102482 Research Smoking-attributable mortality in Bangladesh: proportional mortality study Dewan S Alam,a Prabhat Jha,b Chinthanie Ramasundarahettige,b Peter Kim Streatfield,a Louis W Niessen,a Muhammad Ashique H Chowdhury,a Ali T Siddiquee,a Shyfuddin Ahmeda & Timothy G Evansc Objective To directly estimate how much smoking contributes to cause-specific mortality in Bangladesh. Methods A case–control study was conducted with surveillance data from Matlab, a rural subdistrict. Cases (n = 2213) and controls (n = 261) were men aged 25 to 69 years who had died between 2003 and 2010 from smoking-related and non-smoking-related causes, respectively. Cause-specific odds ratios (ORs) were calculated for “ever-smokers” versus “never-smokers”, with adjustment for education, tobacco chewing status and age. Smoking-attributable deaths among cases, national attributable fractions and cumulative probability of surviving from 25 to 69 years of age among ever-smokers and never-smokers were also calculated. Findings The fraction of ever-smokers was about 84% among cases and 73% among controls (OR: 1.7; 99% confidence interval, CI: 1.1–2.5). ORs were highest for cancers and lower for respiratory, vascular and other diseases. A dose–response relationship was noted between age at smoking initiation and daily number of cigarettes or bidis smoked and the risk of death. Among 25-year-old Bangladeshi men, 32% of ever-smokers will die before reaching 70 years of age, compared with 19% of never-smokers. In 2010, about 25% of all deaths observed in Bangladeshi men aged 25 to 69 years (i.e. 42 000 deaths) were attributable to smoking. Conclusion Smoking causes about 25% of all deaths in Bangladeshi men aged 25 to 69 years and an average loss of seven years of life per smoker. Without a substantial increase in smoking cessation rates, which are low among Bangladeshi men, smoking-attributable deaths in Bangladesh are likely to increase. Introduction The objective of this study was to assess the effects of smoking on cause-specific mortality among Bangladeshi men Deaths attributable to smoking are projected to increase between the ages of 25 and 69 years. We conducted a retro- substantially throughout the 21st century and much of the spective case–control study using data on cause of death and increase will occur in low- and middle-income countries1 such smoking status for all men in this age group who died in the as Bangladesh, whose population of 150 million makes it the subdistrict of Matlab in rural Bangladesh between 2003 and seventh most populous country in the world. More than half 2010. Although tobacco chewing is common in Bangladesh, of Bangladeshi men over the age of 25 years smoke cigarettes its effects will be reported in a separate paper since chewing or bidis, small handmade cigarettes containing about one causes fewer diseases than smoking, most notably oral cancer.7 fourth the amount of tobacco found in cigarettes.2 A nation- ally representative case–control study in neighbouring India showed that in 2010 smoking caused about 20% of all deaths Methods among males aged 30 to 69 years.3 Smoking cessation rates Study design are relatively low in Bangladesh and India,2 but Bangladeshi men are, on average, younger than Indian men when they start As part of the INDEPTH Network, the International Centre for smoking and they smoke more cigarettes or bidis daily than Diarrhoeal Disease Research, Bangladesh (icddr,b), has main- Indian men.2,4 Bidis account for most of the tobacco smoked tained a comprehensive Health and Demographic Surveillance in India, but in Bangladesh cigarettes represent about half of System (HDSS) in Matlab, in the district (zilla) of Chandpur, all the tobacco smoked. since 1966.8,9 Matlab is a rural area (upazila) located about According to recent studies in high-income countries, 55 km south-east of Dhaka, the capital city. It covers 184 km2 men and women who start smoking as young adults and do not and has 142 villages, each with about 1500 people. The total quit have a threefold risk of dying relative to those who have population of Matlab is about 225 000. To track fatal events never smoked.5,6 Prospective studies are required to determine as part of the HDSS, trained field research assistants visit whether the same extreme risks hold true for Bangladesh and households where a death has occurred from 6 to 12 weeks other low- and middle-income countries. Of the estimated 21.9 after the death and administer a structured verbal autopsy million smokers in Bangladesh, 21.2 million are males and questionnaire to any relative who lived with the deceased. only 0.7 million are females. Thus, only the effects of smoking The purpose is to obtain from these respondents information among men can be reliably studied at present. on the symptoms, signs and medical details surrounding the a Centre for Control of Chronic Diseases, International Centre for Diarrhoeal Disease Research Bangladesh, 68 Shaheed Tajuddin Ahmed Sarani, Mohakhali, Dhaka 1212, Bangladesh. b Centre for Global Health Research, St Michael’s Hospital at University of Toronto, Toronto, Canada. c Human Development Network, The World Bank, Washington, United States of America. Correspondence to Dewan S Alam (e-mail: dsalam@icddrb.org). (Submitted: 26 February 2013 – Revised version received: 12 June 2013 – Accepted: 13 June 2013 – Published online: 12 July 2013 ) Bull World Health Organ 2013;91:757–764 | doi: http://dx.doi.org/10.2471/BLT.13.120196 757 Research Smoking-attributable mortality in Bangladesh Dewan S Alam et al. death. Field staff are instructed not to consumption history of the deceased in- 1 try to arrive at a diagnosis of the cause dividual. Smoking history included the 1− (1) of death. Instead, causes of death are type of tobacco smoked (e.g. cigarette RR assigned by two trained physicians at or bidi tobacco [combined into a single icddr,b in accordance with the Interna- question]; tobacco for pipe or hookah) where RR represents the ratio of the odds tional statistical classification of diseases or chewed (e.g. betel leaf with or with- of death in ever-smokers to the odds of and related health problems, tenth revi- out tobacco, tobacco powder, rolled death in never-smokers adjusted for age, sion (ICD-10) and using standardized tobacco); the amount smoked; the age education and chewing tobacco status. To disease coding guidelines developed for (in years) at initiation and cessation of calculate the national attributable fractions other INDEPTH sites.8,10 smoking; and the duration of tobacco we used the following standard formula: smoking or chewing. The field ques- Subjects tionnaire did not separate the number Cases were men aged 25 to 69 years of of cigarettes from the number of bidis age who had died from causes strongly smoked, so we treated them together in p(RR − 1) associated with tobacco smoking:1,7,11 the analysis. We excluded 156 men from (2) cancers of the lung, mouth and larynx the analysis because 34 of them drank 1 + p(RR − 1) (ICD-10 codes: C00–14, C32–34); can- alcohol, 11 smoked only pipe and the cers of the digestive organs (C16–26, remainder had missing information on with the RRs derived from the Matlab C48, D01, D12–13); all other cancers smoking status, education or cause of results and smoking prevalence ( p ) (rest of C00-D48, excluding C60–63 and death. We compared “ever-smokers” – from the GATS survey in 2009.2 About C69–72); chronic lower respiratory dis- i.e. people who smoked at the time of 80% of the Bangladeshi population eases (J40–47); pulmonary tuberculosis the survey (“current smokers”) or who lives in rural areas similar to Matlab, (A15–19); all other respiratory diseases had smoked in the past (“former smok- so the use of the RRs estimated in this (rest of J00–99); stroke (I60–64); isch- ers”) – with “never-smokers”. This was study does not introduce a major bias aemic heart disease (I20–25); all other appropriate because most former smok- in the calculation of overall national vascular diseases (rest of I00–99); cir- ers (214/324) had smoked for more attributable fractions. In Bangladesh, rhosis (K70–77); and other medical than 10 years. Moreover, excess hazard national cause-of-death statistics are causes (rest of A00-R99, excluding the depends strongly not only on recent based on self-reporting, which can be diseases not associated with tobacco smoking habits but also on smoking unreliable.14 For this reason, to generate smoking, which were the causes of death habits in early adult life.1,7,11 national cause-of-death estimates we among controls). Controls were men Statistical methods applied, to the total deaths reported by who had died of causes not associated the United Nations for Bangladesh, the with tobacco, such as intestinal infec- We used proportional mortality to relate proportions of the major causes of death tions (A00–09); other bacterial diseases bidi or cigarette smoking to various found in Matlab.15 In Chandpur zilla, (A30–99); viral infections (B01–77); specific and general medical causes of the rates of all-cause mortality among cancers of the male genital organs death.12,13 The assumption behind pro- adult males are comparable to the rates (C60–63); cancers of the brain and eye portional mortality is that the smoking observed in the whole of Bangladesh, (C69–72); endocrine, nutritional and patterns among the deceased in the con- but child mortality rates are lower.14 We metabolic diseases (E00–07, E25–35, trol group are similar to those observed calculated the cumulative probability of E40–46, E50–64, E70–90); mental in the general population. However, in surviving from 25 until 69 years of age and behavioural disorders (F00–99); reality smoking prevalence was lower for ever-smokers and never-smokers, diseases of the nervous system (G00– among the general population of men with adjustment for any differences in 99), diseases of the eye and adnexa surveyed in 2009,2 than among our de- age, education and use of chewing to- (H00–95); diseases of the appendix and ceased controls, perhaps because smok- bacco. To do so, we combined the RRs hernias (K35–40); diseases of the skin ing caused some excess deaths among estimated from Matlab with national (L00–99); diseases of the musculoskel- controls. Thus, the net effect would be an smoking prevalence1 and age-specific etal system (M00–99); diseases of the underestimation of the differences be- Bangladeshi death rates as reported by genitourinary system (N00–99), and tween ever-smokers and never-smokers. the United Nations,15 following methods injuries (T00-Y99). Other reporting biases should equally described in previous studies.3,6 We excluded women from the affect the cases and controls.12 analyses because the smoking preva- We calculated cause-specific odds lence among Bangladeshi women is very ratios (ORs) (as approximations of rela- Results low. According to the nationally repre- tive risks [RRs]) for ever-smokers versus The Matlab HDSS recorded a total of sentative Global Adult Tobacco Survey, never-smokers using logistic regression, 9708 deaths among individuals aged or GATS, in 2009 smoking prevalence with adjustment for educational status 20 years or older (5296 males, 4412 fe- among living Bangladeshi women 15 (illiterate, below secondary school, males). Of these deaths, 2474 occurred years of age or older was only about secondary school or above), history of in men aged 25 to 69 years. Table 1 1.5%.2 It was somewhat higher, at 5.7%, tobacco chewing (yes, no) and age (in presents data for the 2213 cases and 261 among deceased women aged 20 years continuous years). We calculated the controls. Cases were older than controls of age or older in our study. smoking-attributable deaths among on average, whereas both groups had The field questionnaire contained cases by multiplying the number of ever- similar educational levels and similar questions on the smoking and alcohol smokers among the cases by: rates of use of chewing tobacco. Among 758 Bull World Health Organ 2013;91:757–764 | doi: http://dx.doi.org/10.2471/BLT.13.120196 Research Dewan S Alam et al. Smoking-attributable mortality in Bangladesh Table 1. Demographic characteristics of cases and controls in study on the effects of the deceased, smoking prevalence was smoking on cause-specific mortality, Matlab, Bangladesh highest among those who had been il- literate or between the ages of 40 and 60 Characteristic No. (%) of casesa No. (%) of controlsa years when they died (data not shown). (n = 2 213) (n = 261) About 84% (1855/2213) of the cases were ever-smokers (Table 1), as Age at death (years) compared with 73% (191/261 data not 25–29 27 (1.2) 11 (4.2) shown) of the controls. The OR for all 30–39 110 (5.0) 39 (14.9) causes of death, representing the odds 40–49 330 (14.9) 46 (17.6) in ever-smokers versus the odds in 50–59 580 (26.2) 66 (25.3) never-smokers, was 1.7 (99% confidence 60–69 1166 (52.7) 99 (37.9) interval, CI: 1.1–2.5) after adjusting for Education any differences in age, education and Illiterate 1082 (48.9) 134 (51.3) use of chewing tobacco (Table 2). This Below secondary 1031 (46.6) 117 (44.8) corresponds to 31% of all 2213 deaths Secondary or higher 100 (4.5) 10 (3.8) among the cases, equivalent to an excess Tobacco chewing of 692 deaths among smokers between Yesb 1071 (48.4) 108 (41.4) the ages of 25 and 69 years in Matlab. About 94% of the men who died from No 1142 (51.6) 153 (58.6) cancers of the lung, oral cavity or larynx Smoking status were ever-smokers (162/172); 103 men Current smokerc 1557 (70.4) 165 (63.2) died from lung cancer and of these Former smoker 298 (13.5) 26 (10.0) men, 97 were ever-smokers. The ORs Never-smoker 358 (16.2) 70 (26.8) were notably lower for digestive organ a Cases and controls were men aged 25 to 69 years who had died between 2003 and 2010 from smoking- cancers than for other cancers. The ORs related and non-smoking-related causes, respectively. and the proportions of deaths due to b Includes current and former users of chewing tobacco. c “Current smoker” applies to those who quit smoking within 5 years of their death. Current and former smoking were highest for cancers; they smokers combined represent ever-smokers. were lower or similar for respiratory Table 2. Deaths among ever-smokers and odds of dying among ever-smokers versus never-smokers, by cause, and smoking-associated deaths among men aged 25 to 69 years, Matlab, Bangladesh, 2003–2010 Cause of death No. of deaths in No. (%) of ever-smokers ORb (99% CI) No. (%) of excess 2003–2010 among the deceaseda deaths due to smoking Cancer of lung, oral cavity or larynx 172 162 (94.2) 5.17 (2.0–13.4) 131 (76) Cancers of digestive organs 182 141 (77.5) 1.19 (0.6–2.2) 23 (12) Other cancers 75 64 (85.3) 1.85 (0.7–4.8) 29 (39) All cancersc 429 367 (85.5) 1.94 (1.1–3.3) 183 (43) Chronic lower respiratory diseases 172 146 (84.9) 1.44 (0.7–2.9) 45 (26) Pulmonary tuberculosis 78 67 (85.9) 1.82 (0.7–4.7) 30 (39) Other respiratory diseases 61 52 (85.2) 1.62 (0.6–4.7) 20 (33) All respiratory diseases 311 265 (85.2) 1.58 (0.9–2.8) 95 (30) Stroke 193 157 (81.3) 1.42 (0.8–2.6) 46 (24) Ischaemic heart disease 368 311 (84.5) 1.94 (1.1–3.3) 151 (41) Other vascular diseases 465 387 (83.2) 1.48 (0.9–2.5) 126 (27) All vascular diseases 1026 855 (83.3) 1.66 (1.1–2.6) 323 (31) Cirrhosis 121 100 (82.6) 1.67 (0.8–3.6) 0 (0) Other medical causesd 233 189 (81.1) 1.46 (0.8–2.6) 60 (26) Ill-defined medical causes 93 79 (84.9) 1.66 (0.7–3.9) 31 (34) All medical causese 2213 1855 (83.8) 1.68 (1.1–2.5) 692 (31) CI, confidence interval; OR, odds ratio. a Former and current smokers make up the category of “ever-smokers”. b ORs represent odds of death from specific cause among ever-smokers versus never-smokers, adjusted for age, education and tobacco chewing status. c Except those of the male genital organs, brain and eye. d All other medical deaths excluding cancer, respiratory, vascular and some deaths from causes not associated with tobacco considered as controls in this study. e Deaths from all causes except injury and others not associated with tobacco smoking. Note: See methods for ICD 10 codes used to define the cases. The 261 controls include all injury deaths and deaths that are not attributable to tobacco smoking (see methods for ICD-10 codes). Bull World Health Organ 2013;91:757–764 | doi: http://dx.doi.org/10.2471/BLT.13.120196 759 Research Smoking-attributable mortality in Bangladesh Dewan S Alam et al. diseases, vascular diseases and other Fig. 1. Odds ratios (ORs) and excess deaths due to smoking, by age of smoking initiation medical causes of death. and daily number of cigarettes or bidis smoked by men aged 25 to 69 years, In Bangladesh, men aged 25 to 69 Bangladesh, 2003–2010 years initiate smoking at an average age of 18.6 years (Appendix A, available at: www.cghr.org/tobacco) and nearly ORa (99%CI) No. (%) of smoking- all smokers report having initiated associated excess deaths smoking before the age of 25 years.2 Age started smoking As shown in Fig. 1, in our study, those 25 years or later (30)b 1.4 (0.9–2.2) 219 (25) who started smoking before this age Before 25 years (21)b 1.9 (1.2–3.0) 367 (42) had greater odds of dying than never- smokers (OR: 1.9) and than those who Quantity of cigarettes/bidis smoked per day started smoking after the age of 25 1 to 9 (6)c 1.3 (0.8–2.2) 121 (21) years (OR: 1.4; χ2 test for trend 12.1; P < 0.0001). The mean daily number of 10 to 19 (12)c 1.6 (1.0–2.5) 261 (32) cigarettes or bidis smoked by males who 20 to 50 (23)c 2.7 (1.4–5.3) 219 (59) died was 11. A dose–response analysis (comparing 1757 ever-smokers with 0 1 2 3 4 438 never-smokers for whom informa- Not caused by smoking Excess risk from smoking 99% confidence interval tion on smoking amount was avail- able) showed that those smokers who a ORs represent the odds of dying among ever-smokers relative to the odds among never-smokers, adjusted for age, education and tobacco chewing status. consumed from 20 to 50 (mean: 23) b The figure in parentheses is the median age of smoking initiation. cigarettes or bidis per day had a much c The figure in parentheses is the mean number of cigarettes or bidis smoked. higher odds of dying (OR: 2.7) when compared with never-smokers, than attribute to smoking any deaths from differences in the risk of death between those who consumed 10 to 19 cigarettes cirrhosis, ill-defined conditions or in- ever-smokers and never-smokers were or bidis (mean: 12) daily (OR: 1.6) or jury. Of the smoking-attributable deaths, seen among those as young as 60 years than those who smoked 1 to 9 (mean: about 50% (21 700/42 800) were caused – 14% dead by that age among smokers; 6) cigarettes or bidis a day (OR: 1.3; χ2 by vascular disease and about 26% by only 8% dead among never-smokers. test for trend 16.3; P < 0.001). cancer (11 400). These data suggest that the average Applying the Matlab relative risk The cumulative probability of dying 25-year-old Bangladeshi male smoker to the national death totals suggests between the ages of 25 and 69 years was currently loses an average of seven years that about 42 800 Bangladeshi men much higher for ever-smokers than for of life because of smoking. aged 25 to 69 years died in 2010 from never-smokers (Fig. 2). At this death smoking-attributable diseases. This was rate, 32% of 25-year-old Bangladeshi equivalent to about 25% of the 172 200 male ever-smokers would die before Discussion deaths from all diseases in men in this the age of 70 years, versus only 19% Bangladeshi men between the ages of 25 age group in 2010 (Table 3). We did not of 25-year-old never-smokers. Marked and 69 years who have smoked at some Table 3. Estimated smoking-attributable deaths, by cause, among men aged 25 to 69 years and population-attributable fraction, Bangladesh, 2010 Cause of death No. of deaths in All Bangladesh, 2010 2003–2010 Smoking- attributable Total deaths Population-attributable deaths (thousands)a (thousands)b fraction (%) All vascular diseases 1026 21.7 71.4 30.3 All cancersc 429 11.4 29.9 38.3 Respiratory diseases except tuberculosis 233 4.0 16.2 24.4 Other medical causesd 233 3.8 16.2 23.3 Pulmonary tuberculosis 78 1.9 5.4 35.1 Cirrhosis 121 0.0 8.4 0.0 Ill-defined cause 93 0.0 6.5 0.0 Injury/other cause not associated with tobacco 261 0.0 18.2 0.0 Total 2474 42.8 172.2 24.8 a Smoking prevalence of 66% noted in the GATS2 was used to calculate smoking attributable deaths. b Cause-specific total deaths were estimated by multiplying total deaths as reported by the United Nations by proportions of cause-specific deaths in study. c Except those of the male genital organs, brain and eye. d All other deaths from medical causes except cancer, respiratory diseases, vascular diseases and deaths not associated with tobacco smoking (considered as controls in this study). 760 Bull World Health Organ 2013;91:757–764 | doi: http://dx.doi.org/10.2471/BLT.13.120196 Research Dewan S Alam et al. Smoking-attributable mortality in Bangladesh prompt short-course tuberculosis Fig. 2. Cumulative probability of dying between the ages of 25 and 69 years among male ever-smokers and never-smokers in Bangladesh, at 2010 death rates treatment is available in Matlab.16 Bangladeshi men have higher smoking cessation rates than Indian 40 men but initiate smoking at a younger age and smoke more sticks on a given day.2,4 Smoking may well have caused some of the deaths we defined as our 32% controls, such as deaths from diabetes. 30 Thus, we may be underestimating the true risk of death from smoking in Cumulative probability of death (%) Bangladeshi men. Moreover, there was a higher proportion of ever-smokers among our controls (73%) than among the males of the same age in the GATS 20 19% (66%).2 7 years One of the strengths of the propor- tional mortality method is that most 14% biases, such as recall bias, would apply to cases and controls equally.12 Indeed, we 10 observed few differences between cases and controls in terms of education and 8% 6% tobacco chewing. Few Bangladeshis re- port drinking alcohol17 and we excluded 4% any self-reported drinkers. However, the 121 deaths from cirrhosis suggest that in 0 Bangladesh drinking is underreported 30 40 50 60 70 for cultural reasons. The elevated risk Age (years) of dying from cirrhosis observed in Ever-smoker Never-smoker ever-smokers relative to never-smokers Note: The horizontal arrow shows that the death rates experienced by never-smokers at the age of 70 suggests that most of the heavy alcohol years are experienced seven years earlier by ever-smokers. Thus, smokers lose seven years of life on drinkers also smoked and vice versa. The average. This combines a loss of zero years among smokers not killed by smoking with a loss of far more combination of drinking and smoking than seven years among those killed by smoking. should also exist among controls, many of whom died of injuries. Thus, we might point have a risk of dying of a tobacco- The ORs for ever-smokers versus be slightly underestimating the risks related disease that is 70% higher than never-smokers are slightly higher of smoking. However, the exclusion of the risk among never-smokers of similar among Bangladeshi men than among men who drank alcohol does not explain age, educational level and tobacco chew- Indian men (OR: 1.6). 3 The propor- the marked differences in mortality ing status. Our cases were men who tion of all deaths in men aged 25 to observed between ever-smokers and had died from diseases that have been 69 years that is due to smoking is also never-smokers. Deaths in the control causally associated with smoking in higher in Bangladesh (25%) than in group (n = 261) were fewer than ideal. other studies1,3,5-7,11-12 and we observed a India (20%). Subtle but potentially Consequently, the ORs associated with well-defined dose–response relationship important differences in smoking specific diseases had wide CIs. However, between age at smoking initiation and patterns between the two countries the cause-specific ORs in our study are the number of cigarettes or bidis smoked might have an influence on the risk consistent with those seen in India3,18,19 and the risk of dying. Admittedly, how- of specific diseases. Bangladeshi bidis and other Asian countries.12 Similarly, ever, not all smokers died of diseases are wrapped in cigarette paper rather some deaths may have been assigned to associated with tobacco-attributable than the tendu leaf commonly used to the wrong cause on verbal autopsy.20,21 diseases and not all such diseases oc- wrap Indian bidis. Cigarette smoking This would tend to raise the ratio of the curred among smokers. Thus, smoking is much more common among Ban- risk in smokers to the risk in non-smok- is an important cause of most, but not gladeshi males than Indian males. 2,4 ers in the case of some diseases and to all, of the excess deaths among smok- In our study, lung cancer caused more lower this ratio in the case of others. The ers in Bangladesh. Cumulative survival than 10% of the deaths among smok- proportion of vascular deaths among all analysis revealed that a typical Ban- ers in Bangladesh (5300/42 800) but a deaths in men in Matlab (about 41%) gladeshi smoker currently loses about much smaller proportion of the deaths is higher than the proportion seen in seven years of life because of smoking. among smokers in India.3 By contrast, crude national cause-of-death patterns This combines an average loss of zero pulmonary tuberculosis accounted for in Bangladesh or the proportion seen in years of life for some smokers not killed only 4% (2000) of the deaths among India in men of comparable age.3,14 Thus, by smoking with a loss far in excess of smokers in our study but for a much we might be overestimating the absolute seven years in some smokers who were higher proportion of the deaths in number of smoking-attributable deaths killed by smoking. Indian smokers.3 This may be because from vascular disease (but not the odds). Bull World Health Organ 2013;91:757–764 | doi: http://dx.doi.org/10.2471/BLT.13.120196 761 Research Smoking-attributable mortality in Bangladesh Dewan S Alam et al. However, the computed ORs and the ab- times as high – as ever-smokers and live, on tobacco advertising and promotion, solute number of smoking-attributable on average, one decade less than never- and expanded access to smoking ces- deaths from any cause are affected less, if smokers. 5,6 Indeed, Indian men who sation services are other interventions at all, by the misclassification of specific smoke cigarettes in these large quantities that can be implemented in Bangladesh diseases. Most of the misclassification on already appear to be losing a decade of to encourage men to stop smoking and verbal autopsy is confined to people who life.3 Of Bangladeshi men aged 45 to 64 deter women and youth from taking up die in old age.20 Our study focused on years, only 15% are former smokers and smoking.1,23 ■ men aged 25 to 69 years and fewer than 63% are current smokers.2 By contrast, 4% of the deaths among these men were in the United States of America, where Acknowledgements assigned to ill-defined causes. smoking cessation has become common, This paper is dedicated to the late Sir Bangladesh already has about 20 former smokers are about three times Richard Doll (1912–2005), who would million male smokers and it is likely that more numerous than current smokers have turned 100 on 28 October 2012. they will experience a loss in lifespan among men this same age.6 The opinions expressed here are those of more than the seven years that we About one quarter of the deaths of the authors and do not necessarily estimated in this study. Bangladeshi men in Bangladeshi men between the ages represent those of the institutions where still smoke fewer cigarettes per day and of 25 and 69 years are due to smok- the authors are employed. initiate smoking at a later age than men ing. Tobacco control in Bangladesh is in most high-income countries. Urban still at an early stage; the Smoking and Funding: Funding for this study was pro- male smokers in Bangladesh consume Tobacco Products Usage (Control) Act, vided by the Fogarty International Centre more cigarettes than bidis and smoke which was passed in 2005, has been of the US National Institutes of Health more per day than rural smokers.2 Re- implemented only partially. 22 Efforts (grant TW007939-01), the Canadian cent increases in income in Bangladesh to accelerate tobacco control in Ban- Institute of Health Research (IEG-53506), have also made smoking relatively more gladesh are warranted. Cigarette prices the Bill & Melinda Gates Foundation affordable.22 For cultural reasons, few in the country are among the lowest in (Grant 51447) and the Oxford Health women smoke in Bangladesh, but there the world; bidis are even cheaper than Alliance Vision 2020 (Grant 5444447). is no guarantee that this will hold in the cigarettes.22 Hence, the most effective PJ is supported by a University of To- future. Recent prospective studies in way to encourage smoking cessation in ronto Endowed Chair. DA is supported high-income countries have shown that Bangladesh would probably be to im- by icddr,b. male or female lifelong smokers who pose a substantial excise tax on tobacco start smoking as young adults and do not products. Prominent warning labels Competing interests: None declared. quit have an excess risk of dying 200% using graphic images, bans on smok- as high – corresponding to an RR three ing in public places, strict restrictions ‫ملخص‬ ‫ دراسة معدل الوفيات التناسبي‬:‫معدل الوفيات بسبب التدخني يف بنغالديش‬ ‫األرجحية أعىل معدالهتا ألمراض الرسطان واجلهاز التنفيس‬ ‫الغرض التقدير املبارش ملدى مسامهة التدخني يف التسبب يف‬ ‫ ولوحظ وجود عالقة‬.‫السفيل وأمراض األوعية الدموية وغريها‬ .‫الوفيات يف بنغالديش‬ ‫بني االستجابة للجرعة والعمر عند بدء التدخني والعدد اليومي‬ ،‫الطريقة أجريت دراسة حالة مقارنة مع بيانات ترصد من ماتالب‬ ‫ وبني‬.‫للسجائر أو سجائر البيدي التي يتم تدخينها وخطر املوت‬ )2213 = ‫ وكانت احلاالت (العدد‬.‫وهي منطقة فرعية ريفية‬ % 32 ‫ نجد أن‬،‫ عام ًا‬25 ‫الرجال البنغالدشيني البالغني من العمر‬ ‫ إىل‬25 ‫) لرجال ترتاوح أعامرهم من‬261 = ‫والضوابط (العدد‬ ‫ عاما من‬70 ‫من املدخنني الدائمني سوف يموتون قبل بلوغهم سن‬ ‫ ألسباب تتعلق‬2010‫ و‬2003 ‫ عام ًا لقوا حتفهم بني عامي‬69 ‫ ويف عام‬.‫ من الذين ال يدخنون مطلق ًا‬% 19 ‫ مقارنة بنسبة‬،‫العمر‬ ‫ وتم حساب‬.‫بالتدخني وأسباب ال تتعلق بالتدخني عىل التوايل‬ ‫ من كل الوفيات التي متت مالحظتها يف‬% 25 ‫ كانت نسبة‬،2010 ”‫) لدى “املدخنني الدائمني‬ORs( ‫نسب األرجحية حمددة األسباب‬ ‫ عام ًا‬69 ‫ إىل‬25 ‫الرجال البنغالدشيني الذين ترتاوح أعامرهم من‬ ‫ مع إدخال تعديالت خاصة بالتعليم‬،”‫مقابل “غري املدخنني مطلق ًا‬ .‫ حالة وفاة) ترجع إىل التدخني‬42000 ‫(أي‬ ‫ كام تم أيض ًا حساب الوفيات بسبب‬.‫وحالة مضغ التبغ والسن‬ ‫ من حاالت الوفيات‬% 25 ‫االستنتاج يتسبب التدخني يف حوايل‬ ‫ والكسور التي تنسب ألسباب عىل الصعيد‬،‫التدخني بني احلاالت‬ 69 ‫ إىل‬25 ‫يف الرجال يف بنغالديش الذين ترتاوح أعامرهم من‬ ‫ إىل‬25 ‫ واالحتامل الرتاكمي للبقاء عىل قيد احلياة من سن‬،‫الوطني‬ .‫ وإىل خسارة ما متوسطه سبع سنوات من حياة كل مدخن‬،‫عام ًا‬ .‫ عام ًا بني املدخنني الدائمني وغري املدخنني مطلق ًا‬69 ‫ املنخفضة‬،‫وبدون زيادة كبرية يف معدالت اإلقالع عن التدخني‬ 84 ‫النتائج كانت نسب الكسور لدى املدخنني الدائمني حوايل‬ ‫ من املرجح أن تزداد الوفيات بسبب‬،‫بني الرجال يف بنغالديش‬ :1.7 :‫ بني الضوابط (نسبة األرجحية‬% 73‫ بني احلاالت و‬% .‫التدخني يف بنغالديش‬ ‫ وبلغت نسب‬.)2.5 – 1.1 :‫ فاصل الثقة‬،‫ فاصل ثقة‬% 99 762 Bull World Health Organ 2013;91:757–764 | doi: http://dx.doi.org/10.2471/BLT.13.120196 Research Dewan S Alam et al. Smoking-attributable mortality in Bangladesh 摘要 孟加拉国归因于吸烟的死亡率 : 比例死亡率研究 目的 直接估算吸烟对孟加拉国死因别死亡率有多大的 OR 对于癌症来说最高,对于呼吸道、血管和其他疾 贡献。 病来说较低。在开始吸烟年龄以及每日吸食香烟或比 方法 使用来自农村小区 Matlab 的监测数据进行病例 迪烟数与死亡风险之间有明显的剂量 – 反应关系。 对照研究。病例组 (n = 2213) 和对照组 (n = 261) 分别是 在 25 岁孟加拉国男性中,32% 的曾经吸烟者将活不到 2003 年到 2010 年间因吸烟相关和非吸烟相关原因死 70 岁,非吸烟者的这个比例则是 19%。在 2010 年孟 亡的 25 岁到 69 岁的男性。计算“曾经吸烟者”和“非 加拉国有案可查的 25 岁到 69 岁男性的全部死亡(即 吸烟者” 的死因别优势比 (OR),计算中针对教育程度、 4.2 万例死亡)中,大约 25% 可归因于吸烟。 嚼烟草状态和年龄进行调整。还计算了病例中的吸烟 结论 在孟加拉国 25 岁到 69 岁男性的全部死亡中,吸 归因死亡、全国可归因分数以及 25 岁到 69 岁曾经吸 烟造成的死亡约占 25%,每名吸烟者平均损失七年的 烟者和非吸烟者中的累计生存概率。 寿命。如果没有戒烟率的显著提高(这个比例在孟加 结果 曾经吸烟者的分数在病例组中约为 84%,在对照 拉国男性中很低) ,可归因于吸烟的死亡很可能还会 组中为 73%(OR : 1.7 ;99% 置信区间,CI :1.1–2.5)。 增加。 Résumé Mortalité attribuable au tabagisme au Bangladesh: une étude proportionnelle de la mortalité Objectif Estimer la contribution directe du tabagisme à la mortalité les cancers et plus bas pour les maladies respiratoires, vasculaires et par cause au Bangladesh. autres. Une relation dose-réponse a été notée entre l’âge de l’initiation Méthodes Une étude cas-témoins a été menée avec les données de au tabagisme et le nombre de cigarettes ou de « bidi s» fumées chaque surveillance de Matlab, un sous-district rural. Les cas (n = 2213) et les jour et le risque de mortalité. Chez les hommes du Bangladesh de 25 ans, contrôles (n = 261) étaient des hommes âgés de 25 à 69 ans décédés 32% des fumeurs depuis toujours mourront avant l’âge de 70 ans, par entre 2003 et 2010 des suites ou non du tabagisme, respectivement. Les rapport à 19% pour les non-fumeurs. En 2010, environ 25% de tous les rapports des cotes (RC) par cause ont été calculés pour les « fumeurs décès observés chez les hommes du Bangladesh âgés de 25 à 69 ans depuis toujours » par rapport aux « non-fumeurs », avec un ajustement (c’est-à-dire 42 000 décès) étaient attribuables au tabagisme. pour l’éducation, la consommation de tabac à mâcher et l’âge. On a Conclusion Le tabagisme est à l’origine de 25% de tous les décès chez également calculé les décès attribuables au tabagisme parmi les cas, les les hommes du Bangladesh âgés de 25 à 69 ans, et d’une perte moyenne fractions attribuables à la population nationale et la probabilité de survie de 7 années de vie pour les fumeurs. Sans une augmentation importante de 25 à 69 ans chez les fumeurs depuis toujours et chez les non-fumeurs. des taux d’abandon du tabac, qui sont faibles chez les hommes du Résultats La fraction des fumeurs depuis toujours était d’environ 84% Bangladesh, les décès attribuables au tabagisme augmenteront chez les cas et de 73% chez les contrôles (RC: 1,7; intervalle de confiance probablement dans ce pays. de 99%, IC: 1,1–2,5). Les rapports de cotes étaient les plus élevés pour Резюме Сопряженная с курением смертность в Бангладеш: пропорциональное исследование смертности Цель Напрямую оценить влияние курения на показатели доверительный интервал (ДИ) 99%: 1,1–2,5). Наиболее высокие причинно-обусловленной смертности в Бангладеш. риски были отмечены для онкологических заболеваний, Методы Исследование методом «случай-контроль» проводилось наименее высокие — для респираторных, сердечно-сосудистых по данным наблюдений в аграрном подокруге Матлаб. и других заболеваний. Была установлена связь «доза-эффект» Популяцию исследования (n = 2213) и контрольную группу возраста начала курения и количества ежедневно выкуриваемых (n = 261) составили мужчины в возрасте 25-69 лет, умершие сигарет или биди с риском смерти. 32% постоянно курящих в период между 2003 и 2010 годом в силу связанных и не 25-летних бангладешских мужчин не доживут до 70-летнего связанных с курением причин. Были рассчитаны отношения возраста, тогда как среди некурящих эта доля составит 19%. рисков (ОР) по конкретным причинам смерти для «постоянно В 2010 году около 25% всех смертей бангладешских мужчин в куривших» и «никогда не куривших» лиц, с поправками на возрасте 25-69 лет (т. е. 42 000 смертей) были связаны с курением. уровень полученного образования, отношение к жеванию табака Вывод Курение является причиной примерно 25% смертей и возраст. Также были вычислены доля связанных с курением у бангладешских мужчин в возрасте 25-69 лет и снижает смертей среди включенных в исслеование случаев смерти, продолжительность жизни курильщиков в среднем на семь лет. национальные коэффициенты и интегральная вероятность Весьма вероятно, что в отсутствие существенного увеличения выживания лиц в возрасте 25-69 лет для постоянно курящего и темпов отказа от курения (весьма низких среди бангладешских никогда не курившего населения. мужчин) сопряженная с курением смертность в Бангладеш Результаты Доля постоянно куривших составила 84% среди продолжит расти. исследованных случаев и 73% в контрольной группе (ОР:  1,7; Bull World Health Organ 2013;91:757–764 | doi: http://dx.doi.org/10.2471/BLT.13.120196 763 Research Smoking-attributable mortality in Bangladesh Dewan S Alam et al. Resumen Mortalidad atribuible al tabaquismo en Bangladesh: estudio de mortalidad proporcional Objetivo Estimar directamente la medida en la que el tabaquismo razones de posibilidades fueron las mayores en casos de cánceres contribuye a la mortalidad por causa específica en Bangladesh. y menores en casos de enfermedades vasculares, respiratorias y de Métodos Se realizó un estudio de casos y controles con datos de otros tipos. Se señaló una relación dosis-respuesta entre la edad de vigilancia de Matlab, un subdistrito rural. Los casos (n = 2213) y los la iniciación en el tabaquismo y el número diario de cigarrillos o bidis controles (n = 261) se trataron de hombres de edades entre 25 y 69 fumados y el riesgo de muerte. Entre los hombres de Bangladesh de 25 años que habían fallecido entre 2003 y 2010 por causas relacionadas años, el 32% de los fumadores de siempre fallecerán antes de alcanzar y no relacionadas con el tabaquismo, respectivamente. Las razones de los 70 años, comparado con el 19% de los nunca fumadores. En 2010, posibilidades por causa específica se calcularon para los «fumadores aproximadamente el 25% de las muertes observadas en los hombres de de siempre» frente a los «nunca fumadores», ajustando la educación, Bangladesh de entre 25 y 69 años (42 000 muertes) fueron atribuibles el consumo de tabaco de mascar y la edad. También se calcularon al tabaquismo. las muertes atribuibles al tabaquismo entre los casos, las fracciones Conclusión El tabaquismo causa aproximadamente el 25% de todas nacionales atribuibles y la probabilidad acumulada de sobrevivir de los las muertes en los hombres de Bangladesh de entre 25 y 69 años y una 25 a los 69 años de edad entre los fumadores de siempre y los nunca pérdida media de siete años de vida en cada fumador. Sin un aumento fumadores. significativo en las tasas de abandono del tabaquismo, que son bajas Resultados La fracción de fumadores de siempre fue de entre los hombres de Bangladesh, es probable que aumenten las aproximadamente el 84% entre los casos y el 73% entre los controles muertes atribuibles al tabaquismo en Bangladesh. (razón de posibilidades: 1,7; intervalo de confianza 99%: 1,1–2,5). Las References 1. Jha P. Avoidance. Nat Rev Cancer 2009;9:655–64. doi: http://dx.doi. 13. Zaridze D, Brennan P, Boreham J, Boroda A, Karpov R, Lazarev A et al. org/10.1038/nrc2703 PMID:19693096 Alcohol and cause-specific mortality in Russia: a retrospective case-control 2. Global Tobacco Surveillance System. Global Adult Tobacco Survey – study of 48,557 adult deaths. Lancet 2009;373:2201–14. doi: http://dx.doi. Bangladesh report 2009. Dhaka: World Health Organization, Country office org/10.1016/S0140-6736(09)61034-5 PMID:19560602 for Bangladesh; 2009. 14. Bangladesh Bureau of Statistics. Report of Sample Vital Registration System, 3. Jha P, Jacob B, Gajalakshmi V, Gupta PC, Dhingra N, Kumar R et al.; 2010. Dhaka: Ministry of Planning, Government of the People’s Republic of RGI-CGHR Investigators. A nationally representative case-control study of Bangladesh; 2010. smoking and death in India. N Engl J Med 2008;358:1137–47. doi: http:// 15. United Nations, Department of Economic and Social Affairs, Population dx.doi.org/10.1056/NEJMsa0707719 PMID:18272886 Division. World population prospects, the 2012 revision. New York: UN; 2012. 4. Global Tobacco Surveillance System. Global Adult Tobacco Survey – India Available from: http://esa.un.org/unpd/wpp/Excel-Data/mortality.htm report 2010. New Delhi, India: World Health Organization and Ministry of [accessed 30 June 2013]. Health and Family Welfare; 2010. 16. Chowdhury AM, Chowdhury S, Islam MN, Islam A, Vaughan JP. Control 5. Pirie K, Peto R, Reeves GK, Green J, Beral V; Million Women Study of tuberculosis by community health workers in Bangladesh. Lancet Collaborators. The 21st century hazards of smoking and benefits of 1997;350:169–72. doi: http://dx.doi.org/10.1016/S0140-6736(96)11311-8 stopping: a prospective study of one million women in the UK. Lancet PMID:9250184 2013;381:133–41. doi: http://dx.doi.org/10.1016/S0140-6736(12)61720-6 17. Razzaque A, Nahar L, Abu Haider MGM, Karar ZA, Islam MS, Yunus M. PMID:23107252 Sociodemographic differentials of selected noncommunicable diseases 6. Jha P, Ramasundarahettige C, Landsman V, Rostron B, Thun M, Anderson risk factors among adults in Matlab, Bangladesh: findings from a WHO RN et al. 21st-century hazards of smoking and benefits of cessation in the STEPS survey. Asia Pac J Public Health 2011;23:183–91. doi: http://dx.doi. United States. N Engl J Med 2013;368:341–50. doi: http://dx.doi.org/10.1056/ org/10.1177/1010539510392743 PMID:21159696 NEJMsa1211128 PMID:23343063 18. Gajalakshmi V. Peto R, Kanaka S, Jha P. Smoking and mortality from 7. International Agency for Research on Cancer. Vol. 83. Tobacco smoke tuberculosis and other diseases in India: retrospective study of 43000 adult and involuntary smoking. In: IARC monographs on the evaluation of the males and 35000 controls. Lancet 2003;362:507–15. doi: http://dx.doi. carcinogenic risks of chemicals to humans. Lyon: IARC; 2004. org/10.1016/S0140-6736(03)14109-8 PMID:12932381 8. Centre for health and population research. Health and demographic 19. Gupta PC. Survey of sociodemographic characteristics of tobacco use surveillance system – Matlab, volume.36: registration of health and among 99,598 individuals in Bombay, India using handheld computers. demographic events 2003. Dhaka: ICDDR, B; 2005. Tob Control 1996;5:114–20. doi: http://dx.doi.org/10.1136/tc.5.2.114 9. Alam N, Chowdhury HR, Bhuiyan MA, Streatfield PK. Causes of death of PMID:8910992 adults and elderly and healthcare-seeking before death in rural Bangladesh. 20. Jha P, Gajalakshmi V, Gupta PC, Kumar R, Mony P, Dhingra N et al.; RGI-CGHR J Health Popul Nutr 2010;28:520–8. doi: http://dx.doi.org/10.3329/jhpn. Prospective Study Collaborators. Prospective study of one million deaths v28i5.6161 PMID:20941904 in India: rationale, design, and validation results. PLoS Med 2006;3:e18. doi: 10. International statistical classification of diseases and related health problems, http://dx.doi.org/10.1371/journal.pmed.0030018 PMID:16354108 10th revision. Geneva: World Health Organization; 2008. 21. Dikshit R, Gupta PC, Ramasundarahettige C, Gajalakshmi V, Aleksandrowicz 11. Peto R, Lopez AD, Boreham J, Thun M. Mortality from smoking in developed L, Badwe R et al.; Million Death Study Collaborators. Cancer mortality in countries, 1950–2000. 2nd ed. Available from: www.ctsu.ox.ac.uk/~tobacco/ India: a nationally representative survey. Lancet 2012;379:1807–16. doi: [accessed 30 June 2013]. http://dx.doi.org/10.1016/S0140-6736(12)60358-4 PMID:22460346 12. Liu BQ, Peto R, Chen ZM, Boreham J, Wu Y-P, Li J-Y et al. Emerging tobacco 22. Barkat A, Chowdhury AU, Nargis N, Rahman M, Khan MS, Kumar A et al. The hazards in China: 1. Retrospective proportional mortality study of one economics of tobacco and tobacco taxation in Bangladesh. Paris: International million deaths. BMJ 1998;317:1411–22. doi: http://dx.doi.org/10.1136/ Union Against Tuberculosis and Lung Disease; 2012. bmj.317.7170.1411 PMID:9822393 23. Jha P, Chaloupka FJ. Curbing the epidemic: governments and the economics of tobacco control. Washington: The World Bank; 1999. 764 Bull World Health Organ 2013;91:757–764 | doi: http://dx.doi.org/10.2471/BLT.13.120196