Trends in Bidi and Cigarette Smoking in India from 1998 to 2015, by Age, Gender and Education

Research Trends in bidi and cigarette smoking in India from 1998 to 2015, by age, gender and education

Sujata Mishra,1 Renu Ann Joseph,1 Prakash C Gupta,2 Brendon Pezzack,1 Faujdar Ram,3 Dhirendra N Sinha,4 Rajesh Dikshit,5 Jayadeep Patra,1 Prabhat Jha1

To cite: Mishra S, ABSTRACT et al Key questions Joseph RA, Gupta PC, . Objectives: Smoking of cigarettes or bidis (small, Trends in bidi and cigarette locally manufactured smoked tobacco) in India has smoking in India from 1998 What is already known about this topic? likely changed over the last decade. We sought to to 2015, by age, gender and ▸ India has over 100 million adult smokers, the education. BMJ Global Health document trends in smoking prevalence among second highest number of smokers in the world – 2016;1:e000005. Indians aged 15 69 years between 1998 and 2015. after China. doi:10.1136/bmjgh-2015- Design: Comparison of 3 nationally representative ▸ There are already about 1 million adult deaths 000005 surveys representing 99% of India’s population; the per year from smoking. Special Fertility and Mortality Survey (1998), the Sample Registration System Baseline Survey (2004) What are the new findings? and the Global Adult Tobacco Survey (2010). ▸ The age-standardised prevalence of smoking Setting: India. declined modestly among men aged 15–69 years, ▸ Additional material is Participants: About 14 million residents from 2.5 but the absolute number of male smokers at these published online only. To million homes, representative of India. ages grew from 79 million in 1998 to 108 million view please visit the journal in 2015. This is due to population growth offset- online (http://dx.doi.org/10. Main outcome measures: Age-standardised smoking prevalence and projected absolute numbers of ting modest declines in prevalence. 1136/bmjgh-2015-000005). ▸ smokers in 2015. Trends were stratified by type of Cigarettes are displacing bidis, especially among Received 30 October 2015 tobacco smoked, age, gender and education level. younger men and among illiterate men. This change might further increase the smoker: non Revised 13 January 2016 Findings: The age-standardised prevalence of any smoker relative risks of disease. Accepted 19 January 2016 smoking in men at ages 15–69 years fell from about 27% ▸ Smoking cessation remains uncommon—only in 1998 to 24% in 2010, but rose at ages 15–29 years. about 5% of men aged 45–59 years are During this period, cigarette smoking in men became ex-smokers. India has about 4 current male about twofold more prevalent at ages 15–69 years and smokers for every quitter at these ages. fourfold more prevalent at ages 15–29 years. By contrast, ▸ Female smoking at ages 15–69 years has not bidi smoking among men at ages 15–69 years fell likely risen. modestly. The age-standardised prevalence of any smoking in women at these ages was 2.7% in 2010. The Recommendations for policy smoking prevalence in women born after 1960 was about ▸ More effective tobacco control policy including half of the prevalence in women born before 1950. By 1 higher tobacco taxation for cigarettes needs to be Centre for Global Health contrast, the intergenerational changes in smoking ’ Research, St. Michael’s implemented as a short-term priority. India scom- prevalence in men were much smaller. The absolute plicated tax structure has kept overall taxes on Hospital and Dalla Lana numbers of men smoking any type of tobacco at ages School of Public Health, cigarettes low relative to other countries, with par- 15–69 years rose by about 29 million or 36% in relative University of Toronto, ticularly low taxes on the less-expensive, short Toronto, Ontario, Canada terms from 79 million in 1998 to 108 million in 2015. cigarettes that compete with the bidi market. 2Healis-Sekhsaria Institute of This represents an average increase of about 1.7 million Longer term policies need to raise taxes on bidis. Public Health, Mumbai, male smokers every year. By 2015, there were roughly ▸ Intervention programmes to raise the currently Maharashtra, India equal numbers of men smoking cigarettes or bidis. About low levels of tobacco cessation are needed. 3 – International Institute of 11 million women aged 15 69 smoked in 2015. Among ▸ Use of reliable, representative, large scale popu- Population Studies, Mumbai, illiterate men, the prevalence of smoking rose (most lation surveys can help monitor the evolution of Maharashtra sharply for cigarettes) but fell modestly among men with 4 smoking and its consequences. World Health Organization grade 10 or more education. The ex-smoking prevalence Regional Office of South East in men at ages 45–59 years rose modestly but was low: Asia, New Delhi, India 15 years. Cigarettes are displacing bidi smoking, most only 5% nationally with about 4 current smokers for 5Tata Memorial Hospital, notably among young adult men and illiterate men. every former smoker. Mumbai, Maharashtra, India Tobacco control policies need to adapt to these Conclusions: Despite modest decreases in smoking changes, most notably with higher taxation on tobacco Correspondence to prevalence, the absolute numbers of male smokers aged products, so as to raise the currently low levels of adult Professor Prabhat Jha; 15–69 years has increased substantially over the last smoking cessation. [email protected]

Mishra S, et al. BMJ Glob Health 2016;1:e000005. doi:10.1136/bmjgh-2015-000005 1 BMJ Global Health

INTRODUCTION reported information on their smoking habits.3 The Tobacco smoking is among the largest preventable survey design and sample framework details, field causes of premature deaths globally.1 In 2010, an esti- methods and quality control steps for each survey have – mated 120 million Indian adults smoked, making India been published.31113 We applied Indian census defini- second only to China in number of smokers.23 tions for education (illiterate or with no formal educa- Historically, most of the smoked tobacco in India has tion, less than grade 10 education and grade 10 or more been in the form of bidis, small locally made cigarettes education), and for rural and urban residency status. with tobacco wrapped inside a Tendu leaf. In 2010, smoking caused about 1 million deaths, or 10% of all Definitions of smoking deaths in India, with about 70% of these deaths occur- Most of the respondents in the 1998 SFMS and the 2004 ring at the ages of 30–69 years.45The patterns of use of SRSBS were male heads of the household and provided bidis or manufactured cigarettes vary across different personal information as well as proxy information on regions and socioeconomic levels.67Smoking cessation other household members, including their wives. The is far less common than in high-income countries.3489 SFMS asked respondents if household members aged The consumption pattern of tobacco has likely 10 years or older were current smokers (combining occa- changed over the last decade in response to substantially sional and daily smokers); and if yes, whether they smoked higher income in India paired with population growth cigarette or bidis. The SRSBS asked respondents if and perhaps in response to modest tobacco control persons aged 15 years or older were usual smokers, occa- efforts.10 We examine three nationally representative sional smokers, ex-smokers or never-smokers; we defined surveys covering over 2.5 million homes and 14 million current smokers as usual or occasional smokers. The 2010 people to provide estimates of the changing trends from GATS randomly selected one individual respondent from 1998 to 2015 in any tobacco smoking, in bidi and cigar- each household and asked them to categorise themselves ette smoking, and in smoking cessation. We examine into daily smokers, less than daily smokers, ex-smokers or these trends by gender, age and generation, education never-smokers; we defined current smokers as daily or as level, and urban and rural residence. less than daily smokers. The SRSBS did not publish information on type of tobacco smoked, and reported statistics only for age groups 15–29, 30–44, 45–59 and 60 years or METHODS older. Ex-smoking was reported only in the 2010 GATS Survey populations and in the 2004 SRSBS, the latter only for any smoking. We estimated smoking prevalence using three large, Respondents who smoked cigarettes or bidis exclusively nationally representative surveys in India: the Special were classified as ‘exclusive’ while respondents who Fertility and Mortality Survey (SFMS) (1998), Sample smoked both cigarettes and bidis were classified as ‘both’. Registration System Baseline Survey (SRSBS) (2004) and We also included ‘any smokers’ as a separate category for the Global Adult Tobacco Survey (GATS) (2010). those who were either exclusive or both or smoked any The Registrar General of India (RGI) divides India other type of tobacco including hookah, cheroot, etc.14 into small geographic areas called Sample Registration System (SRS) units, based on the preceding census, to Analysis estimate vital rates nationally and for each state. In 1993, We analysed all states and union territories but retained the RGI randomly selected 6671 of these small areas three states formed after the 1991 census within their (4436 rural and 2235 urban) from the 1991 census to original states (thus, Madhya Pradesh and Chhattisgarh, form the 1993–2003 SRS frame. The SFMS was under- Bihar and Jharkhand, and Uttar Pradesh and taken in these SRS units starting in February 1998, cover- Uttarakhand, are listed as pairs). We combined the ing approximately 6 million people living in smaller states of the Northeast (except Assam) to approximately 1.1 million households across India. include Arunachal Pradesh, Manipur, Meghalaya, Jammu and Kashmir and the rural units of Nagaland Mizoram, Nagaland, Sikkim and Tripura. Sample frame- (<1% of India’s population) were excluded due to oper- work and weights were used to derive prevalence esti- ational problems.11 In 2004, the RGI randomly selected mates. Analyses were conducted in STATAV14.0. 7597 areas (4433 rural and 3164 urban) from the 2001 We standardised smoking prevalence to the age distribu- census, to form the 2004–2013 SRS frame. The SRSBS tion of the 2011 Indian Census.15 We tested statistical sig- was conducted over several months in 2004, covering 1.3 nificance of the differences between the surveys by million households in all Indian states and union terri- age-standardised rate ratios (ASRRs) with 99% CIs. tories.12 In 2010, the GATS surveyed over 79 690 nation- Estimates of the absolute number of smokers for both ally representative households selected from the 2001 genders in 1998 applied the survey age-specificprevalence census enumeration blocks and villages for urban and to the interpolated population totals from the 1991 to rural areas, respectively. The GATS covered all non- 2001 censuses. Estimates of the absolute number of institutionalised adults within 29 states and two union smokers in 2015 applied annual rates of change in territories (Chandigarh and Puducherry). A total of smoking prevalence between 1998 and 2010, for each age 69 296 adults completed the interview and provided self- and state strata for men and 15–69 years for women, to the

2 Mishra S, et al. BMJ Glob Health 2016;1:e000005. doi:10.1136/bmjgh-2015-000005 BMJ Global Health

2015 United Nations population totals.16 Throughout, we The age-standardised prevalence of any smoking in men focus on smoking at ages 15–69 years for three reasons. aged 15–69 years fell from about 27% in 1998 to 24% in First, smoking trends among ages 70 years or older are less 2010 (change per decade −2.0%, ASRR=0.9, 99% CI 0.9 reliably ascertained (due in part to higher absolute mortal- to 0.9; figure 2). The age-standardised prevalence of any ity rates from smoking at older ages).17 Second, India’s smoking in women aged 15–69 years rose from 1.4% in age structure is still quite young, with 80% of India’spopu- 1998 to 2.7% in 2010, but this might well be an artefact of lation being below 45 years of age.16 Third, about 70% of the survey methodology as most of the reported increase tobacco deaths in India are reported at ages 30–69 years.4 occurred at older ages. The SFMS of 199812 13 19 relied on Hence, the more important public health goal is avoid- proxy reporting by male heads of household and they – ance of death in middle versus older age.18 likely under-reported female smoking,20 22 whereas the smoking prevalence was more accurately captured from RESULTS self-reports in the GATS of 2010.3 The changes between We were able to report on men and women in 29 Indian 1998, 2004 and 2010 did not show consistent increases in states representing 99% of India’s population ( Jammu smoking in young adult women, as would be expected if and Kashmir, and 7 of the 8 union territories were there were an increase in female smoking. In each time excluded from 1 or 2 of the surveys). Individual level period, smoking prevalence at ages 30–69 years in women data on smokers analysed about 530 000 men and born after 1960 was about half of the prevalence in 32 000 women above age 15 years in 1998, and 10 000 women born before 1950. By contrast, the intergenera- men and 1200 women in 2010 (see web table 1). tional changes in smoking prevalence in men at these ages The age-specific prevalence of any smoking in men were much smaller (figure 3). As few women smoked, followed a similar pattern in 1998 and 2010, peaking at most of the following results focus on men. about 50 years of age and declining at older ages. At The age-standardised prevalence of any smoking most ages, the age-specific prevalence was significantly among men aged 15–69 years rose significantly in Delhi, lower in 2010 than in 1998 (figure 1). The notable Kerala, Odisha, Punjab and the Northeast States but fell exception was in men aged 15–29 years, among whom modestly in other states (figure 2). The age-standardised the age-specific prevalence was higher in 2010 than in prevalence of any smoking in men aged 15–29 years rose 1998. The age-specific prevalence of any smoking in (9–12%; change per decade +2.4%, ASRR=1.3, 1.2 to women showed a very different pattern than that for 1.4), but fell at older ages. These trends markedly men, with prevalence peaking above 60 years of age. diverged between cigarettes and bidis (figure 4). The observed increases in the age-specific prevalence in Cigarette smoking became more prevalent at ages 15– women between the 1998 and 2010 surveys were concen- 69 years (change per decade +4.8%, ASRR 2.2, 2.1 to trated mostly at these older ages. Age-specific prevalence 2.2) with large increases at ages 15–29 years (change per rates in men in 2004 for any smoking at ages 15–29, 30– decade +4.5%, ASRR 3.9, 3.6 to 4.2) and smaller, but 44 and 45–59 years, were in between the 1998 and 2010 still significant, increases at older ages. By contrast, the results (see web table 2A–C), consistent with modest age-standardised prevalence of bidi smoking among increases at ages 15–29 years and modest declines in the men fell at ages 15–69 years (change per decade −2.3%, latter two age groups. ASRR 0.9, 0.8 to 0.9), and within each age group.

Figure 1 Smoking prevalence by age and gender (with 99% CI): 1998–2010. GATS, Global Adult Tobacco Survey; SFMS, Special Fertility and Mortality Survey.

Mishra S, et al. BMJ Glob Health 2016;1:e000005. doi:10.1136/bmjgh-2015-000005 3 BMJ Global Health

Figure 2 Age-standardised smoking prevalence among adults aged 15-69 years, by state: 1998–2010, absolute change (%) per decade and number of smokers in 2015 (in millions). GATS, Global Adult Tobacco Survey; SFMS, Special Fertility and Mortality Survey. Note that reliable estimates for absolute changes per decade could not be estimated for women. We applied GATS 2010 smoking prevalence for women ages 15-69 years to the 2015 population.

Even using the GATS definition of exclusive smoking, cig- The highest age-standardised prevalence of any smoking arette smoking rose substantially and bidi smoking fell in men aged 15–69 years was in illiterate men in both 1998 modestly (data not shown). Cigarette smoking rose more and 2010 (see web table 3). Among illiterate men at these in rural men than in urban men while bidi smoking fell ages, the prevalence of cigarette, bidi or any smoking rose modestly in both areas for nearly all age groups except at (figure 5), most sharply for cigarettes. By contrast, among ages 15–29 years (see web figures 1A,B). men with grade 10 or more education, the prevalence of

Figure 3 Smoking prevalence by decade of birth cohort and gender: 1998–2010. Note: Numbers near each point represent the sample size. GATS, Global Adult Tobacco Survey; SFMS, Special Fertility and Mortality Survey.

4 Mishra S, et al. BMJ Glob Health 2016;1:e000005. doi:10.1136/bmjgh-2015-000005 BMJ Global Health bidi or any smoking fell. Even among these more educated Smoking cessation among men is best measured at men, cigarette smoking rose modestly. ages 45–59 years, well after the peak of smoking initi- As the age-specific rates in 1998, 2004 and 2010 ation and when smokers are more likely to be aware of showed consistent increases at ages 15–29 years and health hazards that could prompt cessation.17 The age- modest declines at older ages, we were able to make specific ex-smoking prevalence in men at these ages rose reasonably sensible forward projections to 2015. The modestly from 2004 to 2010, from about 2% to 5% absolute numbers of men smoking any type of tobacco nationally (figure 6), with variation in the ex-smoking at ages 15–69 years rose by about 29 million or 36% in prevalence across states in 2010 ranging from 19% in relative terms from 79 million in 1998 to 108 million in Kerala to <1% in Punjab. The ratio of current to former 2015 (figure 4, bottom panel), representing an average smokers in men at these ages was 4 in 2015 for India, annual increase of about 1.7 million male smokers. The ranging from 1 in Kerala and Bihar/Jharkhand to 60 in number of any smokers at these ages rose in urban West Bengal. Very few women were ex-smokers. India by about 68% from 19 to 31 million and the number in rural India increased about 26% from 61 to 77 million. The five largest relative increases over one DISCUSSION decade in the absolute number of male smokers aged The Indian government implemented the Cigarettes and 15–69 years were in Delhi (220%), Punjab (120%), the Other Tobacco Products Act (COTPA)23 in 2003 and rati- North East states excluding Assam (60%), Odisha fied the WHO’s Framework Convention on Tobacco (50%) and Maharashtra (40%). Decreases in the abso- Control in 2004, as well as the Cable Television Networks lute number of male smokers were observed in only (Amendment) Act 2000 prohibiting tobacco advertising two states—Goa and Assam (see web table 1). In 2015, in all state-controlled electronic media and publications, we estimate that approximately 61 million Indian adult including cable television.24 25 Further, the Government men aged 15–69 years smoked cigarettes (40 million has also included tobacco control in the priorities of the exclusively) and 69 million smoked bidis (48 million ongoing National Rural Health Mission.26 Despite these exclusively). At ages 15–69 years, and assuming that the programmes, the major challenge to success is effective prevalence in GATS in 2010 was unchanged by 2015, implementation of the provisions of COTPA, especially in there were about 11 million women who smoked, about enforcement of bans on smoking in public places (which one-tenth the total of men who smoked. Given higher are known to raise cessation rates).124Most importantly, mortality and greater reporting problems at older ages, these trends in smoking reflect the lack of substantial there were, less reliably, an estimated 6 million men and increases in tobacco excise taxes, which have not kept up 2.5 million women smokers at ages 70 years and above. with the increased affordability of cigarettes and bidis.10

Figure 4 Age-standardised rate (ASR) of smoking among men by age group, product between 1998 to 2010 and number of smokers in 2015 (in millions). ASRR, age-standardised rate ratios between the Global Adult Tobacco Survey and the Special Fertility and Mortality Survey.

Mishra S, et al. BMJ Glob Health 2016;1:e000005. doi:10.1136/bmjgh-2015-000005 5 BMJ Global Health

one-third since 1998 to nearly 108 million in 2015. The increase is mostly due to population growth offsetting the modest declines in prevalence over this time period, similar to the pattern observed in other countries.27 There is a clear shift in consumption away from bidis towards manufactured cigarettes. The sharpest relative and absolute increase was for cigarette smoking, particularly in young adult men aged 15–29 years. The increases in cigarette use among younger adult men were seen in rural areas and were greatest among illiterate men. Rapid income growth over the last decade has most likely contributed to the shift in smoking—from the less-expensive bidis to cigarettes. Price is the most important determinant of consumption. Relative to income, cigarettes and bidis have become less costly in the last decade.10 Moreover, and most relevant for Figure 5 Age-standardised rate (ASR) of smoking among policy, India’s complicated tax structure has kept overall men aged 15 to 69 years by product, and by two levels of taxes on cigarettes low relative to other countries, with education: 1998–2010. ASRR, age-standardised rate ratios particularly low taxes on the inexpensive, short cigar- – between the Global Adult Tobacco Survey and the Special ettes that compete with the bidi market.28 30 The Fertility and Mortality Survey. increase in cigarette smoking is consistent also with market reports showing that the absolute volume of Hence, tobacco control in India urgently requires effect- cigarettes sold in India has risen from about 98 billion ive implementation of national policies. sticks in 2000 to 114 billion in 2012.31 Unfortunately, Our study of nationally representative Indian surveys bidi sales data are unavailable, as most bidis are sold by over more than a decade ﬁnds substantial increases in the small cottage industries with little monitoring or regula- number of male smokers aged 15–69 years, rising over tion and attract low or no taxes.32

Figure 6 Ex-smoker rates in India among men (45–59 years) from 2004 to 2010; absolute number (in million) of smokers and ex-smokers in 2015. GATS, Global Adult Tobacco Survey; UN, United Nations. Note that, * it is unlikely that smoking cessation has decreased from 2010-2015. Therefore we assigned GATS 2010 ex-smoking prevalence to Assam, West Bengal and Punjab which reported a decline. Odisha was excluded from the analyses due to few numbers of smokes in the GATS. †We applied GATS 2010 smoking and ex-smoking prevalence for women to estimate 2015 absolute totals. ‡The number of smokers and ex- smokers are based on 2015 UN Population totals for India.

6 Mishra S, et al. BMJ Glob Health 2016;1:e000005. doi:10.1136/bmjgh-2015-000005 BMJ Global Health

The observed increase in female smoking is likely to female smoking prevalence at older ages between the be an artefact of reporting. A true increase in smoking 1998 SFMS or 2004 SRSBS and the 2010 GATS suggests would be expected among younger women, who have a reporting bias from husbands unaware of or unwilling seen more rapid income growth, and are the subject of to disclose their wife’s smoking. Proxy reporting might tobacco industry promotion. However, among younger also result in under-reporting of smoking at younger adult women, there was little increase in smoking ages. However, other surveys42 43 find prevalence fits (figure 1, see web table 1), and indeed smoking preva- with increased consumption among younger adult men. lence was less than half of that seen in older genera- Second, whereas the GATS allowed smokers to self- tions. Among female respondents in the ongoing identify smoking multiple products, the SFMS required Million Death Study (conducted in the same areas as classifying smokers as either cigarette smokers or bidi the SRSBS), there has been no major shift in smoking smokers. Thus, we might have underestimated cigarette patterns among younger women from 2004 to 2013 and bidi smoking prevalence rates in the 1998 baseline, (data not shown). Among selected high schools in par- and exaggerated the actual increase by 2010. Such mis- ticular regions of India, there are reports that female classification of cigarette or bidi smoking would not affect and male smoking at very young ages (13–15 years) are the findings for any smoking prevalence. Moreover, more equal,33 and, in the future, young women, particu- smoking prevalence among living respondents in the larly in urban areas, could take up cigarette smoking. ongoing Million Death Study finds also an increased use A robust measure of successful tobacco control pol- of cigarettes relative to bidis. Lastly, the sampling frame icies is rapid increases in ex-smoking prevalence among for the surveys differed, being much larger for SFMS and later age adults.17 34 We find that smoking cessation SRSBS than GATS, and the GATS method of selecting remains uncommon in India: only about 5% of the men households may have missed (the few) adults living aged 45–59 years were ex-smokers in 2010, a modest alone.3 All three sampling frames were nationally repre- increase from 2004. A similarly helpful statistic is the sentative, with large sizes, and likely accurately captured ratio of current to former smokers. India has about four the true underlying trends in male smoking. current male smokers for every quitter at these ages. By A substantial reduction in smoking is central to achiev- contrast, in the USA and other high-income countries, ing the United Nations’ 2030 goals to decrease prema- there are now as many ex-smokers as current smokers at ture death from non-communicable disease.18 Apart these ages.35 36 Countries that are more comparable to from effective implementation of the WHO’s recom- India in income but adopted tobacco control earlier mended measures, including increasing tobacco tax- (including through tax increases) now have a substan- ation,1 the use of reliable, representative, large scale tially higher prevalence of quitting.37 population surveys is helpful in accelerating tobacco The absolute number of tobacco-attributable deaths control policies and in monitoring their impact, particu- among India’s 120 or so million smokers will continue to larly in raising the low rates of smoking cessation in rise. The exact totals depend on the risks of smoking cigar- India. ettes or bidis, the age of initiation and background mortality rates from causes other than smoking. Indian male Acknowledgements The authors thank Krishna Palipudi for his valuable smokers traditionally start smoking later in life4 and smoke inputs while drafting and revising the manuscript, and Mehak Bhatia for graphics. fewer sticks per day when compared with male smokers in high-income countries.9 Nonetheless, the smoker versus Contributors PJ, SM and RAJ contributed to the conception and design of non-smoker risk ratios for dying from smoking attributable the study. PJ, SM, RAJ and BP contributed to the analysis and interpretation 4 38 of the data. All the authors performed the drafting of the manuscript for causes remain high in India. Unlike in China, we did important intellectual content. not see a dramatic shift to smoking at younger ages (data Funding This work was supported by the University of Toronto and the Bill not shown) among men. Some, but not all, studies have and Melinda Gates Foundation (51447). shown that the relative risk of tobacco attributable causes of death is greater for cigarette than for bidi smoking.414 Disclaimer The funders had no role in study design, analysis, interpretation of results, writing of the report, nor in the decision to submit this article for The risks of death among cigarette smokers who smoke publication. more than eight cigarettes per day, closer to smoking Competing interests None declared. patterns in the USA or Canada, are particularly pro- nounced.139Future risks will also depend on whether the Ethics approval International Institute of Population Sciences, Mumbai and currently low quit rates rise substantially. Ministry of Health, New Delhi, India. Our analyses need to consider a few limitations. First, Provenance and peer review Not commissioned; externally peer reviewed. the SFMS and the SRSBS relied on proxy reporting by Data sharing statement The GATS data are publicly available (Link: http:// the head of the household, which has shown in some www.who.int/tobacco/surveillance/gats/en/; India Report: Global Adult – direct comparisons to be valid20 22 and tightly correlated Tobacco Survey: India Report. Mumbai, India: International Institute for with self-reports. However, other studies find that the Population Sciences (IIPS), Ministry of Health and Family Welfare GoI, 2010). 40 The SFMS and SRS baseline key results have been published, but the data are correlation between proxy and self-responses varies, provided under special agreement with the Registrar General of India and are including by gender or relationship status of the not yet in the public domain. Basic data tables from SFMS, SRS baseline and respondent and subject.41 The reported increase in GATs are provided as web-tables for reference. (Reference 1: Special Fertility

Mishra S, et al. BMJ Glob Health 2016;1:e000005. doi:10.1136/bmjgh-2015-000005 7 BMJ Global Health and Mortality Survey 1998: a report on 1 million homes. New Delhi, India: 21. Hyland A, Cummings KM, Lynn WR, et al. Effect of proxy-reported Registrar General of India, 2005; 2: Sample Registration System: Statistical smoking status on population estimates of smoking prevalence. – Report 2004. New Delhi, India: Registrar General of India, 2005). Am J Epidemiol 1997;145:746 51. 22. Mohan D, Neufeld K, Chopra A, et al. Agreement between Open Access This is an Open Access article distributed in accordance with head of household informant and self-report in a community the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, survey of substance use in India. Drug Alcohol Depend 2003;69:87–94. which permits others to distribute, remix, adapt, build upon this work non- 23. The Cigarettes and Other Tobacco Products (Prohibition of commercially, and license their derivative works on different terms, provided Advertisement and regulation of Trade and Commerce, Production, the original work is properly cited and the use is non-commercial. See: http:// Supply and Distribution) Act; An Act enacted by the Parliament of creativecommons.org/licenses/by-nc/4.0/ Republic of India by notification in the Official Gazette. (Act 32 of 2003). In: India Po, ed. New Delhi, 2003. http://www.who.int/fctc/ reporting/Annexthreeindia.pdf REFERENCES 24. WHO. Framework convention on tobacco control. Geneva: World 1. Jha P, Peto R. Global effects of smoking, of quitting, and of taxing Health Organization, 2004. tobacco. N Engl J Med 2014;370:60–8. 25. The Cable Television Networks (Amendment) Act In: India Go, ed., 2. Shafey O, Eriksen M, Ross H, et al. The Tobacco Atlas. 3rd edn. 2000. http://www.wipo.int/edocs/lexdocs/laws/en/in/in033en.pdf Atlanta, USA: American Cancer Society, 2009. 26. Kaur J, Jain DC. Tobacco control policies in India: implementation 3. Global Adult Tobacco Survey: India Report. Mumbai, India: and challenges. Indian J Public Health 2011;55:220–7. International Institute for Population Sciences (IIPS), Ministry of 27. Ng M, Freeman MK, Fleming TD, et al. Smoking prevalence and Health and Family Welfare GoI, 2010. http://mohfw.nic.in/ cigarette consumption in 187 countries, 1980–2012. JAMA WriteReadData/l892s/1455618937GATS%20India.pdf 2014;311:183–92. 4. Jha P, Jacob B, Gajalakshmi V, et al. A nationally representative 28. Bahri C. How India’s tax system helps heavily taxed cigarettes case-control study of smoking and death in India. N Engl J Med flourish, Mumbai, India: India Spend, September, 2015. http://www. 2008;358:1137–47. indiaspend.com/cover-story/how-indias-tax-system-helps-heavily- 5. Gajalakshmi V, Peto R, Kanaka TS, et al. Smoking and mortality taxed-cigarettes-flourish-33896 from tuberculosis and other diseases in India: retrospective study of 29. Bahri C. Why India ignores a $16-billion smoking-led health crisis. 43000 adult male deaths and 35000 controls. Lancet Mumbai, India: India Spend, September, 2015. http://www. 2003;362:507–15. indiaspend.com/cover-story/why-india-ignores-a-16-billion-smoking- 6. Rani M, Bonu S, Jha P, et al. Tobacco use in India: prevalence and led-health-crisis-37620 predictors of smoking and chewing in a national cross sectional 30. Guindon GE, Mishra S, Jha P. Still ample room to raise India’s household survey. Tob Control 2003;12:e4. tobacco tax. BMJ 2014;349:g4.680. 7. Subramanian SV, Nandy S, Kelly M, et al. Patterns and distribution 31. Cigarettes in India. A market report from ERC. The Canadean of tobacco consumption in India: cross sectional multilevel evidence Group, 2014. from the 1998–9nationalfamilyhealthsurvey.BMJ 2004;328:801–6. 32. Nandi A, Ashok A, Guindon GE, et al. Estimates of the economic 8. Gupta PC. Survey of sociodemographic characteristics of tobacco contributions of the bidi manufacturing industry in India. Tob Control use among 99,598 individuals in Bombay, India using handheld 2015;24:369–75. computers. Tob Control 1996;5:114–20. 33. Sinha DN, Gupta PC, P G. Tobacco use among students and school 9. Jha P, Ranson MK, Nguyen SN, et al. Estimates of global and personnel in India. Asian Pac J Cancer Prev 2007;8:417–21. regional smoking prevalence in 1995, by age and sex. Am J Public 34. Jha P, Chaloupka FJ, Moore J, et al. Tobacco addiction. In: Jamison Health 2002;92:1002–6. DT, Breman JG, Measham AR, et al, eds. Disease control priorities 10. Jha P, Guindon GE, Joseph RA, et al. A rational taxation system of in developing countries. 2nd edn. Washington DC: World Bank; bidis and cigarettes to reduce smoking deaths in India. Econ Polit 2006. 445–493. Wkly 2011;42:44–51. 35. Jha P, Ramasundarahettige C, Landsman V, et al. 21st-century 11. Special Fertility and Mortality Survey 1998: a report on 1 million hazards of smoking and benefits of cessation in the United States. homes. New Delhi, India: Registrar General of India, 2005. N Engl J Med 2013;368:341–50. 12. Sample registration system, baseline report of 2004. New Delhi, 36. Jha P, MacLennan M, Yurekli A, et al. Global hazards of tobacco India: Registrar-General of India, 2007. and the benefits of smoking cessation and tobacco taxes in Cancer. 13. Sample Registration System: Statistical Report 2013. New Delhi, Gellband H, Jha P, et al, eds. Disease Control Priorities. 3rd Edition India: Registrar General of India, 2014. 2015:175–93. 14. Gupta PC, Pednekar MS, Parkin DM, et al. Tobacco associated 37. Giovino GA, Mirza SA, Samet JM, et al. Tobacco use in 3 billion mortality in Mumbai (Bombay) India. Results of the Bombay Cohort individuals from 16 countries: an analysis of nationally representative Study. Int J Epidemiol 2005;34:1395–402. cross-sectional household surveys. Lancet 2012;380:668–79. 15. India Census Data 2011. The Registrar General & Census 38. Chen Z, Peto R, Zhou M, et al. Contrasting male and female Commissioner, India, New Delhi, Ministry of Home Affairs: trends in tobacco-attributed mortality in China: evidence from Government of India. http://www.censusindia.gov.in/ successive nationwide prospective cohort studies. Lancet 16. World Population Prospects Database extract: United Nations, 2015;386:1447–56. Department of Economic and Social Affairs, Population Division, 39. Jha P, Gupta PC, Peto R. Case–control study of smoking and death 2015. http://esa.un.org/unpd/wpp/Download/Standard/Population/ in India. N Engl J Med 2008;358:2842–5. 17. Jha P. Avoidable global cancer deaths and total deaths from 40. Navarro AM. Smoking status by proxy and self report: rate of smoking. Nat Rev Cancer 2009;9:655–64. agreement in different ethnic groups. Tob Control 1999;8:182–5. 18. Norheim OF, Jha P, Admasu K, et al. Avoiding 40% of the premature 41. Harakeh Z, Engels RC, Vries HD, et al. Correspondence between deaths in each country, 2010–30: review of national mortality trends proxy and self-reports on smoking in a full family study. Drug Alcohol to help quantify the UN sustainable development goal for health. Depend 2006;84:40–7. Lancet 2015;385:239–52. 42. National Family Health Survey (NFHS3) 2005–06. In: Ministry of 19. Fu SH, Jha P, Gupta PC, et al. Geospatial analysis on the Health and Family Welfare GoI, and International Institute for distributions of tobacco smoking and alcohol drinking in India. PLoS Population Sciences. India, ed. Mumbai, 2007. https://dhsprogram. ONE 2014;9:e102416. com/pubs/pdf/FRIND3/FRIND3-Vol1AndVol2.pdf 20. Gilpin EA, Pierce JP, Cavin SW, et al. Estimates of population 43. Causes of death in India, 2001–03: Sample Registration System. smoking prevalence: self-vs proxy reports of smoking status. Registrar General of India & Centre for Global Health Research. Am J Public Health 1994;84:1576–9. 2009. http://www.censusindia.gov.in

8 Mishra S, et al. BMJ Glob Health 2016;1:e000005. doi:10.1136/bmjgh-2015-000005 Supplementary Appendix Web Table 1: Number of smokers in India (sample and population): 1998 and 2010. Number of smokers studied Number of smokers (000) State and Region 1998/2010 1998 2015 Annual absolute Men 15-69 Years change (%) Himachal Pradesh 10,998/360 674 839 1.4 Punjab 4095/160 560 1700 12.0 Chandigarh 2175/204 64 84 1.9 Haryana 24,254/364 2291 3231 2.4 Delhi 4815/328 612 2851 21.5 Rajasthan 34,952/370 5286 8035 3.1 Uttar Pradesh & Uttaranchal 54,445/821 14,523 18,014 1.4 Bihar & Jharkhand 38,646/381 7413 7855 0.4 Assam 28,840/713 3044 2340 -1.4 West Bengal 67,614/652 11,058 13,582 1.3 Odisha 15,667/230 1878 3488 5.0 Madhya Pradesh & Chhattisgarh 31,865/479 6168 9165 2.9 Gujarat 20,659/258 3443 4242 1.4 Maharashtra 12,906/326 3441 5707 3.9 Andhra Pradesh 30,599/416 7323 9605 1.8 Karnataka 27,452/217 3863 5333 2.2 Goa 3924/84 76 33 -3.3 Kerala 21,462/230 2270 3534 3.3 Tamil Nadu 20,636/268 3575 5315 2.9 Northeast 44,693/2379 1386 2796 6.0 Rural Men 41,2042/6449 60,957 76,718 1.5 Urban Men 96,965/3321 18,746 31407 4.0 INDIA TOTAL 509,007/9770 79,428 108,125 2.1 Total 70+ 21,864/478 4128 5896 2.5 Total 15+ 530,871/10248 83,437 114,021 2.2 Women 15-69 Years INDIA TOTAL 31,949/1214 11,183 -* Note: * Reliable estimates for annual absolute change (%) for women could not be estimated since we believe that smoking among women has been declining based on our estimation by birth cohorts.

Web Table 2a: The 1998 and 2010 age standardized smoking prevalence for ages 15-69 years and 15-29 years, by state and location of residence and the 2004 non-standardized rates of smoking for 15-29 years *. 15-69 years 15-29 years States/ Regions Total Rural Urban Total Rural Urban 1998 2010 1998 2010 1998 2010 1998 2004 2010 1998 2004 2010 1998 2004 2010 Men Himachal Pradesh 38.2 33.1 38.2 33.1 38 33.1 12.1 8.0 13.7 12.2 7.8 12.7 11.0 8.6 19.6 Punjab 7.5 12.7 7.5 12.8 7.5 12.7 3.0 5.4 8.3 2.6 4.8 5.4 3.7 7.0 12.6 Chandigarh 19.4 18.0 18.9 16.5 19.4 18.1 11.4 11.2 9.6 21.9 17.8 19.3 9.9 7.5 8.3 Haryana 38.2 32.3 37.9 32.0 38.8 32.9 16.0 17.2 12.7 19.2 20.6 14.3 7.4 10.8 9.3 Delhi 13.9 31.4 13.5 31.0 13.9 31.4 4.8 10.3 20.8 6.5 10.5 9.8 4.7 10.3 21.1 Rajasthan 34.7 31.9 34.6 31.7 35.1 32.4 14.3 21.4 12.8 16.0 24.7 13.2 9.0 5.8 11.5 Uttar Pradesh/ Uttaranchal 30.8 25.1 30.8 25.1 30.6 24.9 10.0 13.8 9.9 10.6 15.2 11.0 7.9 8.3 6.0 Bihar/ Jharkhand 25.8 19.6 25.8 19.6 25.6 19.5 7.9 12.0 16.0 8.3 14.1 16.8 5.7 6.8 11.2 Assam 39.8 25.7 39.4 25.5 42.0 26.6 15.0 18.9 15.4 15.7 23.0 14.2 10.7 16.0 21.1 West Bengal 44.7 39.2 44.0 38.8 46.2 40.1 20.7 19.8 23.9 21.7 21.5 23.8 18.0 15.3 24.1 Odisha 17.5 20.4 17.5 20.3 17.4 20.6 6.0 11.7 10.1 6.5 12.7 10.1 3.8 8.2 11.0 Madhya Pradesh/ Chhattisgarh 27.4 25.7 27.3 25.7 27.5 25.8 8.8 12.8 16.2 10.2 16.7 15.4 5.1 5.3 19.6 Gujarat 22.8 18.8 22.9 19.0 22.6 18.7 7.5 9.5 3.7 6.9 10.9 3.5 8.3 5.4 4.0 Maharashtra 11.6 12.2 11.6 12.2 11.6 12.1 2.6 2.8 4.8 2.7 2.0 3.5 2.5 4.2 6.1 Andhra Pradesh 31.8 29.7 31.8 29.8 31.7 29.3 11.0 8.4 12.0 12.3 8.8 11.6 7.7 7.0 12.9 Karnataka 24.4 22.3 24.4 22.3 24.3 22.3 6.8 7.1 9.6 6.7 6.9 8.7 6.8 7.9 10.8 Goa 17.8 9.0 18.1 9.0 17.7 8.9 2.3 2.3 4.0 2.3 2.4 4.4 2.2 2.2 3.7 Kerala 24.6 27.8 24.5 27.8 24.7 27.9 6.9 9.0 16.6 7.7 9.7 14.6 4.7 5.8 22.6 Tamil Nadu 18.0 18.6 17.7 18.3 18.4 18.8 5.2 5.3 7.2 4.7 5.2 6.0 5.9 5.5 8.4 North East 39.4 48.6 39.0 48.4 40.2 49.1 17.6 21.7 34.9 18.2 22.0 34.5 16.0 20.3 35.9 INDIA TOTAL 26.7 24.3 29.3 26.0 21.0 20.4 9.4 11.8 12.3 10.4 13.0 12.8 7.0 8.8 11.1 Women INDIA TOTAL 1.4 2.7 1.6 3.3 0.9 0.4 0.6 1.3 0.5 0.7 1.5 0.7 0.5 0.8 0.2 * Note: The 2004 estimates are age-specific prevalence estimates from SRSBL. SRSBL provided appropriate age specific patterns that are close to 2011 census population which is used as the standard population for age standardized rates using SFMS and GATS.

Web Table 2b: The 1998 and 2010 age standardized smoking prevalence for ages 30-44 years and 45-59 years, by state and location of residence and the 2004 non-standardized rates of smoking at these ages*.

30-44 years 45-59 years Total Rural Urban Total Rural Urban States/ Regions 1998 2004 2010 1998 2004 2010 1998 2004 2010 1998 2004 2010 1998 2004 2010 1998 2004 2010 Men Himachal Pradesh 51.1 38.1 40.2 52.0 41.9 40.9 45.5 30.9 34.6 62.6 47.7 57.3 64.6 54.9 60.2 49.4 34.3 37.5 Punjab 10.5 15.1 17.0 9.7 13.3 16.9 11.7 19.0 17.3 11.6 15.7 12.3 11.3 13.7 11.6 12.1 20.5 12.7 Chandigarh 28.6 28.0 20.5 45.6 31.0 27.1 26.5 26.8 19.6 23.5 26.0 28.2 35.0 28.0 47.2 22.7 25.5 26.4 Haryana 53.7 47.5 42.2 62.9 56.9 44.2 35.1 32.3 38.7 60.0 51.4 55.5 70.7 63.0 58.3 40.6 34.1 49.4 Delhi 20.3 29.6 35.4 26.6 33.6 63.1 20.0 29.1 34.7 23.0 34.5 37.9 31.4 46.4 71.6 22.6 33.0 36.8 Rajasthan 48.6 47.7 46.1 54.5 53.2 51.7 31.9 23.0 30.5 56.3 53.6 54.3 62.1 59.0 61.5 40.7 29.2 35.7 Uttar Pradesh/ Uttaranchal 44.2 40.2 33.7 46.7 43.5 36.8 36.5 27.4 24.1 54.3 45.1 42.9 57.4 48.6 45.3 44.8 32.2 36.6 Bihar/ Jharkhand 36.0 29.9 21.5 36.3 32.4 21.8 34.0 23.6 20.1 44.2 33.1 23.8 44.7 35.6 25.5 42.4 27.2 14.8 Assam 53.9 44.5 29.7 55.4 48.7 28.3 46.3 42.0 36.5 66.1 48.9 38.2 67.7 54.3 40.6 58.0 45.8 28.8 West Bengal 61.3 52.1 48.8 64.7 54.7 49.1 53.8 46.3 47.1 63.8 57.1 52.6 68.3 61.3 54.6 56.5 49.6 48.3 Odisha 24.3 31.3 29.5 25.2 32.5 31.8 19.9 26.9 22.3 27.5 57.5 26.5 27.6 38.3 27.5 26.9 32.2 24.3 Madhya Pradesh/ Chhattisgarh 38.7 33.4 28.5 42.7 39.3 29.7 28.3 22.4 23.4 46.6 39.2 39.1 50.3 44.6 43.4 37.7 29.4 21.4 Gujarat 29.1 31.2 23.0 32.0 35.5 27.9 24.8 18.3 16.6 40.0 42.0 40.5 44.7 47.4 47.5 32.5 23.3 29.2 Maharashtra 16.2 12.4 13.8 19.2 11.7 11.1 13.0 13.6 16.6 21.1 45.1 22.3 24.9 18.7 21.5 16.6 16.7 23.1 Andhra Pradesh 43.5 33.4 34.3 47.0 35.1 35.0 35.0 27.1 32.3 51.6 41.8 48.5 55.0 44.3 51.0 41.9 32.0 41.3 Karnataka 34.0 28.8 28.6 35.8 29.7 32.6 30.9 25.1 22.4 42.7 37.7 39.0 45.3 39.7 46.2 38.3 27.9 26.4 Goa 20.3 12.0 9.2 23.7 11.4 9.0 17.5 13.4 9.2 34.3 21.1 15.8 43.4 21.8 17.9 27.0 19.6 15.5 Kerala 32.0 39.5 32.4 33.8 42.4 35.0 27.2 27.3 25.4 38.2 47.9 37.1 41.0 51.1 38.9 31.0 34.6 34.2 Tamil Nadu 25.8 21.9 22.1 25.2 22.5 22.3 26.4 20.9 22.0 28.3 26.4 30.0 28.6 27.7 29.3 28.0 23.4 29.1 North East 54.1 52.7 57.2 55.7 53.3 60.7 49.2 50.3 47.2 63.6 57.5 62.2 65.8 58.1 67.0 57.6 55.3 48.6 INDIA TOTAL 37.3 34.6 29.8 41.1 37.2 32.1 29.2 28.3 24.9 43.7 40.5 38.2 47.9 43.8 41.6 35.1 32.5 30.8 Women INDIA TOTAL 1.4 2.2 2.7 1.7 1.6 3.4 0.9 0.9 0.3 2.4 3.7 5.0 2.8 2.8 6.2 1.4 1.4 0.8 * Note: The 2004 estimates are age-specific prevalence estimates from SRSBL. SRSBL provided appropriate age specific patterns that are close to 2011 census population which is used as the standard population for age standardized rates using SFMS and GATS.

Web Table 2c: The 1998 and 2010 age standardized smoking prevalence for ages 60-69 years and 70+ years, by state and location of residence.

60-69 years 70+ years Total Rural Urban Total Rural Urban States/ Regions 1998 2010 1998 2010 1998 2010 1998 2010 1998 2010 1998 2010 Men Himachal Pradesh 60.5 41.6 62.4 47.4 39.3 8.6 54.7 47.2 56.5 49.4 30.3 0.0 Punjab 10.3 21.6 10.6 25.6 9.6 13.6 8.6 12.2 9.3 16.5 6.6 7.1 Chandigarh 18.6 29.4 24.2 66.7 18.1 26.9 14.6 28.4 26.6 100 13.6 3.9 Haryana 62.6 52.2 72.7 60.6 34.3 46.0 54.8 56.1 64.5 66.8 27.7 16.9 Delhi 17.5 57.8 35.2 35.4 16.7 57.4 11.8 0.0 32.1 0.0 10.9 0.0 Rajasthan 55.0 30.2 59.3 35.8 40.9 18.8 44.6 30.2 50.2 33.3 30.0 42.4 Uttar Pradesh/ Uttaranchal 50.5 45.7 52.9 49.7 40.6 30.3 40.6 32.2 43.1 30.0 31.3 45.1 Bihar/ Jharkhand 42.1 24.4 42.4 25.5 40.6 17.5 33.3 24.8 33.9 24.8 29.2 25.6 Assam 62.9 33.8 65.9 34.9 46.5 35.7 55.3 33.3 58.0 36.4 41.5 0.0 West Bengal 53.2 42.8 59.5 38.7 41.9 48.9 33.4 28.0 39.7 35.3 24.8 16.9 Odisha 23.0 14.7 23.1 14.4 22.1 21.3 17.8 15.6 18.6 20.4 13.1 0.0 Madhya Pradesh/ Chhattisgarh 42.9 37.0 45.3 34.7 34.4 49.4 31.2 31.9 35.4 36.3 19.3 7.4 Gujarat 40.5 35.4 44.2 46.3 32.5 22.5 33.1 47.1 37.6 58.4 23.8 30.1 Maharashtra 19.0 20.2 23.2 24.5 12.7 15.3 13.3 16.9 17.0 21.1 8.4 7.9 Andhra Pradesh 44.8 51.8 48.2 57.7 32.2 36.0 34.3 39.8 37.5 40.7 22.5 37.4 Karnataka 36.0 26.0 38.5 31.1 31.6 18.4 27.0 45.6 30.1 56.6 22.0 16.5 Goa 34.5 10.9 41.5 14.7 27.9 5.0 26.5 3.6 29.8 5.3 23.2 0.0 Kerala 29.5 31.2 31.2 37.8 24.8 12.2 18.4 29.3 19.0 31.2 16.4 22.6 Tamil Nadu 19.9 24.5 20.9 30.5 18.4 19.0 11.4 29.2 12.1 41.6 10.5 16.2 North East 60.2 56.8 62.3 61.3 53.1 41.1 52.7 55.0 54.6 60.9 46.7 17.7 INDIA TOTAL 39.1 33.4 43.0 36.8 28.7 24.8 30.0 30.0 34.1 33.8 19.9 18.7 Women INDIA TOTAL 3.0 8.7 3.5 10.8 1.8 1.1 2.8 10.6 3.3 12.8 1.6 1.5

Web Table 3: The 1998 and 2010 age standardized rates and age standardized rates ratios of smoking among men aged 15-69 years, by education level and type of product.

Smoke Type Illiterate or no formal education Less than 10th grade 10th grade and above 1998/2010 ASRR 1998/2010 ASRR 1998/2010 ASRR Any Smoking 15-29 years 13.9/22.1 1.6(1.1-2.0) 9.4/14.5 1.5(1.4-1.7) 4.9/7.5 1.5(1.3-1.7) 30-44years 46.5/39.9 0.9(0.8-0.9) 36.0/32.2 0.9(0.8-0.95) 25.5/20.1 0.8(0.7-0.9) 45-59years 51.4/47.9 0.9(0.9-1.0) 41.7/41.2 1.0(0.9-1.1) 30.1/23.1 0.8(0.7-0.9) 60-69years 45.1/40.9 0.9(0.8-1.0) 34.6/32.1 0.9(0.8-1.1) 22.9/20.2 0.9(0.6-1.1) 15-69years 33.9/35.4 1.0(1.0-1.1) 25.1/26.1 1.0(1.0-1.1) 17.8/15.2 0.9(0.8-0.9) 70+years 35.2/35.5 1.0(0.9-1.2) 24.2/28.9 1.2(1.0-1.4) 15.5/11.2 0.7(0.3-1.0) Cigarettes 15-29 years 1.6/10.8 6.8(2.9-9.8) 2.0/8.1 4.1(3.6-4.5) 2.0/6.2 3.1(2.6-3.4) 30-44years 4.3/10.7 2.5(2.1-2.8) 7.8/14.1 1.8(1.6-2.0) 12.5/14 1.1(1.0-1.2) 45-59years 3.7/11.8 3.2(2.4-3.8) 8.2/15.0 1.8(1.6-2.0) 15.3/14.7 1.0(0.8-1.1) 60-69years 2.2/9.3 4.2(3.0-5.1) 5.3/9.3 1.8(1.2-2.2) 9.2/11.8 1.3(0.9-1.6) 15-69years 2.9/10.4 3.6(3.0-4.0) 5.1/11.2 2.2(2.1-2.3) 8.5/10.6 1.3(1.2-1.3) 70+years 1.6/6.8 4.4(2.7-5.6) 3.2/8.4 2.7(1.6-3.5) 5.9/3.8 0.6(0.1-0.99) Bidis 15-29 years 10.9/14 1.3(0.9-1.6) 6.6/8.1 1.2(1.0-1.5) 2.4/1.9 0.8(0.5-0.99) 30-44years 37.2/33.1 0.9(0.8-1.0) 25.4/22.7 0.9(0.8-0.97) 11.0/7.8 0.7(0.6-0.8) 45-59years 40.2/38.7 1.0(0.9-1.1) 29.4/30.2 1.0(0.9-1.1) 12.2/11.1 0.9(0.7-1.1) 60-69years 32.9/33.4 1.0(0.8-1.2) 24.4/25.8 1.1(0.9-1.2) 10.4/11 1.0(0.5-1.5) 15-69years 26.6/27.9 1.0(1.0-1.1) 17.7/18.0 1.0(1.0-1.1) 7.8/6.1 0.8(0.7-0.9) 70+years 22.4/26.8 1.2(1.0-1.4) 15.9/21.1 1.3(1.0-1.6) 6.9/7.4 1.1(0.2-1.6)

Web Figure 1: Age-standardized rate (ASR) of smoking among men by age group and product for 1998-2010 and number of smokers in 2015 (in millions) in rural India.

Note: * ASRR=ASR GATS/ ASR SFMS (99% CI)

Web Figure 1: Age-standardized rate (ASR) of smoking among men by age group and product for 1998-2010 and number of smokers in 2015 (in millions) in urban India.

Note: * ASRR=ASR GATS/ ASR SFMS (99% CI)