (Jafar), Aga Khan University, Karachi, Pakistan; the Division of Nephrol-
conventional cutoff value of 25 kg/m2 resulted in poor sen- ogy, Department of Medicine, New England Medical Center, Tufts University
Medical School, Boston, Mass. (Jafar); and the National Heart and Lung In-
sitivity for both associations. Our findings support the use stitute, Imperial College, London, UK (Chaturvedi)
of a lower cutoff value of BMI (23 kg/m2 or even lower) for
identification of subjects with hypertension and diabetes in Competing interests: None declared.
programs screening for these chronic diseases in Indo- Contributors: Tazeen Jafar was the principal author and contributed to the
Asian populations. study concept and design and to the analysis and interpretation of data. Nish
Our analyses had limitations. Clinical diagnosis of hyper- Chaturvedi contributed to the study concept and the interpretation of data.
Gregory Pappas contributed to the acquisition of data. All of the authors re-
tension requires persistent elevation of blood pressure on re- vised the article critically for important intellectual content and gave final ap-
peated visits. However, the mean of multiple readings on the proval of the version to be published.
same day (2 readings in the National Health Survey of Pak- Acknowledgements: We thank members of the Pakistan Medical Research
istan) is considered acceptable for epidemiologic studies and Council and the US Department of Health and Human Services for their
has been used to diagnose hypertension in other surveys.25,26 assistance in the acquisition of data from the National Health Survey of Pakistan.
In the National Health Survey of Pakistan, blood was drawn This study was supported by a grant from the Fogarty International Cen-
ter, US National Institutes of Health.
without a requirement of fasting. Diabetes was defined as a
blood glucose concentration of 140 mg/dL (7.8 mmol/L) or
greater or a history of diabetes. This definition diverges from
the standard criterion of a fasting blood glucose concentration REFERENCES
1. Friedrich MJ. Epidemic of obesity expands its spread to developing countries.
of more than 126 mg/dL (7.0 mmol/L).19 However, our main JAMA 2002;287:1382-6.
findings were unchanged when we performed a sensitivity 2. Galal OM, Hulett J. Obesity among schoolchildren in developing countries. Food
Nutr Bull 2005;26:S261-6.
analysis using the older blood glucose criterion of 200 mg/dL 3. Reddy KS, Yusuf S. Emerging epidemic of cardiovascular disease in developing
or greater.28 Thus, we believe that these findings are robust. countries. Circulation 1998;97:596-601.
We acknowledge that some dimensions of socioeconomic sta- 4. Reddy KS. Cardiovascular disease in non-Western countries. N Engl J Med 2004;
350:2438-40.
tus may not have been adequately captured, the food-fre- 5. He J, Gu D, Wu X, et al. Major causes of death among men and women in China.
quency questionnaire was not validated, and information on N Engl J Med 2005;353:1124-34.
6. Jafar TH, Jafary FH, Jessani S, et al. Heart disease epidemic in Pakistan: women
physical activity was not collected. In addition, our associa- and men at equal risk. Am Heart J 2005;150:221-6.
tions between obesity and health conditions were cross-sec- 7. Dhurandhar NV, Kulkarni PR. Prevalence of obesity in Bombay. Int J Obes Relat
Metab Disord 1992;16:367-75.
tional. However, there have been few high-quality cohort stud- 8. Deurenberg-Yap M, Schmidt G, van Staveren WA, et al. The paradox of low body
ies in the developing world that have assessed associations mass index and high body fat percentage among Chinese, Malays and Indians in
Singapore. Int J Obes Relat Metab Disord 2000;24:1011-7.
between obesity and incident disease, and we believe that re- 9. Bhargava SK, Sachdev HS, Fall CH, et al. Relation of serial changes in childhood
verse causality is not a major problem for conditions that are body-mass index to impaired glucose tolerance in young adulthood. N Engl J Med
2004;350:865-75.
largely asymptomatic. Moreover, the main objective of our 10. Choo V. WHO reassesses appropriate body-mass index for Asian populations.
analyses was to determine the nationally representative preva- Lancet 2002;360:235.
lence of being overweight. Although we also identified some 11. World Health Organization, Western Pacific Region. The International Associa-
tion for the Study of Obesity and the International Obesity Task Force. The
sociodemographic factors associated with being overweight, Asia–Pacific perspective: redefining obesity and its treatment. Sydney, Australia:
further, in-depth trials are needed to study the impact of Health Communications Australia Pty Limited; 2000. Available: www.diabetes
.com.au/pdf/obesity_report.pdf (accessed 2006 Aug 23).
strategies aimed at behaviour change that would target these 12. Pakistan Medical Research Council. National Health Survey of Pakistan 1990–1994.
factors as well as those identified previously in an attempt to Islamabad, Pakistan: Pakistan Medical Research Council;1998:50.
13. Qureshi AA, Wajid G, Shaikh IA, et al. Ethical considerations for human investiga-
reduce the prevalence of being overweight or having central tion in the National Health Survey of Pakistan. Pakistan J Med Res 1992;31:270-4.
obesity in Pakistan.27 The sound epidemiologic design, high 14. Mallick MD. Sample design for the National Health Survey of Pakistan. Pakistan J
Med Res 1992;31(Suppl):289-90.
response rate and use of the modified Asian-specific defini- 15. Price CP, Koller PU. A multicentre study of the new Reflotron system for the meas-
tion of being overweight or obese are major strengths of our urement of urea, glucose, triacylglycerols, cholesterol, gamma-glutamyltrans-
findings, which are generalizable. ferase and haemoglobin. J Clin Chem Clin Biochem 1988;26:233-50.
16. Pappas G, Akhtar T, Gergen PJ, et al. Health status of the Pakistani population: a health
In conclusion, our study showed that 1 in 4 people in Pak- profile and comparison with the United States. Am J Public Health 2001;91:93-8.
istan over the age of 15 years is overweight or obese. We have 17. Durkin MS, Islam S, Hasan ZM, et al. Measures of socioeconomic status for child
health research: comparative results from Bangladesh and Pakistan. Soc Sci Med
demonstrated alarming rates of obesity among youths and 1994;38:1289-97.
women, indicating where, in particular, interventions should 18. Rolka DB, Narayan KM, Thompson TJ, et al. Performance of recommended
screening tests for undiagnosed diabetes and dysglycemia. Diabetes Care 2001;
be targeted. Owing to the better association of hypertension 24:1899-903.
and diabetes with an even lower BMI cutoff value (less than 19. Kuzuya T. Early diagnosis, early treatment and the new diagnostic criteria of dia-
betes mellitus. Br J Nutr 2000;84(Suppl 2):S177-81.
23 kg/m2), our findings support the use of Asian-specific 20. Jafar TH, Chaturvedi N, Gul A, et al. Ethnic differences and determinants of pro-
thresholds for the definition of being overweight in this pop- teinuria among South Asian subgroups in Pakistan. Kidney Int 2003;64:1437-44.

CMAJ • October 24, 2006 • 175(9) | 1076

 Research

21. Hosmer D, Lemeshow S. Applied logistic regression. Hoboken, New Jersey: Wiley-
Interscience, John Wiley & Sons; 1980. p. 1–134.
22. WHO Expert Consultation. Appropriate body-mass index for Asian populations and Editor’s take
its implications for policy and intervention strategies. Lancet 2004;363:157-63.
23. Oh SW, Shin SA, Yun YH, et al. Cut-off point of BMI and obesity-related comor- • The rising prevalence of obesity in industrialized countries is
bidities and mortality in middle-aged Koreans. Obes Res 2004;12:2031-40.
24. Gu D, He J, Duan X, et al. Body weight and mortality among men and women in now well documented. It is generally known that people or
China. JAMA 2006;295:776-83. societies that move from famine to feast go through a time,
25. Hypertension control. Report of a WHO Expert Committee [Technical Report Se- often a generation, of overeating.
ries, no. 862]. Geneva: WHO; 1996.
26. Hajjar I, Kotchen TA. Trends in prevalence, awareness, treatment, and control of
hypertension in the United States, 1988–2000. JAMA 2003;290:199-206. • The authors, in reviewing population survey data, estimate
27. Jafar TH, Levey AS, White FM, et al. Ethnic differences and determinants of diabetes
and central obesity among South Asians of Pakistan. Diabet Med 2004;21:716-23.
that Pakistan has a high prevalence of overweight and obe-
28. Lebovitz HE. Diagnosis, classification, and pathogenesis of diabetes mellitus. [dis- sity and find that the trend is similar to that already noted in
cussion 40-1]. J Clin Psychiatry 2001;62(Suppl 27):5-9. China and other South Asian countries.

• The concern over rates of overweight and obesity in industri-

Correspondence to: Dr. Tazeen H. Jafar, Director, Clinical alized countries ought to include the entire globe. In
Epidemiology Unit, and Head, Section of Nephrology, Canada, many people are recent immigrants who came from
Medicine and Community Health Sciences, Aga Khan University, a context of inadequacy to one of plenty. They likely face the
PO Box 3500, Stadium Road, Karachi, Pakistan; fax 92-21-4934294; same risk of obesity and all its complications.
tazeen.jafar@aku.edu

The complete
Publications Mail Agreement no. 41190522; PAP registration no. 9848. USPS #0762-530. USPS periodical postage paid at Champlain, NY. Return undeliverable Canadian addresses to CMA Member Service Centre, 1867 Alta Vista Dr., Ottawa ON K1G 3Y6

CMAJ•JAMC
picture
October 24, 2006, Vol. 175, No. 9 • Le 24 octobre 2006, vol. 175, no 9

on research.
PRACTICAL. RELEVANT. CMAJ IS NOW
MORE COMPREHENSIVE THAN EVER. Deep-vein thrombosis
Recent advances in diagnosis and treatment

CMAJ’s reputation and wide Canadian and international RESEARCH

Obesity in Indo-Asians: Which
RESEARCH
Clues to diagnosing
ANALYSIS
Opinions in the aftermath

reach make it THE place to publish leading Canadian BMI cutoffs should be used? angioedema without urticaria of Chaoulli v. Quebec

research. CMAJ publishes important peer-reviewed www.cmaj.ca

research within weeks of submission and ranks fifth among

international general medical journals by impact factor,
a measure of the scientific importance of a journal.

The essential read.™

CMAJ • October 24, 2006 • 175(9) | 1077