Article Text
Abstract
Background: In Japan, tobacco smoking is one of the main avoidable causes of disease and death. Although the benefits of smoking cessation for reducing disease risk and increasing longevity have been extensively documented, a relatively low proportion of Japanese smokers currently express a willingness to quit. This study attempted to quantify future reduction in the burden of smoking-attributable disease that could result from increases in smoking cessation.
Methods: A simulation model was developed to project changes in mortality in Japan associated with increased quit attempts and use of nicotine replacement therapy (NRT) among smokers, incorporating data on smoking prevalence, cause-specific mortality rates, quitting behaviour and NRT use and effectiveness.
Results: Approximately 46 000 lung cancer deaths and 56 000 cardiovascular disease deaths could be avoided over 20 years if the proportion of smokers making a quit attempt per year gradually increased to current US levels over 20 years. If each of these quit attempts were aided by NRT, the estimates of avoidable deaths would increase to 64 000 for lung cancer and 78 000 for cardiovascular disease. In this model, negligible deaths were avoided due to decreased smoking initiation over the 20-year simulation.
Conclusion: Smoking cessation can have measurable short-term impacts on the smoking-related mortality burden in Japan. However, to achieve these gains, tobacco control policies should focus both on increasing smokers’ willingness to quit and providing the support and therapies to increase the likelihood that smoking cessation attempts will succeed.
Statistics from Altmetric.com
Worldwide, tobacco use is a leading cause of preventable disease and premature death. For over a century, large corporations have manufactured and sold cigarettes, a highly addictive tobacco product. Today, over one billion people smoke.1 In recent decades, as smoking has declined in developed countries, the multinational tobacco companies have aggressively sought new markets in the developing countries. They also look to retain their markets in developed countries by seeking new customers among young adults, both men and women.
In Japan, tobacco smoking is one of the main avoidable causes of disease and death. About half of young men smoke and because the numbers of cigarettes Japanese men smoked increased until the early 1970s, the burden of smoking-related disease can be expected to rise.2 Fortunately, most Japanese women do not smoke; however, the number of smokers among young women, who are a target of tobacco advertising and promotion, is rising. If the disease epidemic caused by smoking is to be slowed quickly, a substantial proportion of male smokers need to quit permanently. For the longer term, prevention of smoking initiation is needed to reduce the widespread smoking among men and to assure that tobacco use does not increase among women.
The benefits of smoking cessation for reducing disease risk and increasing longevity have been extensively documented.3 4 However, a relatively low proportion of Japanese smokers express a willingness to quit at present. In 2003, 24.6% of men and 32.7% of women reported that they would like to quit smoking,5 but for the period 2002–5, on average, only 8% of Japanese smokers had made a quit attempt in the last year (Fujii A, GSK, personal communication). These percentages are low in comparison to the estimate of almost 42.5% for US smokers in 2005 (fig 1).6 The average smoker will make many quit attempts before quitting smoking successfully, although the effectiveness of a given quit attempt can be increased using proved approaches, such as pharmacotherapy. If the number of quit attempts and the long-term success could be increased in Japan, we can confidently predict a significant reduction in the burden of smoking-attributable disease.
In this report, we provide the results of modelling the burden of avoidable deaths among Japanese smokers under various scenarios of future cessation. We focus on lung cancer and cardiovascular disease (CVD), leading causes of deaths among smokers and major causes of death in Japan. The success of smoking cessation can be enhanced with pharmacotherapy, including nicotine replacement therapy (NRT). One aspect of our modelling considers scenarios of NRT use and the potential impact of this therapy on reducing future disease burden. The model incorporates Japanese estimates of smoking prevalence, cessation rates and smoking-related mortality risks.7 The analysis illustrates the magnitude of avoidable premature death in a population of addicted smokers who presently have a low rate of successful quitting.
METHODS
We have developed a simulation model to project changes in mortality in Japan associated with increased quit attempts and NRT use among smokers. This model incorporates data on smoking prevalence, cause-specific mortality rates, quitting behaviour and nicotine therapy use and effectiveness.
We examined a set of future scenarios associated with different patterns of youth smoking prevalence, quit attempts among current smokers and NRT use and effectiveness. The difference in total mortality between the “baseline” and each “alternative” scenario is the net reduction in premature mortality associated with increased quit attempts. The specific data sources and methods used to develop the model are shown in table 1 and are described below.
Data sources
Smoking prevalence and Initiation
We used the 2003 National Health and Nutrition Survey in Japan (NHNS-J) to estimate the prevalence of current smoking among Japanese adults (20+ years old) by gender and 10-year age groups. The NHNS-J is a nationwide survey conducted since 1948 by the Japanese Ministry of Health, Labour, and Welfare.5 Current smokers were defined as those who answered “Total of 100 or more cigarettes or for 6 months or more” to the question “Have you ever smoked?” and also answered that they smoked every day or sometimes within the past month.
We incorporated several assumptions regarding the introduction of new smokers in the model. First, we assume that smoking initiation among adults occurs only before 25 years of age. The average age of initiation in Japan was 21 years of age for males and 28 years of age for females during the period of 1990–3 and has been declining, especially in women.8 9 Second, we have modelled smoking initiation at a constant rate into the future, such that the smoking prevalence in the 18–24-year-old age group remains constant, unless otherwise specified in the model. Third, the population of 18–24-year-olds declines at a rate of about 62 000 people per year for each gender, based on predictions for the period 1990–2030 from the National Institute of Population and Social Security.12
Cessation attempt rates
We obtained rates of quit attempts from a series of NRT tracking surveys conducted by Glaxo-SmithKline (GSK) from 2002–5 (Fujii A, GSK, personal communication). The NRT tracking survey, a mail survey sent to approximately 2000 people age 20–59 years old, has been conducted at least twice a year since 2002. Current smokers were defined as those who answered “Almost every day” to the question “Do you smoke in your daily life?” Smoking cessation attempts in the past year among current smokers were defined by an affirmative response to “Have you ever tried to stop smoking?” along with reporting that their last quit attempt was in the previous year. We calculated the proportion of current smokers and former smokers who made at least one quit attempt in the past year by gender and 10-year age groups.
Nicotine replacement therapy use and effectiveness
We obtained data on current NRT use in Japan, by age group and gender, from the NRT tracking survey conducted by GSK. NRT use among almost daily smokers who have attempted to quit in the past year was based on the responses of “NRT gum” or “NRT patch” to the question “Which method did you try last time?” Rates of NRT use among smokers and former smokers who made a quit attempt during the last year were estimated by age group and gender.
NRT use has been shown to increase the likelihood of long-term success in quitting. The probability of long-term abstinence for a given quit attempt with and without the use of NRT was obtained from the scientific literature. A recent review suggests that the long-term prolonged abstinence rate for self-quitters with no assistance ranges from 3% to 5%.11 A systematic review of NRT effectiveness was used to characterise the increased chance of long-term success in quitting when using NRT versus no therapy during a quit attempt. These results suggest an approximate doubling in the long-term success rate with the use of NRT.12 In the simulation model, we assume a 5% long-term success rate for an unaided quit attempt and a 10% long-term success rate for an NRT-aided quit attempt.
Smoking-related mortality
We conducted a literature search to identify cohort studies with mortality data suitable for use in the simulation. For our previous work on the TobaccoFree*Japan report,2 we created a literature database that includes abstracts, qualitative and quantitative data from key articles regarding active smoking, exposure to secondhand smoke and disease outcomes that are relevant for Japan and that were referenced in the report. The articles in the report and database reflect the most recent findings in the scientific literature up to March 2003. The report was written to provide a perspective on the state of the tobacco epidemic in Japan, including the burden of disease caused by smoking in Japan.
We conducted an update of the literature for lung cancer and CVD in May 2005 using PubMed. For lung cancer, we used the search terms smoking OR smoker OR smoke OR tobacco AND Japan* AND cancer OR tumor OR neoplasm AND human AND cohort OR case OR epidemiol* NOT case report AND lung. A total of 81 articles resulted from the PubMed search and six articles were reviewed and abstracted for the database. For CVD, we used the search terms smoking OR smoker OR smoke OR tobacco AND Japan* AND cardiovascular AND human AND cohort OR case OR epidemiol* NOT case report. A total of 117 articles resulted from the PubMed search and six articles were reviewed and abstracted in the database. Currently, the Tobacco Free*Japan database contains 446 articles on smoking and disease of relevance for Japan.
We used mortality rates among non-smokers and relative risks (RRs) associated with smoking for lung cancer and CVD from the Japan Collaborative Cohort (JACC) study.13 14 The JACC study was selected based on the review of cohort studies because it was one of the most recent studies, had the largest number of cases, and had the most detailed information about smoking, age and gender. The JACC study is a prospective cohort study that was established from 1988 to 1990, with 46 465 males and 64 327 females aged 40–79 years. The results obtained were for follow-up until the end of 1997 for lung cancer deaths and 1999 for CVD deaths. In the JACC study, never-smokers were defined as those who reported never having smoked cigarettes, former smokers were those who stopped cigarette smoking before participating in the JACC study and current smokers were those who had a continuing smoking habit. Mortality rates for lung cancer among never-smokers were reported by 10-year age group and gender. For CVD, the mortality rates among never-smokers were available for two broad age groups, 40–64 and 65–79 years. To generate estimates by smaller age groups, we interpolated cardiovascular mortality by age among never-smokers by fitting an exponential function to data from the American Cancer Society’s Cancer Prevention II (CPS-II) cohort study (Michael Thun, American Cancer Society, personal communication) and anchoring this to the observed JACC estimates. CPS-II is a nationwide, prospective cohort study begun in 1982, comprising 1.2 million US adults aged 30 and older.15 Results were available for follow-up from 1984–91.
We obtained unadjusted RRs from the JACC study for current smokers and former smokers by 10-year age groups. Limited data were available on lung cancer mortality relative risks for female smokers. As a result, we conducted the simulation using two different assumptions: (1) female smokers experience the same RR as males, and (2) female smokers’ RRs are equivalent to half the excess risk of males. Because the majority of smokers are male, these two approaches yielded estimates of avoided deaths that differed by only about 6%. Thus, for the main analysis, we present results assuming an equivalent RR in males and females. The simulation model estimates avoided deaths from ages 40 to 84. Relative risk estimates were available for age 40 through age 79 years, and we assumed that people aged 80–84 have the same risk as those in the age group 70–79.
We estimated risk after smoking cessation for the JACC study using data for lung cancer and cardiovascular mortality among males from CPS-II. For four different age-at-cessation groups (<40, 40–49, 50–59, 60–69), we modelled the RR among former smokers (compared with never smokers) as a function of time since quitting using a log-linear model with inverse variance weighting. From this model, we estimated the percentage reduction in RR per year of age since quitting. We applied the coefficients obtained from the model to current smoker RRs from the JACC study to estimate the trend in risk after cessation among Japanese smokers.
Simulation model
We developed a simulation model using the data inputs described above. The simulation model was run out to 20 years into the future. The simulation begins with the population of current smokers in Japan, estimated for 2003, as its baseline. As the population of smokers advances in age, some smokers will die, attempt to quit or remain as current smokers. Among those who attempt to quit, some will use NRT and some will not. Those successfully quitting will move to being former smokers while the others will continue as smokers. New smokers also join the current smokers and mortality is estimated for current and former smokers (fig 2). The extent to which current smokers become former smokers depends on assumed baseline quit attempt rates, the prevalence of NRT use, the rate of long-term quitting success in the absence of NRT and the effectiveness of NRT in maintaining long-term quitting success. As the simulated population was advanced through time, age-specific and gender-specific quit attempt rates and prevalence of NRT use were applied in the model. The mortality risk faced by the simulated population is determined by age, gender, smoking status and time since quitting among former smokers.
Each simulation included a “baseline” scenario in which quitting behaviour and NRT use remained constant into the future at baseline levels. Alternative scenarios were modelled in which quitting behaviour and NRT use gradually increase over time or smoking initiation decreases over time (fig 2). We report the difference in mortality between the two scenarios as the number of avoided deaths associated with increased rates of quit attempts and NRT use. The specific alternative scenarios for increasing the rate of quit attempts were: (1) doubling over a 20-year period; (2) increasing to current US levels (48%) over a 20-year period; and (3) increasing to 100% of smokers making a quit attempt per year over a 20-year period. For NRT use, the alternative scenarios modelled were: (1) doubling over a 20-year period; (2) increasing to 50% of quit attempts using NRT; (3) increasing to 100% of quit attempts using NRT. In addition to these individual scenarios, we modelled the impact of concurrent increases in quit attempts and NRT use.
Statistical analyses
Data obtained on smoking prevalence from the 2003 NHNS-J5 and on quit attempt rates and NRT use (Fujii A, GSK, personal communication) from the GSK 2002-5 NRT tracking studies as a Microsoft Excel spreadsheets, were imported into SAS version 9.1 for analysis.16 SAS also was used to develop the simulation model. Non-smokers’ mortality rates and relative risks by smoking status were obtained from published results of the Japan Collaborative Cohort.13
RESULTS
Figures 3–6 show the cumulative number of avoided premature deaths over time predicted with increases in rates of quit attempts and NRT use. Tables 2 and 3 show estimates of the cumulative number of avoided premature deaths from lung cancer and CVD over the 20-year period of the simulation. The estimates of avoided premature deaths from lung cancer were slightly lower than for CVD, predominantly due to the higher prevalence of CVD in the Japanese population.17 If quit attempt rates remained constant over the 20-year simulation period, the number of avoided deaths from lung cancer would increase modestly, even if every smoker who attempted to quit used NRT. The estimated number of avoided premature deaths from lung cancer was nearly 8000, assuming that the quit attempt rates gradually doubled in Japan and NRT use stayed constant over 20 years. If the rate of quit attempts among smokers gradually increased to current US levels, the estimated number of lung cancer deaths avoided over 20 years would increase to approximately 46 000 deaths. This number would jump to over 100 000 if every smoker made at least one quit attempt per year. If each of these quit attempts were aided by NRT, the number of avoided lung cancer deaths would increase by about 30 000 to over 130 000.
For cardiovascular disease, an increase in quit attempt rate to US levels and to 100% by the end of a 20-year period would lead to over 50 000 and 130 000 avoided premature deaths, respectively. In the former scenario, if the rate of NRT-aided quit attempts increased to 100%, a total of approximately 78 000 premature cardiovascular deaths would be avoided. In the latter scenario, an increase in NRT-aided quit attempts to 100% would lead to a total of about 170 000 premature cardiovascular deaths avoided. These estimates assume that no other changes in smoking behaviour occur over the simulation time period.
In our model, negligible deaths were avoided due to decreased smoking initiation over the 20-year simulation (data not shown). Given the limited data on smoking relative risk before age 40, we only estimated avoidable deaths between the ages of 40 and 84. Because initiation was assumed to occur only before age 25, there were very few years in the simulation in which avoidable deaths among new smokers would have been estimated. However, owing to the low baseline risk of death among non-smokers before age 40, there are likely to be few smoking-attributable deaths before age 40.
DISCUSSION
This analysis describes the reduction in premature mortality in Japan over the next 20 years that could be achieved through increased quit attempts and use of NRT. The largest estimates of avoided deaths would be associated with an increase in the quit attempt rate and moderate additional gains could be achieved from increased NRT use. Conversely, we have shown that reduced smoking initiation would not have a measurable impact on mortality over a 20-year period. We estimate that over 110 000 deaths due to lung cancer and cardiovascular disease combined could be avoided by achieving an increase in the quit attempt rate to current US levels and a doubling of NRT use among smokers making a quit attempt. These findings suggest that smoking cessation can have measurable short-term impacts on the smoking-related mortality burden. However, to achieve these gains, tobacco control policies should focus both on increasing smokers’ willingness to quit and providing the support and therapies to increase the likelihood that smoking cessation attempts will succeed.
Any attempt to predict future mortality patterns will, of necessity, require many assumptions to bridge evidence gaps and related uncertainties. In this analysis, we examined the number of premature deaths from lung cancer and cardiovascular disease that could be avoided with increased smoking cessation. These two end-points were selected because they are major contributors to smoking-related premature death and disease and because recent data were available from a Japanese cohort study to describe the increased risk of death from these conditions as a result of smoking.13 14 Insufficient data were available to describe risk from all-cause mortality and other smoking-related diseases, such as chronic obstructive pulmonary disease (COPD). However, the relative risks associated with smoking for lung cancer and CVD were available only for broad age groups, relative risks for lung cancer were not available for females, and limited Japanese data were available to describe the decline in risk after cessation. For this reason, we applied data from the ACS CPS-II to fill in these gaps.
We also assume that the mortality rates among never-smokers observed in the JACC study remain constant into the future. However, numerous factors may result in changing rates over time, including improvements in detection and treatment and trends in the prevalence of risk factors over time. For lung cancer, which has few prevalent non-smoking-related risk factors, no effective screening modalities available18 and very poor and only slowly improving survival,19 this assumption is reasonable. However, changes in risk-factor prevalence may be especially important for multifactor diseases such as CVD. Similarly, we assumed that the relative risks of smoking from the JACC study will reflect the disease risk of current smokers into the future, regardless of smoking frequency and duration. Factors such as changes in the average age at initiation and the average number of cigarettes consumed per day could influence the magnitude of the effect of smoking on mortality risk. In the model, changes in risk occur immediately when individuals successfully quit. This probably overestimates the magnitude of avoidable cancer mortality since risk reduction does not occur immediately after cessation.
Additional uncertainty stems from the use of estimates of quitting behaviour and NRT use based on small and not necessarily representative samples of the Japanese population. NRT effectiveness was based on a systematic review of many studies, most of which were not conducted in Japan.12 In this analysis, we assumed that the effectiveness of NRT would be the same in Japan. These limitations and data gaps increase the uncertainty in our estimates and point to future research opportunities to better describe the burden of tobacco-related disease in Japan.
Japan is at a critical and pivotal moment in the course of its tobacco epidemic that began 50 years earlier. The majority of its men are addicted smokers who will inevitably face premature onset of chronic, crippling and fatal diseases. Fortunately, smoking is uncommon among Japanese women, with only about 11% currently smoking (fig 1).15 Unfortunately, smoking is increasing among young women, who are a target for aggressive marketing by the Japanese tobacco industry—majority-controlled by the government—and by other multinational corporations.
Given the large numbers of male smokers at high risk for tobacco-related disease and death, steps should be taken to promote cessation, which would have immediate and long-term health benefits. Efforts should be directed at changing the social environment so that the smoker’s desire to quit increases; with the help of NRT availability, the smoker’s chances of successfully quitting also increases. Currently, the proportion of smokers making a quit attempt in the past year in Japan is quite low in comparison to the United States, which has a long history in the development, implementation and enforcement of tobacco control policies. Tobacco control policies in the United States have been effective in reducing smoking prevalence. As a result, population-wide lung cancer mortality rates began to decline among males in the early 1990s and recent evidence suggests a decline the rates among females as well.20 Japan, on the other hand, is in the earliest stage of tobacco control, with male smoking prevalence at the level of US males in the 1970s and with quitting attempts lagging behind US rates by over 30 years. This comparison points to the need for increasing awareness of the harms of smoking and the benefits of quitting for quality of life and longevity. It also highlights the long time frame necessary to curb the epidemic of smoking-related cancer deaths in a population with a significant smoking history.
Although the need for tobacco control in Japan has long been evident, activities have been limited and local, rather than organised and national. Tobacco control may be uniquely difficult in Japan because the government controls the Japanese tobacco industry. In fact, the Tobacco Industry Law of 1984 calls for activities that promote the development of tobacco in Japan, and tobacco sales provides the government with substantial tax revenues. The government only recently established a focal point office in the Ministry of Health, Labor, and Welfare for tobacco control.
Japan signed the Framework Convention for Tobacco Control (FCTC) on 9 March 2004, and ratified it on 8 June 2004. The FCTC, a needed platform for addressing a global epidemic, commits Japan to a series of measures related to tobacco control. Since recognising the significant health and economic burden imposed by tobacco on its citizens and acknowledging the policy requirements of the FCTC, the Japanese government should now move to implement an effective national tobacco control policy, using the most appropriate tobacco control measures.
What this paper adds
This study shows the future reduction in the burden of smoking-attributable mortality that could result from different scenarios of increased smoking cessation in Japan. Although greater access and use of nicotine replacement therapy should increase the rate of successful smoking cessation in Japan, the greatest impact on mortality burden would be achieved by increasing the rate at which smokers make a quit attempt.
We estimate that if the proportion of smokers making a quit attempt increased to current US levels over the next 20 years in Japan, approximately 102 000 premature deaths from lung cancer and cardiovascular disease would be avoided.
In summary, this report provides an estimate of the premature mortality that may be avoided through increased cessation. Despite the limitations described above, these estimates provide a basis for understanding the burden of smoking-related premature death that could be reduced by effective tobacco control policies aimed at smoking cessation. To achieve the greatest health benefits to the population over the long term, however, a dual strategy based on cessation for addicted smokers and prevention for youths will be necessary.
Acknowledgments
We thank the American Cancer Society for providing age-specific, sex-specific and disease-specific death rates from Cancer Prevention Study II. The authors would also like to thank Mr Akira Fujii for his support and help throughout the study, Ms Georgiana Onicescu for her technical assistance, Ms Carrie Mattson for her administrative support and Ms Charlotte Gerczak for her editing assistance.
Footnotes
Competing interests: None.
Funding: This research was supported by a grant from GlaxoSmithKline. The funding organisation had no role in the outcome of this study or the preparation of the manuscript.