Articles

Tobacco use and risk of myocardial infarction in

52 countries in the INTERHEART study: a case-control study
Koon K Teo, Stephanie Ounpuu, Steven Hawken, MR Pandey, Vicent Valentin, David Hunt, Rafael Diaz, Wafa Rashed, Rosario Freeman,
Lixin Jiang, Xiaofei Zhang, Salim Yusuf, on behalf of the INTERHEART Study Investigators

Summary
Background Tobacco use is one of the major avoidable causes of cardiovascular diseases. We aimed to assess the risks
associated with tobacco use (both smoking and non-smoking) and second hand tobacco smoke (SHS) worldwide.
Methods We did a standardised case-control study of acute myocardial infarction (AMI) with 27 089 participants in
52 countries (12 461 cases, 14 637 controls). We assessed relation between risk of AMI and current or former
smoking, type of tobacco, amount smoked, eect of smokeless tobacco, and exposure to SHS. We controlled for
confounders such as dierences in lifestyles between smokers and non-smokers.
Findings Current smoking was associated with a greater risk of non-fatal AMI (odds ratio [OR] 295, 95% CI
277314, p<00001) compared with never smoking; risk increased by 56% for every additional cigarette smoked.
The OR associated with former smoking fell to 187 (95% CI 155224) within 3 years of quitting. A residual
excess risk remained 20 or more years after quitting (122, 109137). Exclusion of individuals exposed to SHS in
the never smoker reference group raised the risk in former smokers by about 10%. Smoking beedies alone
(indigenous to South Asia) was associated with increased risk (289, 211396) similar to that associated with
cigarette smoking. Chewing tobacco alone was associated with OR 223 (141352), and smokers who also chewed
tobacco had the highest increase in risk (409, 298561). SHS was associated with a graded increase in risk
related to exposure; OR was 124 (117132) in individuals who were least exposed (17 h per week) and 162
(145181) in people who were most exposed (>21 h per week). Young male current smokers had the highest
population attributable risk (583%; 95% CI 550616) and older women the lowest (62%, 4192). Population
attributable risk for exposure to SHS for more than 1 h per week in never smokers was 154% (121193).
Conclusion Tobacco use is one of the most important causes of AMI globally, especially in men. All forms of tobacco
use, including dierent types of smoking and chewing tobacco and inhalation of SHS, should be discouraged to
prevent cardiovascular diseases.

Introduction
Tobacco use is one of the most important avoidable
causes of cardiovascular diseases worldwide.1 The
number of smokers worldwide is currently estimated to
be 13 billion, of which 82% are in developing countries.2
During the 20th century, 100 million individuals died
worldwide as a result of tobacco-related diseases.3 This
number is expected to increase to 1 billion during the
21st century.4 About half of these deaths will occur
among middle-aged adults (3569 years old), who will
lose on average 22 years of life.5 Most tobacco-related
deaths occur among men, but female mortality from
tobacco is expected to increase substantially as a result
of large increases in smoking among women in many
developed countries, and high rates of use of
non-smoking tobacco, especially in women, in several
developing countries.2
The risk of coronary heart disease associated with
smoking has been documented in studies in developed
countries,68 few large studies have been done to
examine the eects of tobacco in other geographical
regions. The available studies are dicult to compare,
and extrapolations from studies in developed countries
to other regions of the world might not be appropriate
www.thelancet.com Vol 368 August 19, 2006

because of varying methods and markedly dierent

patterns of tobacco use (eg, the type of smoking varies
in dierent countries: cigarettes or beedies, smoking or
chewing, dierent ages at starting the habit, numbers
smoked).916 Moreover, the few studies that have been
done in developing countries include relatively small
numbers of cases, so that there is uncertainty about the
magnitude of the risk associated with smoking in these
countries.
Emerging data suggest that second hand smoke
(SHS) is associated with adverse health eects,
including coronary heart disease,17,18 but the available
studies recorded relatively few events, and might not
have fully adjusted for other lifestyle factors such as
diet. However, some of the research that suggested a
failure to adjust for such confounders has itself been
challenged on both methodological grounds and
undisclosed links to the tobacco industry.19 Therefore,
considerable uncertainty exists about the size of the
eect of SHS. We undertook a study to document the
risks associated with various forms of active tobacco
use (both smoking and non-smoking) and SHS in all
regions of the world, controlling for potential
confounders.

Lancet 2006; 368: 64758

See Comment page 621
Population Health Research
Institute, McMaster
University-Hamilton Health
Sciences, Hamilton, ON
L8L 2X2, Canada
(Prof K K Teo MB, S Ounpuu PhD,
S Hawken MSc,
Prof S Yusuf DPhil); Nepal
Health Research Council,
Kathmandu, Nepal
(M R Pandey MB); Hospital
Universitari Dr Peset, Valencia,
Spain (V Valentin MD);
University of Melbourne, Royal
Melbourne Hospital,
Melbourne, Australia
(D Hunt MD); Department of
Cardiology, Instituto
Cardiovascular de Rosario,
Rosario, Argentina (R Diaz MD);
Kuwait University, Kuwait City,
Kuwait (W Rashed MD);
University of Washington,
Seattle, USA (R Freeman MD);
and Cardiovascular Institute,
Beijing, China (L Jiang MD,
X Zhang MD)
Correspondence to:
Prof Salim Yusuf
[email protected]

647

Articles

Methods
Participants
INTERHEART was a standardised case-control study of
15 152 cases of rst acute myocardial infarction (AMI)
and 14 820 age-matched and sex-matched controls who
were recruited from 262 centers in 52 countries in Asia,
Europe, Middle East Crescent, Africa, Australia, North
and South America. Details have been published
previously.20,21 Consecutive cases of rst AMI presenting
within 24 h of symptom onset were eligible. We included
all consenting cases without cardiogenic shock or history
of major chronic diseases. At least one age-matched (plus
or minus 5 years) and sex-matched control (without a
history of heart disease or exertional chest pain) was
recruited per case by use of specic criteria. A
community-based control was either a visitor or relative
of a patient from a non-cardiac ward or an unrelated
visitor of another cardiac patient. A hospital-based control
was selected from those at the same centre with illnesses
not obviously related to coronary heart disease or its risk
factors. Of the cases, 1531 were excluded because they
had unstable angina alone and were recruited into a
substudy, 205 did not have a conrmed AMI, 695 had a
previous AMI, and 260 had insucient data. Of the
controls, 74 were excluded because of insucient data,
and 109 had a previous history of AMI. Therefore,
27 098 participants (12 461 cases and 14 637 controls) were
available for the study; among these individuals, data on
tobacco use was missing in 530, so that complete data
were available in 12 133 cases and 14 435 controls.

Procedures
Trained sta administered a structured questionnaire
and did physical examinations for cases and controls in
the same manner. Participants were asked if they
regularly used any of the following tobacco products:
cigarettes, beedies, pipes or cigars, chewing tobacco,
paan, snu, sheesha or water pipe, and other forms of

smoked or non-smoked tobacco. Beedies consist of a small

amount of tobacco wrapped in a dried temburini leaf and
tied with a string. Paan is a form of tobacco preparation
that is used with lime, with areca nut or a betel quid. For
cigarettes (and beedies), the number smoked per day, the
type of cigarettes (lter, nonlter, or both), and the brands
of cigarettes commonly smoked were recorded. Categories
of tobacco use were dened as follows: current smokers
were individuals who smoked any tobacco (including
beedies, pipes, and other forms) in the previous 12 months,
and included those who had quit within the past year.
Former smokers had quit more than a year earlier. Never
smokers were those who responded that they had never
used tobacco products regularly. Regular users were
individuals who used at least one tobacco product daily.
Exposure to SHS was recorded by asking about the
smoking habits of family members, friends, or co-workers,
whether these individuals smoked regularly in the
participants presence, the number of times per day that
SHS exposure exceeded 5 consecutive minutes, the
average number of hours per week of exposure over the
past 12 months, and smoking habits of the spouse. Height,
weight, waist and hip circumferences, blood pressure, and
heart rate were measured with a standardised protocol.
Concentrations of apolipoproteins B and A1 in serum
were measured with standardised approaches. Information
on dietary patterns, physical activity, alcohol consumption,
education, income, psychosocial factors, personal and
family history of cardiovascular disease, and risk factors
(hypertension, diabetes mellitus) were recorded.
All data were transferred to the Population Health
Research Institute, McMaster University and Hamilton
Health Sciences, Canada, where extensive data checks
were done. The study protocol was approved by appropriate
regulatory and ethics committees in all participating
countries and centres. All participants provided informed
consent before participating in the study.

Statistical methods
Number of individuals
Female
Mean age (SD)

Cases

Controls

12 461

14 637

3005 (24%)
581 (122)

3786 (26%)
569 (122)

Current smokers

452%

Diabetes

185%

268%
75%

Hypertension

390%

219%

Daily consumption of vegetables and fruits

358%

424%

Daily exercise

143%

193%

Mean body-mass index, kg/m2 (SD)

261 (42)

Mean waist-hip ratio (SD)

093 (008)

091 (008)

Median ApoB, g/L (IQR)

095 (078113)

090 (074107)

Median ApoA1, g/L (IQR)

110 (096136)

119 (103137)

Median ApoB/ApoA1 ratio (IQR)

087 (070105)

075 (060093)

Apo=apolipoprotein.

Table 1: Demographics and prevalence of risk factors in cases and controls

648

258 (42)

Details of statistical analysis have been described in

previously published reports.20,21 Univariate associations
were investigated with frequency tables. For comparisons
of prevalence of tobacco use across subgroups (eg, by
region or ethnicity), the potential dierences in age
structure of the populations were accounted for by direct
standardisation of the frequencies to the overall
INTERHEART age distribution with a ve-level age
stratication (<45, 4555, 5665, 6670, >70 years).
Continuous variables were summarised by means or
medians. For comparison of means across subgroups,
values were adjusted for age and sex with analysis of
covariance (ANCOVA) models. Sex-specic quintile
values in controls were used to categorise continuous
variables. Unconditional logistic regression with
adjustment for matching factors (ie, age, sex, and
geographic region included in all models as block eects)
was used to control for confounding by other risk factors.
www.thelancet.com Vol 368 August 19, 2006

Articles

Ever

Former

Current
Any

Female

Male

Female

Male

Female

19 cig/day

1019 cig/day

20 cig/day

Male

Female

Male

Female

Male

Female

Male

Overall
Young

219

573

108

160

111

413

46

119

36

119

28

176

Older

192

579

127

299

65

280

26

70

21

92

18

118

North America
Young

635

518

405

357

230

161

108

71

81

18

41

71

Older

407

800

391

628

16

172

00

41

00

21

16

110

Young

471

567

240

189

231

378

77

111

96

100

58

167

Older

288

651

173

431

115

220

43

65

50

69

22

86

Western Europe

Australia/New Zealand
Young

424

539

289

313

135

226

39

35

29

35

67

157

Older

360

641

288

544

72

97

36

12

09

26

27

60

Eastern/central Europe
Young

350

714

168

188

182

526

42

38

64

121

76

367

Older

182

656

101

331

81

325

29

33

17

120

35

172

Middle East
Young

44

538

10

154

34

384

19

92

10

69

05

224

Older

00

645

00

173

00

372

00

77

00

81

00

214

Africa
Young

388

677

116

148

272

529

192

335

54

136

27

58

Older

298

652

213

271

85

381

43

194

43

93

00

93

South Asia
Young

48

451

09

128

39

323

09

156

30

106

00

62

Older

34

479

17

221

17

258

17

100

00

104

00

54

Young

55

647

16

59

39

588

10

70

18

224

10

294

Older

113

499

52

169

61

330

14

58

33

124

14

148

China/Hong Kong

Japan/southeast Asia
Young

83

655

59

210

24

445

06

123

06

154

12

169

Older

162

709

103

358

59

351

59

77

00

116

00

158

Young

282

540

180

267

102

273

49

135

32

74

21

64

Older

216

576

168

410

48

166

29

77

10

47

10

43

Latin America

Young: female 65 years, male 55 years. Older: female >65 years, male >55 years. cig/day=cigaretters per day.

Table 2: Percentages of ever (former+current), former, and current smokers in controls by region, sex, and age

Results from unconditional analyses were similar to

those from conditional and mixed eect models analyses
(<5% variation). Analyses adjusted for the other
modiable INTERHEART risk factors (apolipoprotein
B/apolipoprotein A ratio, obesity, history of hypertension,
history of diabetes, dietary pattern, activity, and alcohol
use) are also presented.
Population attributable risks (PAR) and their 95% CI
were calculated by a method based on unconditional
logistic regression using the methods of Benichou and
Gail,22,23 and with the Interactive Risk Attributable
Program by the US National Cancer Institute, 2002.24 The
PAR presented are adjusted for confounders in a similar
fashion to the corresponding logistic regression models
for odds ratio (OR) estimates and, where indicated,
www.thelancet.com Vol 368 August 19, 2006

stratied by subgroups of interest. The PAR calculation

uses the prevalence rates of risk factors reported in the
control group, as well as the estimate of relative risk. In
the analysis of PAR in subgroups (eg, males vs females or
by age groups, etc) the actual prevalence and relative
risks observed in the specic subgroups were used and
are presented here.

Results
A previous report showed that abnormal lipid proles,
smoking, hypertension, diabetes, abdominal obesity,
psychosocial factors, consumption of fruits, vegetables,
and alcohol, and physical activity accounted for most of
the risk of AMI worldwide.21 The demographic, health
history and characteristics for the 27 098 participants
649

Articles

100

Current smoking
Former smoking

90
80

Prevalence (%)

70
60
50
40
30
20
10
0
F

N Am

W Eur

Aus/NZ

E/C Eur

M East

M
Afr

S Asia

Ch/HK

Jap/SEA

Latin Am

Figure 1: Prevalence of smoking by region and sex

F=female. M=male. N Am=North America. W Eur=western Europe. Aus/NZ=Australia and New Zealand. E/C Eur=eastern and central Europe. M East=Middle East.
Afr=Africa. S Asia=South Asia. Ch/HK=China and Hong Kong. Jap/SEA=Japan and southeast Asia. Latin Am=Latin America.

(12 461 cases and 14 637 controls) have been reported.21

The distribution of various risk factors between cases
and controls are summarised in table 1. Here, we focus
on the use of tobacco by participants in the study.
Smoking habits varied greatly among controls and
cases, between men and women, and in dierent regions
of the world. Table 2 summarises the smoking status in
controls overall and by region, sex and age. In control
women, nearly 80% never smoked and less than 10%
were current smokers. Younger (65 years of age) women
were more likely to have smoked at any time compared
with older (>65 years) women (219% vs 192%), and
were more likely to be current smokers (111% vs 65%).
Smoking patterns varied markedly by region, especially
10
9
8

OR (95% Cl)

7
6
5
4
3
2
1
0
Nev Form 119 20 Nev Form 119 20 Nev Form 119 20 Nev Form 119 20 Nev Form 119 20
Age <40 years

Age 4049 years

Age 5059 years

Age 6069 years

Age 70 years

Figure 2: Risk of AMI associated with numbers smoked, by age group

p for interaction < 00001. Nev=never smokers. Form=former smokers. 119=currently smoking 119 cigarettes
per day. 20=currently smoking 20 or more cigarettes per day.

650

in younger women. Prevalence of current smoking was

fairly high (>20%) among younger women in three
regions (North America, western Europe, and Africa),
very low (<5%) in Asia and the Middle East, and
intermediate (1020%) in Australia/New Zealand, eastern
and central Europe, and Latin America (gure 1, table 2).
Among older women, the prevalence of current smoking
was very low in most regions, with the highest rate
(115%) in Western Europe (table 2)
The pattern of smoking in female controls diered
substantially from that in male controls, among whom
about a third were current smokers with a tendency to
smoke heavily, and only about two fths never smoked.
The prevalence of ever smoking did not dier between
younger (55 years of age) and older (>55 years) male
controls (573% vs 579%), but current smoking was more
frequent in younger than in older men (413% vs 280%)
a pattern seen in every region except North America. The
highest rates of current smoking among young male
controls were reported in China and Hong Kong, eastern
and central Europe, and Africa (all >50%). The lowest
rates were in North America and Australia and New
Zealand (<25%), which were also regions with the highest
proportions of former smokers in both younger (>30%)
and older men (>50%). By contrast, in South Asia, the
Middle East, and China and Hong Kong, only 15% or less
of younger men and less than 25% of older men were
former smokers. (gure 1, table 2)
Overall, current smoking was associated with a
three-fold increase in odds of a non-fatal AMI, compared
with never smokers (odds ratio [OR] 295, 95% CI
277314, p<00001). The risks were higher depending
on the number of cigarettes smoked, with people who
smoked one to nine cigarettes per day having an OR of
163 (95% CI 145182, p<00001), 1019 per day an OR
www.thelancet.com Vol 368 August 19, 2006

Articles

www.thelancet.com Vol 368 August 19, 2006

OR (95% Cl)

1
075
Never

Filter

Non-lter

Beedies

Pipes

Chew

Chew and
smoke

Figure 3: Risk of AMI associated with type of tobacco used

OR for current smokers=295 (95% CI 277314) indicated by broken horizontal line. Never=never smokers.
Filter=lter cigarettes. Non-lter=non-lter cigarettes. Beedies=smoking beedies alone. Pipes=smoking pipes/
cigars. Chew=chewing tobacco alone. Chew and smoke=both chewing and smoking tobacco.

4
OR (95% Cl)

of 259 (235285, p<00001), and 20 or more per day

an OR of 459 (421500, p<00001). Similar trends
were noted in men and women. Women who smoked
119 cigarettes per day had an OR of 211 (177252,
p<00001) and 20 or more per day an OR of 511
(398656, p<00001) for AMI. For men, the
corresponding ORs were 206 (190224, p<00001)
and 448 (413487, p<00001), respectively.
The eect of current smoking was much larger in
younger (OR 353, 95% CI 323386) than in older
individuals (255, 2235276; p<00001 for interaction);
especially in heavy smokers (20 cigarettes per day) in
whom ORs were 560 (95% CI 51620) for younger and
360 (325398, p<00001 for interaction) for older
smokers. A consistent interaction between age and
smoking was noted in risk of AMI (gure 2), with groups
of smokers younger than 40 years and those aged
4049 years of age showing much higher levels of risk of
AMI associated with number of cigarettes smoked. The
excess risks associated with tobacco were similar between
smokers who used lter cigarettes (OR 292, 95% CI
273312) and non-lter cigarettes (235, 196282;
gure 3).
A clear dose-response relation existed between number
of cigarettes smoked per day and risk of AMI (gure 4).
Regression analysis of the number of cigarettes smoked
per day as a continuous variable showed that the odds of
developing AMI were increased by 1056 (95% CI
105106) for every additional cigarette smoked per day.
This risk did not change when exposure to SHS was
removed in the control never smokers. The odds were
nine-fold higher in those who smoked 40 or more
cigarettes a day (OR 916, 95% CI 6791236) than in
never smokers.
Compared with never smokers, former smokers had a
moderately higher risk (OR 149, 95% CI 139159).
There was an apparent dierence in risks between female
(104, 089122, p=062) and male former smokers
(162, 149175, p<00001; p=00002 for heterogeneity)
after adjusting for age, region, physical activities, and
consumption of fruits, vegetables, and alcohol. After
further adjustment for numbers of cigarettes smoked per
day before quitting, this interaction was only marginally
signicant (p=003), and therefore might be simply due
to chance. This apparent dierence in risk cannot be
readily explained.
The risk associated with smoking was much reduced
within a few years of quitting (table 3, gure 5). When
compared with control never smokers as the reference,
the OR for AMI for those who quit smoking 13 years
earlier was substantially lower than that observed among
current smokers after similar adjustment. Risk of AMI
fell progressively with time after smoking cessation but
even in people who had quit 20 or more years ago, there
was a residual excess risk of about 22%.
When these analyses were repeated after removing
individuals who had been exposed to SHS from the

1
075
Never

12

34

56

78

910 1112

1314 1516

1718

1920

21

Number smoked per day

Figure 4: Risk of AMI with increasing numbers of cigarettes smoked, compared with never smokers
21 cigarettes smoked per day represents about 15 pack of cigarettes per day, associated with OR 600700.

control reference never smokers group, the risks in

former smokers were higher by about 10%, with a similar
graded lowering in risks the longer the individuals had
quit smoking (table 3). When individuals exposed to SHS
were removed from both reference never smokers and
former smokers, the levels of risk in the former smokers
were lower by about 5% compared with the levels of risk
observed when there was no adjustment for SHS (table 3).
In these analyses, the risks for current smokers were not
aected substantially when adjusted for exposure to SHS
(table 3).
Among those who smoked 20 or more cigarettes per
day before quitting, the largest decrease in risk occurred
651

Articles

OR (95% CI)
Adjustment 1

Adjustment 2

Adjustment 3

Current smokers

304 (285325)

308 (286333)

283 (254315)

Quit 13 years

187 (155224)

193 (160232)

149 (109202)

Quit >35 years

157 (125197)

165 (132207)

151 (107212)

Quit >510 years

151 (129176)

163 (140191)

136 (109170)

Quit >1015 years

145 (125169)

160 (137187)

140 (113174)

Quit >1520 years

155 (130184)

169 (142202)

147 (115197)

Quit >20 years

122 (109137)

141 (125159)

131 (113151)

p for slope

<00001

<00001

<00001

Adjustment 1=adjusted for age, sex, region, physical activities and consumption of fruits, vegetables, and alcohol.
Adjustment 2=never smokers not exposed to SHS as reference group; adjusted for sex, region, physical activities, and
consumption of fruits, vegetables, and alcohol. Adjustment 3=included only reference group (never smokers) and
former smokers not exposed to SHS; adjusted for sex, region, physical activities, and consumption of fruits, vegetables,
and alcohol.

Table 3: Risk of AMI associated with various durations of cessation in former smokers

during the rst 3 years after quitting. Thereafter, the ORs

decreased more gradually. However, in this group, the
risk of AMI was still raised 20 or more years after quitting
(gure 5). In light smokers (<10 cigarettes per day), the
excess risk fell rapidly, with no apparent excess risk
3 years after quitting (gure 5). When levels of risk were
assessed at 1, 2, and 3 years after quitting, we found that
the risks remained high at 1 year, with a rapid lowering of
risk at 2 years after quitting.
The use of other forms of tobacco was largely conned
to specic regions. Smoking of beedies was reported in
59% of male controls in South Asia (Bangladesh, India,
and Sri Lanka). Of these, 51% also smoked cigarettes. In
western Europe and North America, 31% and 44%
Overall
19 cigs per day
1019 cigs per day
20 cigs per day

OR (95% Cl)

1
075

050
>13
Current

>35

>510

>1015

>1520

>20

Former smokers (years since cessation)

Figure 5: Diminishing risk of AMI associated with quitting in former smokers, with never smokers as
reference
ORs adjusted for sex, region, diet, alcohol, and physical activity.

652

male controls, respectively, reported smoking cigars,

pipes, or both; 67% and 73% of these individuals also
smoked cigarettes. In the Middle East, 04% of female
controls smoked sheesha (none smoked cigarettes), and
31% male controls smoked sheesha; 53% also smoked
cigarettes. In Iran, 23% female controls (no male
controls) smoked sheesha, but none of these women also
smoked cigarettes. Conversely, in Egypt, 82% men
smoked sheesha, with 56% also smoking cigarettes.
In South Asia, 47% of controls chewed tobacco
(34% women, 50% men). Of these, 40% (8% women
and 44% men) also smoked cigarettes. Chewing tobacco
was most frequently reported in India (46%; 23%
women, 51% men), Nepal (42%; 21%, 51%), Pakistan
(48%; 53%, 59%) and Sri Lanka (116%; 111%,
117%). In these countries, very few or none of the
women who chewed tobacco also smoked cigarettes or
beedies, whereas men who chewed tobacco often also
smoked cigarettes or beedies (ranging from 17% in India
to 86% in Sri Lanka). In the USA, 17% of male controls
both chewed tobacco and smoked cigarettes. The use of
paan in controls who also smoked cigarettes or beedies
was mostly reported in Bangladesh (73%; 45% women,
75% men), India (31%; 16%, 34%), Nepal (34%, only
in men), and Pakistan (59%; 77% women, 54% men).
All individuals who used paan without other forms of
chewing tobacco also smoked cigarettes or beedies.
The use of non-smoking tobacco, including chewing
tobacco and paan, was reported mainly from South Asia
(73% controls; of whom 40% also smoked cigarettes or
beedies). This high rate of use was observed in women
(55%; 5% also smoked) and men (76%; 44% also
smoked). Countries with the highest rates of use of
non-smoking tobacco included Bangladesh (76%; 87%
also smoked), India (64%; 24% smoked), Nepal (56%;
73% smoked), Pakistan (76%; 28% smoked), and Sri
Lanka (136%; 77% smoked). In these countries, very few
women who used non-smoking tobacco also smoked
cigarettes or beedies, whereas men chewed and smoked
tobacco frequently, ranging from 23% in India to 88% in
Sri Lanka. In Western Europe, 74% Scandinavian men
reported using non-smoking tobacco, mostly snu
(68%), and 35% of these individuals also smoked
cigarettes. The use of snu was reported in Africa, mainly
in Nigeria (in 68%, 100% also smoked cigarettes) and
Cameroon (156%, 57% also smoked cigarettes) but the
numbers of participants from these countries were
relatively small.
Signicant risks were associated with forms of tobacco
use other than cigarette smoking. Smoking beedies alone,
without cigarette smoking, was associated with an
age-adjusted and sex-adjusted OR of 289 for AMI (95%
CI 211396) compared with never smokers who did not
use other tobacco products. The risk for South Asia alone
did not dier (OR 273, 95% CI 190392). This risk was
similar to that associated with current cigarette smoking.
The risk was graded; the ORs for people who smoked one
www.thelancet.com Vol 368 August 19, 2006

Articles

www.thelancet.com Vol 368 August 19, 2006

Overall

Female Male Never smokers Former smokers Current

Current
119 cig/day >20 cig/day

Never

440

489

422

536

473

270

180

17 h

390

375

395

357

390

481

406
169

814 h

82

77

84

57

69

120

1521 h

36

23

40

21

27

55

94

>21 h

53

36

58

29

42

74

152

cig/day=cigaretters per day.

Table 4: Prevalence (%) of exposure to SHS (per week) in controls by sex and smoking status

In younger individuals (women aged <65 years and men

aged <55 years), the PAR was 454% (428479) and for
older individuals (women 65 years, men 55 years), it
was 308% (285331). The PAR for younger women
was 224% (194257) and for younger men was 583%
(550616). For older women, the PAR was 62%
(4192) and for older men, it was 379% (351408).
The PAR was much lower in women than in men,
reecting the lower prevalence of smoking in women,
even though they had the same odds of AMI as in men.
The PAR for current smokers ranged from a low of
331% (95% CI 297366) in China/Hong Kong to a
high of 533% (462603) in Australia/New Zealand.
The PAR for the other regions fell within this narrow
range, showing that the PAR for current tobacco smoking
is consistent and high across all the regions. When the
PAR for current smoking was examined by sex and age in
each of the regions, we noted that in men and in younger
smokers, and to some degree in older smokers, the PAR
was consistently high. In female current smokers, higher
PAR was seen in North America, Australia/New Zealand,
8
Overall, adjusted for smoking status
In never smokers

OR (95% Cl)

to nine beedies per day was 206 (95% CI 096442), for

1019 beedies was 251 (160394), and for 20 or more
beedies was 399 (237674). The risk of AMI associated
with pipe and cigar smoking, in the absence of smoking
cigarettes or beedies, was also high (330, 220496;
gure 3). Similarly, the use of sheesha was associated with
an excess risk (216, 106439). Participants who were
current smokers of cigarettes or beedies and who chewed
tobacco had an OR of 409 for AMI (95% CI 298561).
When adjusted for smoking status, the OR in all
participants who chewed tobacco was 157 (124200).
Individuals who chewed tobacco alone had an OR of 223
(141352) compared with those who never used tobacco
(gure 3). This eect of chewing tobacco was undiminished
after further adjustment for diabetes, abdominal obesity,
hypertension, exercise, and diet. Because of the small
number of users of paan and snu in this study, we were
unable to draw conclusions about the eects of these
forms of tobacco on AMI risk.
Overall, 44% of controls reported no exposure to SHS,
39% of controls reported exposures of between 1 to
7 hours/week, 82% exposed for 8 to 14 hours, 36% for
15 to 21 hours and 53% for 22 or more hours per week.
Slightly lower proportions of females reported exposures
to SHS than males (table 4). Half (536%) of controls
who were never smokers were also not exposed to SHS,
whereas slightly less than half of former smokers and
much fewer of the current smokers were not also exposed
to SHS. Current smokers reported much greater exposure
to SHS than never and former smokers (table 4)
Exposure to SHS increased the risk of non-fatal AMI
in a graded manner. For example, after adjusting for age,
sex, region, physical activity, and consumption of fruits,
vegetables, and alcohol, individuals with the lowest level
of exposure (17 h per week), compared with those who
had never used tobacco or been exposed to SHS, had an
OR of 124 (95% CI 117132). The levels of risk then
increased in a graded manner with increasing exposure
(gure 6). These patterns were most marked among
never smokers (gure 6) and in former smokers. The
risk due to SHS was least marked among heavy smokers
(20 or more cigarettes per day), with an OR of 105
(088126) for those with the lowest exposure and 130
(104162) at the highest level of exposure to SHS
(p=0128 for slope). About a fth of controls were
exposed to SHS from a spouses tobacco smoking, with
women more frequently exposed than men. The risk of
AMI associated with exposure to SHS due to spousal
smoking was signicant (OR 128, 95% CI 112147)
and did not dier from the overall risks due to SHS
exposure. The increased risk of AMI associated with
tobacco use was consistently noted in the presence or
absence of the other risk factors (table 5).
Current smoking, compared with never smoking, was
responsible for 376% (95% CI 359394) of the PAR of
AMI in the overall population. In women the PAR was
154% (134176) and in men, it was 463% (441485).

1
075

Never

17

814

1521

22

SHS exposure (h per week)

Figure 6: Risk of AMI associated with extent of exposure to SHS, adjusted for
smoking status in all individuals and in never smokers
Graded increase in risk occurs with increasing exposure. In never smokers, the
decreasing number of individuals who were exposed to higher levels of SHS
resulted in loss of robustness of the data at the highest level of exposure.

653

Articles

who were not exposed to SHS, the overall PAR was

154% (95% CI 121193); for women in this group, it
was 108% (62181) and for men, it was 186%
(144236). In the younger individuals in the group,
the PAR was 137% (86211) and for older people it
was 149% (110199). In younger women who were
never smokers but were exposed to SHS, PAR was 96%
(38224) and for younger men, it was 182%
(107290). The corresponding PAR for older women
was 108% (51214) and for older men was 170%
(123230).

OR (95% CI)
Current smokers

Former smokers

Overall

295 (277314)

149 (139159)

Diabetes

235 (192288)

130 (107158)

Hypertension

253 (224287)

131 (160148)

No regular fruits or vegetables

295 (253343)

145 (122173)

No regular physical activities

279 (246316)

143 (124165)

1st tertile ApoB/ApoA1

241 (208278)

174 (147206)

2nd tertile ApoB/ApoA1

279 (246316)

143 (124165)

3rd tertile ApoB/ApoA1

295 (265329)

147 (130167)

1st tertile waist-hip ratio

290 (257327)

164 (142190)

2nd tertile waist-hip ratio

297 (265332)

142 (125162)

Discussion

3rd tertile waist-hip ratio

314 (283348)

145 (130162)

1st tertile BMI

300 (268335)

157 (138180)

2nd tertile BMI

291 (262324)

153 (135172)

3rd tertile BMI

318 (285355)

139 (124156)

Our study resulted in several key ndings about tobacco

use. First, use of tobacco is associated with increased
risk of AMI, consistently across all regions. Tobacco use
is one of the largest contributors to AMI worldwide. The
risk is greater in the young than in the old, for men and
women. Although the PAR of AMI was low in women
because of the low prevalence of smoking, the excess
risk associated with smoking in women was similar to
that in men. Second, smoking patterns diered
signicantly between men and women and across
dierent regions. Third, the magnitude of risk is closely,
and linearly, related to the number of cigarettes smoked,
with even low levels of smoking (eg, ve cigarettes per
day) being associated with an appreciable risk of AMI.
Fourth, former smokers have a higher risk of AMI than
do non-smokers, but this risk decreases with time after
stopping smoking. A large part of the excess risk of AMI
associated with smoking dissipates within 5 years, and
among light smokers there was no excess risk after
35 years of quitting. By contrast, moderate and heavy
smokers still had an increased risk even 20 years after
quitting. Fifth, exposure to SHS increases the risk of
AMI, in non-smokers and former smokers. The eect of
tobacco use on AMI risk was consistent in the presence
and absence of the other risk factors.
Several epidemiological studies done in developed
countries, mainly with European populations, have

Apo=apolipoprotein. BMI=body-mass index.

Table 5: Eect of tobacco use in strata dened by presence of other risk factors

Africa, and Latin America than in the other regions

(table 6) Although the odds for AMI in beedies smokers
did not dier from that of cigarette smokers, the PAR for
beedies smoking alone in South Asia, where this form of
tobacco use was more common, was about 6%.
The PAR for former smokers was much smaller,
reecting the lower risks found in former smokers, and
to low prevalence of former smokers in most regions.
Overall, the PAR for former smokers was 124%
(104146); for women it was 25% (1254) and for
men 182% (154213). In younger former smokers,
PAR was 81% (58112) and in older individuals, it was
152% (127182). In younger women who were former
smokers, PAR was 28% (1168) and in younger men,
it was 153% (111206). In older women who had quit
smoking, PAR was 25% (0779) and in older men it
was 200% (167238).
For never smokers who were exposed to SHS for
1 hour per week or longer, compared with never smokers
Current smokers

Current female smokers

Current male smokers

Current younger smokers

Current older smokers

Overall

376 (359394)

154 (134176)

463 (441485)

454 (428479)

308 (285331)

North America

454 (337576)

376 (208580)

511 (364657)

614 (460748)

324 (169531)

Western Europe

346 (268433)

133 ( 51303)

448 (351550)

532 (392667)

262 (176371)

Australia/New Zealand

533 (462603)

429 (304564)

564 (479646)

657 (550750)

451 (369536)

Eastern/central Europe

346 (295400)

161 (113 223)

452 (378527)

515 (435595)

209 (148286)
402 (328480)

Middle East

455 (414500)

57 ( 26121)

527 (475578)

490 (436544)

Africa

393 (304488)

289 (184423)

450 (326581)

426 (315545)

327 (195493)

South Asia

361 (320404)

16 (01167)

422 (375472)

388 (334445)

326 (265392)

China/Hong Kong

331 (297366)

106 (81139)

436 (389484)

387 (328451)

290 (250332)

Japan/southeast Asia

381 (311457)

144 (89226)

432 (342528)

457 (368549)

294 (191423)

Latin America

403 (337576)

271 (207347)

453 (391517)

532 (458605)

314 (257378)

Younger: female 65 years, male 55 years. Older: female >65 years, male >55 years.

Table 6: PAR as percentage (95% CI) from current smoking by region, sex, and age

654

www.thelancet.com Vol 368 August 19, 2006

Articles

documented the risks for cardiovascular disease and

cancers associated with tobacco smoking.68 These
studies generally include populations where the average
levels of other risk factors are high. Few data come from
non-European populations, among whom the levels of
other risk factors might be lower. INTERHEART used a
standardised approach across several countries, and
showed that the increased risk for AMI associated with
smoking is seen in individuals from several ethnic
groups and regions of the world. The present study
provides useful data on the association of tobacco use
with risk of AMI from all regions of the world. The
similarity of the increased risks in all regions lends
strong support to the need for a global approach to
tobacco control.
World Bank data from 2000 indicate that globally,
48% of men and 10% of women aged 15 years and older
smoked daily with wide variations by region and sex.25
Our data also show a decrease in the prevalence of
current smoking with increasing age. In view of the
higher mean age of subjects in INTERHEART
(557 years for men and 626 years for women),
compared with the average age in the populations of
the participating countries, the rates of smoking we
present here reect behaviour among older individuals.
Our data for prevalence of current smoking in controls
in various regions are close to those noted among older
individuals in some of the national surveys that reported
such data.2628 For example, whereas overall smoking
prevalence among Australian adults is about 23%, in
2001 the prevalence in those aged 5059 years was 20%
and in those older than 60 years was 10%,27 rates that
are consistent with the prevalence of smoking in
controls (14%) in INTERHEART. A national population
health survey in the Russian Federation reported that
34% of males over the age of 65 years smoked,28 which
is similar to the 35% rate of smoking in male controls
older than 65 years in the Russian INTERHEART
sample. Therefore, the estimates of smoking prevalence
in the control group of INTERHEART are generally
similar to those in populations at risk of AMI in the
regions studied.
In this study, the higher prevalence of smoking in
men than in women in low and middle income
countries, and the similarity between rates in men and
women in high income countries, show changing
patterns of smoking habits. These patterns are
consistent with the stages in the evolving pandemic
described by WHO.2 In several developed countries, the
gap between the sexes is narrowing because of an
increase in smoking among women and a decline
among men.29 In the older age groups represented in
INTERHEART, lower smoking rates were observed in
women than in men in Western Europe, North America,
Australia, and New Zealand, although smoking
prevalence among women in these regions were high.
This pattern probably reects trends 40 years ago, since
www.thelancet.com Vol 368 August 19, 2006

smoking rates in these countries tend to be higher in

younger women than in older women. In Asia and the
Middle East, smoking rates in women at all ages were
very low, but the rates of smoking in younger women in
Africa and Latin America were higher than in older
women, suggesting an increase in the rates of smoking
at younger ages. It is projected that the prevalence of
smoking among women in developing countries will
more than double from 8% to about 20% by the
year 2025.30 If smoking rates increase in women in the
future, AMI rates in women are also likely to increase
greatly, especially in developing countries, and tobacco
could become a much bigger cause of cardiovascular
disease among women in several parts of the world.
The odds ratio for AMI associated with smoking was
markedly raised in younger age groups. Because of the
higher prevalence of smoking, and the higher numbers
of cigarettes smoked per day among smokers, the PAR
for AMI in young male smokers is higher than in
middle-aged or older men. Even though the PAR for
younger women was smaller than that for younger
men, it was much higher than that for older women.
This nding suggests that there is a need and
opportunity to drastically reduce the high risks faced by
younger smokers, through active and eective tobacco
control programmes that encourage smoking
cessation.
Our data show the benets of stopping smoking.
Detailed analyses showed that the excess risk remained
high during year 1 and fell substantially during year 2.
The rate of change was almost identical to that reported
previously.31 However, the excess risk does not seem to
have completely disappeared even 20 years after quitting
in those who are moderate or heavy smokers. By
contrast, it seems to completely dissipate in light
smokers (<10 cigarettes per day) after about 35 years of
cessation. By removing from our analyses individuals
who reported exposure to SHS in the never smoker
reference group, and then in the former smokers group,
we found that the eects of SHS could explain some,
but not all, of the long-term excess risk persisting in
former smokers.
Our ndings show that use of tobacco in any form is
harmful. Beedies smoking is eight to ten times more
prevalent than cigarette smoking in South Asia.1215
Beedies are not subject to taxes, and the packages do
not carry health warnings. Sheesha smoking is now
used by many auent people from the Middle East and
by women in developed countries, as many people
wrongly think that the water removes the toxins. Control
of all forms of tobacco should be as an integral part of
any programme to reduce use, but use of indigenous
forms of inexpensive tobacco is probably much more
dicult to control than that of other tobacco products.
Findings of other studies have also shown enhanced
risk of mortality associated with use of non-smoking
tobacco.15 Our study provides important new data on
655

Articles

the link between use of non-smoking tobacco and

increased risk of AMI. Men in the two Cancer Prevention
Studies (CPS I and II)32 who chewed tobacco at baseline
had signicantly increased risk of death from ischaemic
heart disease, with relative risks 112 in CPS I and 126 in
CPS II.32 A large study in Swedish construction workers
found a 40% increased risk of death from cardiovascular
disease in users of non-smoking tobacco.33 Previous
case-control studies have not, however, shown a
signicant association.34,35 However, despite the size of
the present study, the number of individuals who used
paan and snu was still too few to allow conclusive
results on the risks associated with the use of these
specic forms of tobacco.
The mechanisms by which non-smoking tobacco
increases the risk of AMI are poorly understood. The
increase in risk of AMI associated with use of
non-smoking tobacco and smoked tobacco suggests a
role for toxins that are intrinsic to tobacco itself, and not
just conned to the smoked form. It seems that tobacco
use is associated not only with arterial damage36 but also
with short-term increases in blood pressure.37 The risk of
AMI associated with both chewing and smoking tobacco
is higher than either habit alone. This pattern of dual
tobacco use is common in several parts of South Asia,
especially in rural areas, and could lead to a large health
burden in these countries.
Our ndings clearly show the harmful eects of SHS, a
conclusion that has been widely accepted by authoritative
scientic bodies for about 10 years.38 Smokers, as well as
non-smokers, are exposed to SHS, thus possibly
compounding these adverse eects. Non-smokers
exposed to a spouses SHS had an increased risk of AMI,
suggesting that SHS has an important adverse eect, on
family members of the individual who smokes. Our
ndings are consistent with those of many previous
studies that have shown increased risk of AMI in
non-smokers reporting exposure to SHS,17,18 with a graded
increase in risk associated with extent of exposure.
However, the method used in this study to assess SHS
exposure might underestimate the dangers of SHS.
When degree of exposure is more precisely speciied,
using cotinine levels, a graded increase has been shown,
with those most exposed experiencing an almost 60%
increase in risk.39 Our data also show that individuals
with the highest level of exposure to SHS had a 62%
increase in AMI risk, but the condence intervals are
wide and might be consistent with a smaller 45% increase
in oddswhich is nevertheless of great public-health
importance. Tobacco industry research has shown how
the combustion products from cigarettes smouldering at
low temperatures are much more toxic than when air is
drawn through them,40 which might explain the risks
associated with exposure to SHS.
The focus of any anti-tobacco programme must not
only prevent young adults from starting smoking, but
also promote quitting in current smokers. The latter are
656

at high risk of developing a major vascular event (or

tobacco-related cancer) in the next decade or two.
Furthermore, the present ndings conrm earlier
reports that the association between smoking and
disease is asymmetrical; the adverse health eects of
exposure can take many years to appear, whereas the
benets of withdrawal appear much more rapidly.8
Therefore, a strategy to facilitate quitting in current
middle-aged smokers is likely to have a substantial
health benet within a relatively short time. But typically,
middle and low income countries do not have well
developed tobacco control policies and activities. The
tobacco industry has expended great eorts to create
uncertainty about the harm caused by its product.41
Consequently, public and political support for
implementation of eective tobacco control policies is
often low. In general, high income countries have better
developed control policies in place at the national and
community levels.2 These policies,42 and ongoing health
education about the diseases caused by smoking, have
resulted in reduced acceptance of smoking by the public,
especially among those with more education. The
INTERHEART data provide support for intensifying
tobacco control policies worldwide.
This study has potential limitations. One limitation is
that it involved participants who have survived AMI, and
by using a case-control design we could not elucidate the
relation between tobacco use and risk of AMI in
individuals who died in the early phase of the AMI or
because of serious complications or co-morbidity. Second,
patients with AMI might overreport exposure to tobacco
smoke, but other reports suggest that compared with
biochemical monitoring with cotinine levels, self
reporting in AMI survivors tended to underestimate the
prevalence of smoking.43 In any case, self reporting of
tobacco use is thought to be reliable and our results are
directionally similar to those of other studies with
dierent designs, such as cohort studies. The relative
risk for AMI of about 300 for AMI survivors in our study
is consistent with the relative total mortality risk of 219
for current smokers in the British Doctors study.8 The
ndings of excess risk of AMI associated with
non-smoking tobacco in our study are consistent with
those of of the large US Cancer Prevention Studies32 and
the Swedish Construction Workers study.33 Furthermore,
the risks associated with exposure to SHS are close to
those reported by a meta-analysis of 18 previous studies,
which indicated that passive smoking increases the risk
of coronary heart disease by 125.18
Tobacco use is one of the most important causes of
AMI globally, especially in men. Among smokers, even
low levels of smoking (about eight to ten cigarettes per
day) increase the risk of AMI two-fold. The excess risk of
AMI increases with all forms of tobacco smoking,
including inhalation of SHS. Chewing tobacco, which is
increasingly being promoted as a safe alternative to
smoking, is also harmful, and the risk of AMI is even
www.thelancet.com Vol 368 August 19, 2006

Articles

higher in those who chew and smoke tobacco. Since the

risks of AMI associated with smoking dissipate
substantially after smoking cessation, public-health
eorts to prevent people from starting the habit, and
promote quitting in current smokers, will have a large
impact in prevention of AMI worldwide.

11

Contributors
S Yusuf initiated the INTERHEART study, and supervised its conduct,
data analysis, and writing the report. K Teo coordinated the study in
Canada, and had the main responsibility for writing this report.
S Ounpuu coordinated the project worldwide, and reviewed and
commented on drafts. Hawken did the statistical analyses, reviewed and
commented on drafts. All other authors coordinated the study in their
respective countries and provided comments on drafts of the
manuscript.

14

Conict of interest statement

We declare that we have no conict of interest.
Acknowledgments
The INTERHEART study was funded by the Canadian Institute of Health
Research, the Heart and Stroke Foundation of Ontario, the International
Clinical Epidemiology Network (INCLEN), and through unrestricted
grants from several pharmaceutical companies, with major contributions
from AstraZeneca, Novartis, Hoechst Marion Roussel (now Aventis),
Knoll Pharmaceuticals (now Abbott), Bristol Myers Squibb, King
Pharma, and Sano-Sythelabo, and by various national bodies in
dierent countries: Chile: Universidad de la Frontera, Sociedad Chilena
de Cardiologia Filial Sur; Columbia: Colciencias, Ministerio de Salud;
Croatia: Croatian Ministry of Science and Technology; Guatemala: Liga
Guatemalteca del Corazon; Hungary: Astra Hassle, National Health
Science Council, George Gabor Foundation; Iran: Iran Ministry of
Health; Italy: Boehringer-Ingelheim; Japan: Sankyo Pharmaceutical Co,
Banyu Pharmaceutical Co, Astra Japan; Kuwait: Endowment Fund for
Health Development in Kuwait; Pakistan: ATCO Laboratories;
Philippines: Philippine Council For Health Research and Development,
Pzer Philippines Foundation, Astra Pharmaceutical, and the Astra Fund
for Clinical Research and Continuing Medical Education, Pharmacia and
Upjohn; Poland: Foundation PROCLINICA, State Committee for
Scientic Research; Singapore: Singapore National Heart Association;
South Africa: MRC South Africa, Warner-Parke-Davis Pharmaceuticals,
Aventis; Sweden: Grant from the Swedish State under LUA Agreement,
Swedish Heart and Lung Foundation; Thailand: The Heart Foundation of
Thailand, Thailand Research Fund; USA: King Pharma.
References
1
Pechacek TF, Asma S, Blair N, Eriksen MP. Tobacco: global and
community solutions. In: Yusuf S, Cairns JA, Camm AJ, Fallen EL,
Gersh BJ, eds. Evidence-based cardiology. 2nd edn. London:
BMJ Books, 2003: 10313.
2
Thun MJ, da Costa e Silva VL. Introduction and overview of global
tobacco surveillance. In: Tobacco control country proles. 2nd edn.
Geneva: World Health Organization, 2003.
3
Mackay J, Ericksen MP. The tobacco atlas. Geneva: World Health
Organization, 2002.
4
Peto R, Lopez A. The future worldwide health eects of current
smoking patterns. In: Koop CE, Pearson CE, Schwarz MR, eds.
Critical issues in global health. San Francisco: Jossey-Bass, 2001.
5
World Bank. Curbing the epidemic: government and the economics
of tobacco control. Washington DC: World Bank, 1999.
6
Negri E, La Vecchia CL, Franzosi MG, Tognoni G: Attributable risks
for nonfatal myocardial infarction in Italy. Prev Med 1995; 24: 60309.
7
Parish S, Collins R, Peto R, et al. Cigarette smoking, tar yields and
non-fatal myocardial infarction: 14 000 UK cases and
32 000 controls. BMJ 1995; 311: 47177.
8
Doll R, Peto R, Boreham J, Sutherland I. Mortality in relation to
smoking: 50 years observations on male British doctors. BMJ 2004;
328: 151928.
9
WHO. Smoking in China: a time bomb for the 21st century. Fact
sheet no 177. Geneva: World Health Organization, 1997.
10 Lam TH, He Y, Sun Li L, Shou Li L, He SF, Liang BQ. Mortality
attributable to cigarette smoking in China. JAMA 1997; 278: 150508.

www.thelancet.com Vol 368 August 19, 2006

12
13

15

16

17
18

19
20

21

22

23
24

25
26
27

28
29

30
31

32

33

34

35

36

Chen ZM, Xu Z, Collins R, Li, WX, Peto R. Early health eects of the
emerging tobacco epidemic in China. JAMA 1997; 278: 150004.
Kiyohara Y, Ueda K, Fujishima. Smoking and cardiovascular disease
in the general population in Japan. J Hypertension 1990; 8: S9S15.
Subramanian SV, Nandy S, Kelly M, Gordon D, Davey Smith G.
Patterns and distribution of tobacco consumption in India: cross
sectional multilevel evidence from the 19989 national family health
survey. BMJ 2004; 328: 80106.
Gupta PC. Survey of sociodemographic characteristics of tobacco use
among 99,598 individuals in Bombay, India using handheld
computers. Tobacco Control 1996; 5: 11420.
Rani M, Bonu S, Jha P, Nguyen SN, Jamjoum L. Tobacco use in
India: prevalence and predictors of smoking and chewing in a
national cross sectional household survey. Tob Control 2003; 12: e4.
Gupta PC, Mehta HC. Cohort study of all-cause mortality among
tobacco users in Mumbai, India. Bull World Health Organ 2000; 78:
87783.
Law MR, Wald NJ. Environmental tobacco smoke and ischemic
heart disease. Prog Cardiovasc Dis 2003; 46: 3138.
He J, Vupputuri S, Allen K, Prerost MR, Hughes J, Whelton PK.
Passive smoking and the risk of coronary heart diseasea
meta-analysis of epidemiological studies. N Engl J Med 1999; 340:
92026.
McKee M. Smoke and mirrors: clearing the air to expose the tactics
of the tobacco industry. Eur J Public Health 2000; 10: 16163.
Ounpuu S, Negassa A, Yusuf S, for the INTER-HEART
investigators. INTER-HEART: a global study of risk factors for acute
myocardial infarction. Am Heart J 2001; 141: 71121.
Yusuf S, Hawken S, Ounpuu S, et al, on behalf of the INTERHEART
Study Investigators. Eects of potentially modiable risk factors
associated with myocardial infarction in 52 countries (the
INTERHEART study): case-control study. Lancet 2004; 364: 93752.
Breslow N, Day N. Statistical methods in cancer research, vol 1: the
analysis of case-control studies. Lyon: IARC Scientic Publications,
1980.
Walter CD. The distribution of Levins measure of attributable risk.
Biometrika 1975; 62: 37174.
Engels LS, Chow WH, Vaughan TL, et al. Population attributable
risks of esophageal and gastric cancers. J Natl Cancer Inst 2003; 95:
140413.
Guindon GE, Boisclair D. Past, current and future trends in tobacco
use. Washington DC: The World Bank, 2003.
Yang G, Fan L, Tan J, et al. Smoking in China. Findings of the 1996
National Prevalence Survey. JAMA 1999; 282: 124753.
Australian Institute of Health and Welfare. 2001 National Drug
Strategy Household Survey: rst results. Canberra: Australian
Institute of Health and Welfare (Drug Statistics Series No 9), 2002
McKee M, Bobak M, Rose R, Shkolnikov V, Chenet L, Leon D.
Patterns of smoking in Russia. Tobacco Control 1998; 7: 2226.
Kumra V, Marko BA. Whos smoking now? The epidemiology of
tobacco use in the United States and abroad. Clinics Chest Med 2000;
21: 19.
Mackay J. The global tobacco epidemic: the next 25 years.
Public Health Rep 1998; 113: 1421.
Lightwood JM, Glantz SA. Short-term economic and health benets
of smoking cessation. Myocardial infarction and stroke. Circulation
1997; 96: 108996.
Henley SJ, Thun MJ, Connell C, Galle EE. Two large prospective
studies of mortality among men who use snu or chewing tobacco
(United States). Cancer Causes Control 2005; 16: 34758.
Bolinder G, Alfredsson L, Englund A, de Faire U. Smokeless tobacco
use and increased cardiovascular mortality among Swedish
construction workers. Am J Public Health 1994; 84: 399404.
Huhtasaari F, Lundberg V, Eliasson M, Janlert U, Asplund K.
Smokeless tobacco as a possible risk factor for myocardial infarction:
a population-based study in middle-aged men. Am J Coll Cardiol
1999; 34: 178490.
Hergens MP, Ahlbom A, Andersson T, Pershagen G. Swedish moist
snu and myocardial infarction among men. Epidemiology 2005; 16:
1216.
Bolinder G, Noren A, de Faire U, Wahren J. Smokeless tobacco use
and atherosclerosis: an ultrasonographic investigation of carotid
intima media thickness in healthy middle-aged men. Atherosclerosis
1997; 132: 95103.

657

Articles

37

38

39

658

Bolinder G, de Faire U. Ambulatory 24-h blood pressure monitoring

in healthy, middle-aged smokeless tobacco users, smokers and non
tobacco users. Am J Hypertens 1998; 11: 115363.
Scientic Committee on Tobacco and Health (SCOTH).
Secondhand smoke: review of evidence since 1998. Update of
evidence on health eects of secondhand smoke. UK Department
of Health. http://www.advisorybodies.doh.gov.uk/scoth/PDFS/
scothnov2004.pdf (accessed July 9, 2006).
Whincup PH, Gile JA, Emberson JR, et al. Passive smoking and
risk of coronary heart disease and stroke: prospective study with
cotinine measurement. BMJ 2004; 329: 20005.

40

41
42
43

Diethelm PA, Rielle JC, McKee M. The whole truth and nothing but
the truth? The research that Philip Morris did not want you to see.
Lancet 2005; 366: 8692.
Glantz SA, Hanauer P, Barnes DE, Slade J. Cigarette papers.
Berkeley, CA: University of California Press, 1996.
WHO. WHO framework convention on tobacco control. Geneva:
World Health Organization 2003.
Woodward M, Tunstall-Pedoe H. Biochemical evidence of persistent
heavy smoking after a coronary diagnosis despite self-reported
reduction: analysis from the Scottish Heart Health Study.
Eur Heart J 1992; 13: 16065.

www.thelancet.com Vol 368 August 19, 2006