Article Text
PDF
Abstract
Objective: To provide the most reliable evidence as to the nature of the associations between smoking and cause-specific illness, as well as the expected benefits from quitting smoking, in studies conducted in Asia, where smoking remains popular among men.
Data sources: Studies published between January 1966 and October 2008, identified in the Medline search strategy with medical subject headings, in addition to studies from the Asia Pacific Cohort Studies Collaboration.
Study selection: Studies were considered to be relevant if they were prospective studies, in an Asian setting that reported on the association between smoking, quitting and cause-specific illness.
Data extraction: Two reviewers independently screened all identified articles for possible inclusion and extracted data.
Data synthesis: The pooled relative risks (RRs) for incidence or mortality, comparing current to never smokers were always significantly higher than unity; the highest was for lung cancer: 3.54 (95% confidence interval 3.00 to 4.17). The pooled RRs for former smokers (compared to never smokers) were also always significantly higher than unity, and were lower than in current smokers, for coronary heart disease, stroke, lung and upper aero-digestive tract cancer. Only for respiratory disease was the RR for former smokers higher than that for current smokers.
Conclusions: This meta-analysis has shown that, despite the relative immaturity of the smoking epidemic in Asia, smoking is unquestionably a major contributor to ill health and death. However, the beneficial effects of quitting are not yet always apparent, most probably because quitting is a consequence of ill health and the relative unpopularity of smoking cessation in many Asian populations.
Statistics from Altmetric.com
The seminal work of Doll and Hill1 2 in the 1950s convincingly demonstrated the causal role of smoking in lung cancer and prompted a sharp decline in the popularity of the habit, particularly in countries such as the United Kingdom and the United States.3 4 5 Fifty years on, thanks largely to highly effective mass-media campaigns,4 5 there is now widespread acceptance among most Western populations of the causal role of smoking in a plethora of diseases, such as many cancers, cardiovascular and respiratory diseases.6 7 Combined with broad legislative measures and tobacco tax policies,4 5 such campaigns have been instrumental in convincing hundreds of thousands of smokers to quit and dissuading thousands of others from taking up the habit.
Despite these enormous efforts to curb the smoking pandemic, it remains the second leading cause of death (after high blood pressure), accounting for 12% of all deaths worldwide, as well as one of the major causes of disability.8 9 Paradoxically, most of these deaths occur in Western populations that now have some of the lowest smoking rates (typically less than 30% in men), whereas the burden of smoking-related illness in Asia (where between 50–60% of men smoke depending on the country3 4 5 10) remains disproportionately low.11 This partly reflects the often long latency period between smoking and onset of illness, but over the coming decades this global pattern of morbidity and mortality is likely to reverse if current smoking trends in Asia persist. Moreover, the enormous public health burden that is the result of smoking will have a substantial negative impact on a country’s economy, an effect that will be particularly great in Asian countries relative to the West over the coming decades.4 12
Although there are several countries in Asia where the prevalence of smoking among men has been on the decline in recent years, including Hong Kong, Thailand, Singapore and Japan,3 4 5 10 13 in other countries such as China and Indonesia, there is no evidence to indicate a similar decline in the popularity of the habit. There are perhaps two main reasons for this, a low level of awareness as to the harms associated with the habit7 14 15 16 and a lack of population-wide smoking cessation strategies. An essential prerequisite for any smoking cessation campaign is a sound evidence base. To this end, we conducted a systematic review of the literature for all Asian studies that reported on the association between smoking, quitting and illness in order to provide the most reliable evidence as to the nature of the associations between smoking and cause-specific illness, as well as the expected benefits of quitting smoking, in studies conducted in Asia.
Methods
Search strategy
We performed a systematic search for relevant articles published from January 1966 to October 2008, using Medline. We searched with medical subject headings (MeSH): ([Smoking (MeSH)] or [Tobacco (MeSH)] or [Tobacco use cessation (MeSH)] or [Smoking cessation (MeSH)]) and ([Cardiovascular diseases (MeSH), including MeSH terms found below this term in the MeSH tree] or [Neoplasms (MeSH), including MeSH terms found below this term in the MeSH tree] or [Respiratory tract diseases (MeSH), including MeSH terms found below this term in the MeSH tree]) and ([Asia (MeSH), including MeSH terms found below this term in the MeSH tree]). We restricted the search to English language articles and studies of human subjects. We read the titles and abstracts of all the articles identified in the Medline search to exclude any articles that seemed irrelevant. The full texts of the remaining articles were read to determine if they met our criteria for inclusion. In addition, we manually searched for extra relevant articles in the reference lists of the identified articles and other publications.
Data extraction
Articles were considered to be relevant if they were prospective studies (cohort studies or nested case-control studies), in an Asian setting, that calculated the approximate relative risks (RRs) (that is, hazard ratios, rate ratios or odds ratios) for coronary heart disease (CHD), stroke, site-specific cancer and respiratory disease incidence and/or mortality in both current smokers and former smokers with never smokers serving as the referent. Articles were excluded if they did not provide the RR with 95% confidence interval in both current smokers and former smokers with adjustment at least for age. Articles were also excluded if the study population solely consisted of individuals with a particular high-risk condition (for example, high blood pressure or diabetes) or of hospital patients. In the case of multiple publications from the same or overlapping cohorts, we selected only the article with the largest person-years of follow-up or, if the articles had exactly the same person-years of follow-up, only the article with the most exhaustive level of adjustment for potential confounders.
The Asia Pacific Cohort Studies Collaboration (APCSC) is a collaboration of prospective studies in the Asia-Pacific region17 18 that has previously published results on smoking, quitting and several diseases.19 20 For the current work we included only Asian studies in APCSC that were not otherwise included in the search of published literature, to avoid duplication.
Statistical analysis
Outcomes treated in this article are CHD, stroke, cancer and respiratory diseases. Cancer was classified into four categories: lung cancer, upper aero-digestive tract (UADT) cancer, gastric cancer and other cancers. No other cancers, or subsets within our chosen classifications, provided large enough numbers of events for reliable estimation of associations. In most relevant articles, including APCSC, outcomes were defined using the International Classification of Diseases or the International Classification of Diseases for Oncology.
The primary analysis was to estimate pooled RRs for incidence (that is, fatal or non-fatal events) and mortality (that is, where fatal events alone were reported) combined for each of the seven diseases. Should an article report RRs for both incidence and mortality, only the RR for incidence was used for the primary analysis. As a secondary analysis, summary estimates were obtained for mortality only. Further analyses repeated the above for separate countries. Again, to avoid small numbers, and thus unreliable results, only China, Japan and Korea were separated, although we included a category of other Asian countries for completeness.
For current smokers and former smokers, pooled RRs and their corresponding 95% confidence intervals for the outcomes of interest were estimated using a random effects model with inverse variance weighting.21 Statistical heterogeneity within studies was assessed using the Cochran’s Q test for heterogeneity and Higgins’s I2 statistic.21 I2 is interpreted as the percentage of variability between studies due to heterogeneity, rather than chance, with 0% representing no heterogeneity. These meta-analyses were carried out using Stata, version 9.
Results
Our search strategy identified a total of 2705 articles (fig 1), of which 60 were considered as relevant studies that met our criteria for inclusion. Seven additional relevant articles were identified from reference lists. Of these 67 relevant articles, 14 articles were excluded because of multiple publications. In addition, APCSC provided 24 additional cohort studies. In all, this gave 77 individual studies for inclusion in the meta-analysis: 73 cohort studies and four nested case-control studies. The characteristics of these studies are summarised in table 1, according to country and initial year of baseline survey.22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 Of the 77 studies, 42 studies were from Japan; 16 from China, five from Singapore, five from Taiwan, four from Korea, three from Hong Kong, one from India and one from Thailand. In most studies, the prevalence of current smoking among men was over 50%, whereas for women it was typically below 10%. In the 1990s, China had a higher rate of current smokers (defined as the number of current smokers among total participants) and a lower rate of quitters (defined as the number of former smokers among ever smokers) among men, compared to Japan and Korea.
Flow chart of search strategy and data extraction.
Characteristics of the 53 published studies and the Asia Pacific Cohort Studies Collaboration involving 24 studies
The pooled RRs for incidence or mortality, comparing current to never smokers were always significantly higher than unity, ranging from a value of 1.23 for other cancers to 3.54 for lung cancer (fig 2). The pooled RRs for former smokers (compared to never smokers) were also always significantly higher than unity, but were lower than in current smokers for CHD, stroke, lung cancer and UADT cancer, although not always significantly so. Former smokers had much the same risk as never smokers for gastric and other cancers and for respiratory diseases. The pooled RRs for mortality presented a similar pattern to the pooled RRs for incidence and mortality combined (data not shown). There was a high degree of heterogeneity in the estimates between studies for all except other cancers and respiratory diseases. With these two exceptions, I2 statistics ranged from 32% to 89% and all Q tests were significant at the 10% level. This justifies the use of random effects methodology.
Pooled relative risks (RRs) and 95% confidence intervals (CIs) for incidence and mortality combined in former smokers (broken lines) and current smokers (solid lines) with never smokers serving as the referent. Upper aero-digestive tract (UADT) cancer included oesophageal, oral, nasopharyngeal and laryngeal cancer. Other cancers included colon, rectum, liver, gallbladder, pancreas, skin, breast, uterus, ovary, prostate, bladder, kidney, ureter, brain, thyroid cancer, lymphoma, multiple myeloma and leukaemia. Asterisks indicate that one or two studies did not report the number of events.
In a sensitivity analysis in which the four nested case-control studies were excluded, including one stroke study, two lung cancer studies and one liver cancer study, the pooled RR (95% confidence interval) for current smokers and former smokers (compared to never smokers) was 1.36 (1.23 to 1.50) and 1.10 (1.01 to 1.21) for stroke, 3.48 (2.95 to 4.10) and 2.11 (1.86 to 2.04) for lung cancer and 1.23 (1.13 to 1.34) and 1.22 (1.16 to 1.24) for other cancers, respectively. These results were broadly similar to those shown in figure 2, indicating that these studies did not bias the overall summary estimates.
In country-specific analyses, the pooled RRs (compared to never smokers) from studies conducted in Japan and Korea tended to be lower in former smokers than in current smokers for CHD, stroke, lung cancer, UADT cancer and respiratory diseases (table 2). However, former smokers had higher pooled RRs compared with current smokers for respiratory diseases in China and for stroke and respiratory diseases in other Asian countries, although the pooled RRs for CHD, stroke (in China), lung cancer and UADT cancer were lower in former, than current, smokers (table 2).
Pooled relative risks for incidence and mortality combined in former smokers and current smokers with never smokers serving as the referent group by country
Discussion
In the present study, we have provided the most reliable estimates to date as to the health hazards of smoking and the benefits of quitting for various diseases in Asian populations. Our summary estimates indicate that smoking is certainly hazardous for cardiovascular diseases (especially CHD), cancer (especially lung cancer and UADT cancer) and respiratory diseases in Asia. Importantly, in most instances, we were able to show that quitting smoking reverses the harmful effects of smoking on most diseases that are closely associated with smoking, except that we were unable to show these benefits completely for China. One explanation of why the beneficial effects of quitting smoking were not wholly apparent in Chinese studies may have been because of the underlying presence of disease in smokers who had reportedly quit smoking. In support of this, there is some evidence to suggest that the reasons for quitting smoking differ considerably between developed and developing countries. In the former, smokers are more likely to quit for health reasons,75 whereas in developing countries, where the hazards of smoking are much less widely known, the primary reason for quitting is ill health.15 Consequently, the beneficial effects of smoking cessation are likely to be underestimated among such individuals.20 76 77
Our overall summary estimates are specific to Asian populations (who have many different physiological and non-physiological characteristics from Western populations), but even among these studies, there was evidence of significant heterogeneity of the impact of smoking and smoking cessation on health outcomes. This may be because of the differences in smoking behaviour and/or genetic, environmental or lifestyle-related factors between countries; although the differences in factors related to research methodology (for example, range of age, method for identification of the events, levels of adjustment) between studies may also have played a part. For instance, in a study where smokers have a high average consumption of cigarettes per smoker the health hazard of smoking will be proportionally large. On the other hand, extensive passive exposure to environmental tobacco smoke either at home or in the workplace may have resulted in an underestimation of the true harms of smoking, owing to the classification of such individuals as “never smokers” in most studies (thereby ignoring their exposure to environmental tobacco smoke).78 Other environmental air pollutants (for example, indoor coal burning), which is also a major health concern associated as it is with mortality and disability in developing regions,8 79 may also attenuate the risk for lung cancer, UADT cancer and respiratory diseases attributed to smoking.79 80 81 82
The present study suggests that the health hazards of smoking for CHD, stroke, lung cancer and UADT cancer have been underestimated in China, where tobacco control is particularly crucial, given that China has approximately 20% of the world’s population and the Chinese consume about one-third of cigarettes smoked in the world.4 5 This underestimation could, in part, be due to the potentially extreme passive exposure to cigarette and/or other environmental air pollution in China.4 79 80 81 82 83 It may well have led to a lack of awareness of the harmful effects of smoking among the Chinese; about three-quarters of smokers and two-thirds of non-smokers in China regard smoking as confering negligible harm, and less than 10% of Chinese people know that smoking causes CHD.14 To some extent, this is also true in other Asian countries.7 15 16 This unfavourable phenomenon among Asians contrasts with the almost universal recognition of the causal associations of smoking and (particularly) CHD, stroke and lung cancer among people in the West.6 7
This study has some limitations. First, a number of the included studies explored the RRs for disease-specific mortality without excluding participants who had a history of the disease of interest at baseline, so that true incidence was not always found. Second, smoking status was only recorded at study baseline, but is certain to have changed over follow-up for many participants. Third, smoking status is crudely defined—for example, not including number of years of smoking or average amounts smoked, but the broad categorisation has increased the number of studies available for inclusion and otherwise controlled variability. Forth, the categories of UADT cancer and other cancers were heterogeneously composed, and the RRs for smoking in different anatomical sites may have varied.63 Finally, this meta-analysis was restricted to English-language publications in order to minimise bias that may have been introduced by our inability to access local-language journals and to maximise study quality.84 However, we acknowledge that preclusion of non-English studies has the potential to have introduced publication bias into the analyses.
In contrast to the West which has witnessed a steady decline in the prevalence of smoking over the last few decades owing in part to widespread awareness of the harms that smoking does, many tens of millions of smokers across Asia remain oblivious to the hazards of cigarette smoking. This review provides the most reliable evidence to date of the impact of smoking on a broad array of illnesses specifically in Asian populations and offers convincing evidence as to the likely benefits from quitting the habit. Such information is essential for effective public health campaigns that aim to convince smokers across Asia to quit and dissuade others from taking up the habit, in order to prevent the many millions of smoking-related deaths that are predicted to occur in the next couple of decades, if current smoking patterns in Asia persist.
What is already known on this subject
Smoking is hazardous for cardiovascular diseases, cancer and respiratory diseases, but quitting smoking reverses the harmful effects of smoking on diseases that are closely associated with smoking.
What this study adds
This study provides the most reliable estimates as to the association between smoking, quitting and cause-specific illness in Asian populations, indicating that smoking is unquestionably a major contributor to ill health and death; however, the beneficial effects of quitting are not yet always apparent, most probably because quitting is a consequence of ill health and the relative unpopularity of smoking cessation in many Asian populations.
REFERENCES
Footnotes
Funding None.
Competing interests None.
Contributors KN, RH and MW conceived and designed the research. KN and AAM conducted the systematic review. KN, RH, AAM and MW interpreted the data. AAM performed statistical analysis. MW handled supervision. KN drafted the manuscript. RH, AAM and MW made critical revision of the manuscript. KN is guarantor.
Provenance and peer review Not commissioned; externally peer reviewed.