Bulletin of the World Health Organization

Tobacco use among adults in Cambodia: evidence for a ‎tobacco epidemic among women

Pramil N Singh a, Daravuth Yel b, Sovann Sin c, Sothy Khieng d, Jaime Lopez a, Jayakaran Job a, Linda Ferry e & Synnove Knutsen a

a. School of Public Health, Loma Linda University, 2005 Nichol Hall, Loma Linda, CA 92350, United States of America.
b. WHO Tobacco Free Initiative, Phnom Penh, Cambodia.
c. National Center for Health Promotion, Ministry of Health, Phnom Penh, Cambodia.
d. Adventist Development and Relief Agency, Phnom Penh, Cambodia.
e. School of Public Health, Loma Linda University, Loma Linda, CA, USA.

Correspondence to Pramil N Singh (e-mail: psingh@llu.edu).

(Submitted: 10 September 2008 – Revised version received: 24 February 2009 – Accepted: 24 March 2009 – Published online: 02 September 2009.)

Bulletin of the World Health Organization 2009;87:905-912. doi: 10.2471/BLT.08.058917


National prevalence data on adult tobacco use in south-east Asia (Indonesia, Malaysia, the Philippines, Thailand and Viet Nam) indicate that cigarette smoking is common among men (39–75%) but not women (3–18%).1 This gender-specific pattern of cigarette consumption is also seen in other parts of Asia such as China and India.1

WHO has outlined the stages of a global tobacco epidemic that occurs after the introduction of tobacco into the economy. The first stage is a 50-year period when cigarette smoking increases to more than 50% among men. During the second stage that occurs over the next 50 years, smoking decreases among men but increases among women.2,3 The model for this later stage is based on trends in Europe and North America, where cigarette smoking increased among women after changes in social norms as a result of industrialization.4 This paradigm has been used to classify south-east Asia as being in the second stage of a tobacco epidemic, characterized by the increase in tobacco use among women.3

In analyses of the stage of the tobacco epidemic in south-east Asia, it is important to consider that the staging paradigm focuses on cigarette smoking and does not consider the contribution to gender-related trends of other forms of tobacco, such as smokeless tobacco and pipe tobacco. Moreover, the WHO Framework Convention on Tobacco Control (FCTC) Treaty tends to call for measures focused on curbing the purchase and consumption of commercial cigarettes, such as price and tax measures, regulation of tobacco content, and restrictions on package and product labelling and advertising.5 It remains to be seen whether the successful restriction of commercial cigarette consumption transfers the foci of addictive behaviour to other tobacco products, such as smokeless tobacco, hand-rolled cigarettes and pipes.

In Cambodia, published reports on adult cigarette consumption from economic surveys6 during 1999 and 2003 and a Demographic and Health Survey among women completed in 20007 confirm the trend seen in other parts of south-east Asia: cigarette smoking was highly prevalent among men (> 50%) but not women (3%). Until 2005, accurate prevalence data on all forms of tobacco use in a complete nationwide sample were not available.

During 2005–2006 we completed the largest nationwide survey of adult tobacco use ever conducted in Cambodia (n = 13 988) during the Tobacco Control Leadership Training (TCLT) programme – a collaborative effort of the Fogarty International Center, National Institutes of Health (Bethesda, MD, United States of America), Loma Linda University (Loma Linda, CA, USA), the National Institute of Statistics (Ministry of Planning, Phnom Penh, Cambodia)8. In preparation for this national prevalence survey, we conducted validity and reliability studies of the survey items and pictograms of commercial and hand rolled cigarettes, chewing tobacco and pipe tobacco that were used by survey interviewers. Our aims in the present study were (i) to identify the demographic characteristics of current tobacco users in Cambodia, especially women, and (ii) to examine the reasons for current tobacco use in demographic subgroups of the smoking population. The implications of the findings for the design of prevention programmes and tobacco control policies will be discussed.


Study population

For the TCLT survey we assembled a nationwide, representative sample of 13 988 adults aged 18 years and older. We conducted a stratified three-stage cluster sampling of the population using the 1998 Cambodia General Population Census as a sampling frame. Specifically, we stratified the country into 17 sampling domains consisting of 12 individual provinces and five groups of similar provinces. Within each domain, we then randomly selected 26 villages (we use the term “village” to represent a rural village or urban area of similarly small size) by circular systematic sampling, with the probability of inclusion of the village proportionate to its size. Each village was further subdivided into “enumeration areas,” i.e. blocks of about 110 households that were created during cartographic research for the 1998 census. For each village, one enumeration area was randomly selected for surveying. We used circular systematic sampling with a random start point to randomly select 11 households from urban areas and 15 households from rural areas.

Survey teams consisted of 4 or 5 interviewers and enumerators from the National Institute of Statistics (Ministry of Planning, Phnom Penh, Cambodia) who worked in each of the 17 census-derived regions described above. A total of 92 people were trained by the National Institute of Statistics and one of the report authors (PNS) in the pretesting and administration of the survey. In addition to conducting the interviews, survey teams updated the household enumeration and the map of the sampling units that was obtained during the 1998 census.

The survey teams visited all private households including single-member households. The survey did not cover institutional households such as military barracks, prisons and hospitals, or the residents of temples (monks). In each household, all adults aged 18 years and older were asked to participate in the study, and less than 3% declined.

The final sample consisted of 13 988 adults. This large sample provided excellent statistical power (> 90%) to estimate prevalences as low as 5% to within about 1% sampling error in separate analyses of women and men.

Written informed consent was obtained from each subject and the protocols for the national survey and substudies (i.e. salivary cotinine validation) were approved by the Institutional Review Board of Loma Linda University and the National Ethics Committee on Health (Ministry of Health) in Cambodia. An incentive of about US$ 0.50 was provided to each participant.


Survey items were designed with three sources of input: focus groups conducted on tobacco and health issues in rural and urban centres, graduate-level training in survey research of Ministry of Health personnel to select and modify items from tobacco-use surveys in other countries,8 and consultation with local nongovernmental organizations that had conducted provincial surveys on tobacco use. The final survey was conducted in the local language (Khmer) and the written survey items were translated and back-translated to verify content, criteria and semantic equivalence by bilingual and monolingual experts who used the methods described by Flaherty et al.9

The national prevalence survey administered in 2005–2006 included items on demographics, tobacco use (commercial cigarettes, hand-rolled cigarettes, chewing tobacco and pipe tobacco), age at the start of tobacco use, reasons for starting to smoke, knowledge and attitudes about tobacco, smoking cessation, anthropometric data, diet, current health, women’s health and media exposure. The items about tobacco use were in the form of interviewer-administered items adapted from WHO surveys and other national prevalence surveys that measured the intensity and duration of tobacco use, as well as interviewer-administered pictograms of commercial and local tobacco products. The pictograms were designed on the basis of the findings of qualitative research (focus groups, key informants) involving rural and urban tobacco users in Cambodia.

Validity and reliability studies

For the validity substudy we randomly sampled 201 adults aged 21 to 84 years from a rural province. They completed the TCLT survey and provided saliva samples for cotinine testing with the NicAlert test (Nymox Pharmaceutical Corporation, Hasbrouck Heights, NJ, USA). For the reliability study we randomly sampled 30 adults from 10 provinces who had completed the national prevalence survey. We re-interviewed these participants 2 to 3 weeks later with the same survey administered by a different interviewer.

The survey items and pictograms showed excellent validity when compared with salivary cotinine levels: sensitivity, 86% (95% confidence interval, CI: 78.9–93.1); specificity, 94% (95% CI: 87.6–98.4); positive predictive value, 93% (95% CI: 89.7–98.7). Reliability analysis showed excellent retest results for current smoking (κ = 0.93) and the pictograms of 37 commercial cigarette brands (κ = 1), hand-rolled cigarettes (κ = 0.80), chewing tobacco (κ = 0.86) and pipes (κ = 1).

Statistical analysis

Data analysis for this study needed to account for the stratified, multistage cluster sampling protocol described above. The 95% CIs for prevalence, means and odds ratios (ORs) for tobacco use and health variables were calculated with a Taylor series linearized method. This allowed the computation of between-cluster variance estimators that accounted for the intracluster correlation among subjects within the same village. Point estimates for prevalence, means and ORs were further adjusted by sample weights to account for different sampling fractions within each of the 17 domains described above. ORs were derived from logistic regression models where tobacco use was the dependent variable and the independent variables had to do with demographic characteristics, anthropometric data, diet and baseline health status. Interaction terms in these models were tested with a log likelihood ratio test. These statistical analyses were performed with SUDAAN software release 9.0 (RTI International, Research Triangle Park, NC, USA).


The demographic characteristics of our sample of 13 988 adults from all provinces of Cambodia are shown in Table 1. The population was predominantly of Khmer ethnicity (95.1%) and Buddhist (95.7%). Most of the sample had completed 12 years of education or less (97.6%), earned less than US$ 2 per day (86.7%) and engaged in farming (55%) or labour (8.6%) as their primary economic activity.

Fig. 1 shows that the prevalence of current tobacco use was 49.0% among men and 20.5% among women. The figure also shows the prevalence of current use of specific forms of tobacco (cigarettes, chewing tobacco, pipe tobacco) by women and men. A small proportion (0.3% of men; 0.4% of women) of participants were mixed users who chewed tobacco and smoked cigarettes.

Fig. 1. Use of tobacco,a by gender, in nationwide sample of 13 988 adults (aged ≥ 18 years) from urban and rural regions, Cambodia, 2005–2006
Fig. 1. Use of tobacco,<sup>a</sup> by gender, in nationwide sample of 13 988 adults (aged ≥ 18 years) from urban and rural regions, Cambodia, 2005–2006
a Percentage and 95% confidence intervals.

Among current tobacco users, the median age at initiation was 18.2 years (interquartile range, IQR: 15.6–21.1) among men and 29.0 years (IQR: 19.5–39.1) among women. Sample weights were used to estimate that there are 1 924 112 adults (95% CI: 1 799 228–2 048 996) who currently use tobacco in Cambodia. The two largest subgroups by gender and form of tobacco were 1 222 831 men (95% CI: 1 138 823–1 306 839) who smoked cigarettes and 560 482 women (95% CI: 504 783–616 180) who chewed tobacco. Cigarette smoking and tobacco chewing.

In Table 2 and Table 3 (available at: http://www.who.int/bulletin/volumes/87/12/08-058917/en/index.html) we show how tobacco use varied by age, gender and other demographic variables. Men primarily smoked cigarettes, and after the age of 25 years the prevalence of smoking doubled to more than 50% (Table 2). Women primarily chewed tobacco, and after the age of 25 years the prevalence of this habit more than doubled with each passing decade. As a result 43.4% of all older women (> 48 years) and almost half of all rural women (48.0%) chewed tobacco at the time of the survey (Table 3).

Among men, cigarette smokers tended to be older, rural-dwelling, married and less educated (Table 2). More than half of the men who earned < US$ 2 per day at the time of the survey smoked cigarettes. Cigarette smoking was highly prevalent (> 50%) among men engaged in tobacco farming or preparation, other farming, or the armed forces as their primary economic activity.

Women who chewed tobacco at the time of the survey tended to be lower-income, older, rural-dwelling, widowed, divorced or separated and less educated (Table 3). Among rural women, chewing tobacco was most common among midwives (67.9%), traditional healers (47.2%) and those working in tobacco farming or preparation (28.6%) as their primary economic activity.

Pipe tobacco use

We found that pipe tobacco was used primarily among local tribes (12.2% of men, 30.9% of women), other ethnic minorities (10.3% of men, 12.5% of women) or religious minorities (30.1% of men, 48.6% of women). On further analysis by province we found that pipe tobacco was only used in three provinces (Mondol Kiri, Rotanak Kiri, Stueng Reng) on the border with the Lao People’s Democratic Republic.

Multivariable model

Among all adults, the odds of using tobacco more than doubled during each decade after the age of 25 years – an effect that was particularly evident for tobacco chewing among women (OR: 5.7, 15.8–51.5) for ages 26–36 years, 37–48 years and > 48 years). Because of the high prevalence of pipe tobacco use among ethnic or religious minorities, the odds of using tobacco were particularly high among these minorities (OR: 1.6 for men, 16.2 for women), local tribes (OR: 4.2 for men, 3.7 for women) and “other adults” who were mostly immigrants (OR: 5.0 for men, 1.5 for women) (Table 4). Also, those who were married, divorced, separated or widowed were at least twice as likely to use tobacco, whereas those who had completed at least 12 years of education or who earned > US$ 3 per day were one half to one third as likely to use tobacco as participants in the reference category.

Men who engaged in tobacco farming (OR: 4.6), other farming (OR: 2.5) or labour (OR: 2.0) were more likely to be smoking cigarettes at the time of the survey. Among women, those who were midwives (OR: 5.1) or involved in tobacco farming or preparation (OR: 2.0) were more likely to be chewing tobacco at the time of the survey.

Reasons for tobacco use

We examined the reasons for starting and continuing (Table 5 and Table 6, available at: http://www.who.int/bulletin/volumes/87/12/08-058917/en/index.html) to use tobacco among the 5081 current tobacco users. Men gave as the most common reasons for starting to smoke (primarily cigarettes) the influence of older relatives (26.1%), the wish to reduce appetite (13.3%) and the desire to meet new people (12.8%). The most common reasons given by women for starting to use tobacco (primarily to chew it) were the influence of older relatives (31.9%), the need to alleviate morning sickness during pregnancy (17.0%) and the wish to experiment (13.9%). The perception that tobacco use would alleviate morning sickness was more prevalent among rural women (17.8%) than urban women (7.5%). The main reasons given for continuing to use tobacco resembled the reasons given for starting to use it, but in addition, 10% to 20% of rural men believed that cigarette smoke repelled mosquitoes and provided warmth during agricultural work in the monsoon season. As shown in Table 5 and Table 6, more than 70% of the adults who smoked pipe tobacco cited wanting to reduce appetite as their reason for starting and continuing to consume tobacco of this type.


Our findings from Cambodia confirm regional trends in south-east Asia suggesting that cigarette smoking occurs in half of all men, virtually no young women and about 3% to 6% of middle-aged and older women.1

An epidemic among women

Among Cambodian women we found an important trend towards increasing tobacco use. After the age of 18 years, the prevalence of tobacco chewing more than doubles during each decade of life up to the point that after the age of 50 years, about half of all rural women chewed tobacco at the time of the survey (Table 3). The use of a large nationwide sample lends precision to our estimate that of the 1.9 million adults who currently use tobacco in Cambodia, 560 000 are women who chew tobacco.

These data are not consistent with tobacco epidemic models24 that predict the emergence of a high prevalence of tobacco use among Cambodian women after industrialization and a demographic shift towards more cigarette-smoking. Rather, we found that if cigarette smoking does emerge as a major form of tobacco use among Cambodian women, the habit will either be added to or replace an already widespread habit of chewing tobacco during adulthood.

Why start?

It is important to note that the habit includes a centuries-old practice of chewing the tobacco leaf in combination with a betel nut (Areca catechu) and a betel leaf wrapping.10,11 The resulting “betel quid” is a mixture of these substances that remains in contact with the oral mucosa for an extended period. Arecoline, one of the active ingredients of the betel nut, is an alkaloid similar to nicotine in its effects on the central nervous system.12,13 It is a known agonist of muscarinic acetylcholine receptors involved in parasympathetic responses such as papillary and bronchial constriction.14 The betel quid mixture of tobacco-derived and areca-derived alkaloid and phenolic compounds is known to exhibit psychoactive and addictive properties due to its effects on the sympathetic and parasympathetic nervous systems. These effects include increased body temperature, increased pulse rate, increased mental alertness, improved concentration, reduced appetite and mood elevation.14,15 Additionally, the betel nut (with or without tobacco) is known throughout Asia for its medicinal effects as a stimulant, a digestive aid, an antiseptic, a deworming agent and an analgesic for headaches, arthritis and joint pain.16,17

We found that Cambodian women, particularly those in rural areas, tended to start and continue chewing tobacco because of the influence of older relatives and to alleviate morning sickness and reduce appetite. These data suggest three reasons for the use of chewing tobacco and betel nut among Cambodian women: (i) as an addictive stimulant, (ii) as part of a female rite of passage into adulthood and reproductive age, and (iii) as a remedy to relieve pregnancy-related symptoms.

It is particularly alarming that among rural Cambodian women (Table 5), about 1 out of 5 cigarette smokers and tobacco chewers started their habit in response to pregnancy-related symptoms. In addition to the wide-ranging effects of fetal tobacco syndrome, emerging data suggest that the use of the areca nut among pregnant mothers results in increased infant mortality.18,19 The specific effects (synergistic or otherwise) of various combinations of betel nut and tobacco exposure during pregnancy on neonates and infants merit further study.

The increased use of chewing tobacco among older Cambodian women may be part of a regional trend in south-east Asia. Among Indonesian women the prevalence of chewing tobacco increases from about 4% at age 25–29 years to more than 50% after the age of 60 years.20 In samples of rural women in Malaysia the prevalence of tobacco chewing was about 55–77%.21,22 In other parts of Asia and the Pacific region (Bangladesh, India, Palau, Province of Taiwan in China), similar trends in tobacco chewing are seen.23

Programme design

When considering the design of tobacco prevention programmes that target Cambodian women who are pregnant or of reproductive age, it is important to note that the use of chewing tobacco appears to be strongly influenced by beliefs passed on by older relatives. In a small focus group of 10 Cambodian women who chewed tobacco and betel nut, the behaviour was identified as a “rite of passage from girlhood into womanhood”.16 Also noteworthy is the extremely high prevalence of tobacco use (47–68%) by rural women who were midwives or traditional healers. Further research is needed on whether these village health workers actively promote the medicinal use of betel nut or tobacco.

Taken together, our findings point to the possibility that generations of young Cambodian women are adopting a tobacco habit during their reproductive and prenatal years under the direct or indirect influence of an older generation of relatives and midwives. For tobacco control programmes this poses the immediate challenge of “educating” rural women against long-held cultural, familial and traditional medical beliefs. Tobacco control programmes to decrease tobacco use in pregnant women may also need to include health education for their older relatives and health-care providers.

Use among men

Among Cambodian men, tobacco use doubled in prevalence after the age of 25 years, and the odds of smoking at the time of the survey were four-fold higher among the oldest men. Cigarettes were heavily favoured by male tobacco users, and our findings clearly indicate that most of the cigarette smokers were rural men who earned less than US$ 2 per day and worked in farming. This trend indicates that implementation of FCTC price and tax measures could potentially curb use of the more expensive commercial cigarette brands.5 Cambodia ratified the FCTC in 2004 but has yet to develop a national plan for implementation.

Use among ethnic minorities

We found that the use of pipe tobacco (i.e. water pipes) occurred entirely among ethnic and religious minorities living in Mondol Kiri, Rotanak Kiri and Stueng Reng provinces on the border between Cambodia and the Lao People’s Democratic Republic. These minorities are indigenous, mountain-dwelling peoples of the region who, unlike the Khmer–Buddhist majority, practice animism and are ethnically related to tribes that live in communities with as many as 90 000 inhabitants in the Lao People’s Democratic Republic, Thailand and Viet Nam. Our data indicate a contrast with the Khmer Buddhist majority in their use of pipe tobacco, their higher rates of cigarette smoking and their frequent (> 70%) use of smoking to reduce appetite.

Tobacco control policy

Our findings indicate that the tobacco epidemic in Cambodia extends far beyond the use of cigarettes among men. Specifically, we found that about 560 000 Cambodian women started chewing tobacco in the third decade of life and that pipe tobacco use was widespread among women and men in the minority groups (i.e. hill tribes) of Cambodia, which are also found throughout south-east Asia.

These data indicate that implementation of FCTC policies in Cambodia with a focus primarily on commercially available cigarettes will not affect a large proportion of tobacco users: those who chew tobacco (typically with betel nut) or smoke pipe tobacco. The impact of the recent tobacco control measures proposed by WHO may also be limited because banning non-commercial sources of betel nut and chewing tobacco in rural areas cannot be directly enforced through taxation or warning labels.24 Moreover, cessation strategies need to take into account the addictive properties of both arecoline and nicotine in the betel quid.

A more comprehensive national policy would require gender- and ethnic-specific tobacco control measures that target generations of women through programmes focused on prevention, cessation and education about the harms of chewing tobacco. Such programmes must take into account the use of chewing tobacco and pipe tobacco in rural areas as part of cultural, familial and traditional medicine practices. ■

Funding: The National Institutes of Health and the Fogarty International Center provided funding through the Asian Leadership Training for Tobacco Control Research programme (grant number R01 TW05964-01).

Competing interests: None declared.