The Association of Smoking and Cardiovascular Disease in a Population With Low Cholesterol Levels
A Study of 648 346 Men From the Korean National Health System Prospective Cohort Study
Background and Purpose— The association of smoking with cardiovascular diseases in populations with low cholesterol levels is unclear and this lack of clarity may have contributed to the very high prevalence of smoking in East Asian men. The effect of smoking on stroke subtypes is particularly unclear.
Methods— We used data from a prospective cohort study of 648 346 Korean men aged 30 to 64 years at their baseline assessment in 1992.
Results— Mean (SD) total cholesterol at baseline was 4.9 (1.0) mmol/L and 59% of the men were current smokers. Over the follow-up period of 10 years, 9475 men experienced any type of stroke, 4768 experienced an ischemic stroke, 2380 an intracerebral hemorrhage, 786 a subarachnoid hemorrhage, 3329 men experienced a myocardial infarction, and 269 an aortic aneurysm. Smoking at baseline was associated with marked increases in risk of ischemic stroke (hazard ratio comparing current smokers with never smokers 1.58 [95% CI: 1.49 to 1.68]), subarachnoid hemorrhage (1.91 [1.63 to 2.24]), myocardial infarction (2.01 [1.87 to 2.17]), and aortic aneurysm (1.47 [1.14 to 1.90]) with these effects remaining after adjustment for potential confounding factors and with the first 5 years of events excluded from the analyses. The effect of smoking on these outcomes was the same among men with cholesterol levels below the median value for this population as among those with higher levels. There was no association of smoking with intracerebral hemorrhage. Smoking accounted for 20% of all cardiovascular disease outcomes in this population.
Conclusions— Smoking is a major risk factor for cardiovascular diseases in populations with low cholesterol levels and global interventions are required to halt the emerging tobacco epidemic in low- and middle-income countries.
Studies from developed countries have demonstrated strong associations between smoking and increased risk of coronary heart disease (CHD).1–3 However, some studies suggest that cigarette smoking has a weak or no effect on CHD in populations such as East Asian populations in which cholesterol levels are low when compared with Western populations.4–7 It has been suggested that the reporting of these findings has contributed to the very high, and increasing levels of tobacco consumption in these populations.8–10 Although studies in Western populations show a consistent increase in risk of CHD with smoking, there are inconsistencies between studies in the association of smoking with stroke, which may in part be related to differences in the proportion of all strokes that are due to intracerebral hemorrhage between different study populations.11–13 The first systematic review and meta-analysis to consider differential effects of smoking by stroke subtype was published 18 years ago.13 It reported that smoking increased the risk of ischemic stroke and subarachnoid hemorrhage but protected against intracerebral hemorrhage.13 However, a subsequent review suggested that when studies were restricted to those with CT scan-confirmed outcomes, the balance of evidence suggested an overall null effect of smoking on intracerebral hemorrhage.11 Intracerebral hemorrhage is more common and contributes a greater proportion of stroke subtypes in East Asian populations, but few studies have examined the association of smoking with stroke subtypes in these populations.
Among the limited number of studies that have explored the association of smoking with cardiovascular outcomes in East Asian populations, the Asia Pacific Cohort Studies Collaboration found that smoking was positively associated with CHD, ischemic and hemorrhagic stroke, with the association of smoking with CHD and ischemic stroke being stronger than that for hemorrhagic stroke, but region-specific analyses (ie, separating Australian and New Zealand participants from those from East Asia) were not presented for stroke subtypes.8 A prospective Japanese study found that smoking was positively associated with CHD and all stroke subtypes, but absolute effects were not presented.14 Finally, in a prospective study of Korean men (a 25% random sample of the men included in the current study), smoking amount and duration were both positively associated with CHD and stroke, but analyses by stroke subtype were limited to a sentence stating that relative effects were similar for ischemic and hemorrhagic strokes.9 That study was unable to remove individuals with prevalent cardiovascular disease at baseline, which could have resulted in underestimations of the effects if individuals who have experienced a cardiovascular disease event quit smoking.9
In this article, we examine the relative and absolute associations of smoking with all major cardiovascular disease outcomes (ischemic, intracerebral and subarachnoid stroke, myocardial infarction [MI], and aortic aneurysm) in a very large prospective cohort study of Korean men, a population with low mean cholesterol levels and high smoking prevalence.
Study participants were Korean male public servants aged 30 to 64 years who underwent a health examination provided by the Korean Medical Insurance Corporation between April 1, 1992, and September 30, 1992. Full details of the study have been previously published.15 The Korean Medical Insurance Corporation was responsible for all public service personnel, teachers and other staff in private schools; it was merged into National Health Insurance Corporation in 1998. Females were not included because the proportion of public service workers who are female in Korea is low and very few of the women have ever smoked (0.63% at baseline assessment in 1992). The Internal Review Board of Samsung Medical Center and Korea National Health Insurance Corporation reviewed and approved this study.
We related smoking assessed at the baseline examination (between April 1, 1992, and September 30, 1992) to cardiovascular events (fatal and nonfatal) occurring between October 1, 1992, and July 31, 2001. Participants with known cardiovascular disease (any of the outcomes considered here) at baseline were excluded. A total of 688 170 men were examined in 1992, of whom 1556 (0.3%) experienced a death or cardiovascular event before the start of the follow-up period, 13 898 (2.0%) had missing data on their smoking status, and 24 370 (3.5%) had missing data on other covariables considered in the analyses. These men were excluded from the study leaving 648 346 men (94.2%) for the analyses.
For our main analyses, we examined the associations of all outcomes with smoking status at the baseline examination classified into 5 mutually exclusive categories (never smoker, past smoker, current smoker <10 cigarettes/d, 10 to 19 cigarettes per day, and ≥20 cigarettes per day; participants were asked to record which of these prespecified categories was correct for their smoking status at baseline). In additional analyses, we also examined the effect of duration of smoking in those men who were current smokers at baseline (4 mutually exclusive categories: smoked for <10 years, 10 to 19 years, 20 to 29 years, or ≥30 years). Men who were past smokers at baseline were not asked about their previous smoking duration and therefore the effect of duration in past smoker could not be examined.
Fasting venous blood samples were taken and used to assess total cholesterol and glucose levels. A single measurement of blood pressure was taken using a standard mercury or electronic sphygmomanometer with the man rested and in the seated position. Weight and height (used to determine body mass index) were assessed using standard protocols with the men in light clothing and without shoes. At baseline, the men were asked about regular engagement in exercise and were categorized as engaging in regular exercise or not. The amount of alcohol consumed was calculated from questions that asked about drinking frequency per week and usual amount of drink (including questions about amount of the most popular Korean liquor, Soju) consumed at each drinking time. The weekly alcohol consumption level was divided into 5 categories: <30, 30 to 104, 105 to 209, 210 to 419, ≥420 g/week. Socioeconomic status was grouped into 4 levels based on quarters of the distribution of the man’s monthly salary level in 1992, and area of residence (capital, large city, small city, rural) was also considered as a potential confounding factor.
All nonfatal and fatal strokes, MI, and aortic aneurysms occurring between October 1, 1992, and July 31, 2001, were included in the analysis. The following codes in the Tenth Revision of International Classification of Diseases were used to identify and classify our outcomes: all strokes (I60–I69), ischemic stroke (I63, I67.8), intracerebral hemorrhagic stroke (I61), subarachnoid stroke (I60), MI (I21–I24), and aortic aneurysm (I71). Analyses of all cardiovascular disease included all of these outcomes. In this article, we assume that the clinical diagnoses of stroke subtypes and MI is correct; this is based on the findings from previous studies evaluating the accuracy of diagnosis of ischemic stroke (83.4%), hemorrhagic stroke (85.7%), and MI (85.6%) in a sample of hospital-admitted Korean public servants.16,17 Fatal cases were ascertained through data linkage with the mortality files of the Korean National Statistical Office. Nonfatal cases were ascertained through linkage with medical claim data of Korean National Health Insurance System, which provides information on all hospital admissions. To ensure that we only included participants with an actual diagnosis of a cardiovascular disease outcome (as opposed to those who may have had this included as a differential diagnosis but later discounted), we only included as a case those individuals who were admitted to the hospital for at least 48 hours. The easy access to hospital care for public servants together with the 10-year duration of follow-up increase the likelihood that the majority of nonfatal cases was obtained.
Follow-up, for the assessment of outcomes, began October 1, 1992 (the day after completion of the baseline assessment), and the participants were censored at the date of admission or death attributable to MI, any stroke, or aortic aneurysm; other causes of death; or July 31, 2001, whichever occurred first. Follow-up was complete for all participants. Hazard ratios for each cardiovascular outcome by current smoking status and duration of smoking were estimated by Cox proportional hazards regression analysis with an adjustment for age (year) at first and then for other covariates such as height (cm), blood pressure (categories: <120/80, 120 to 139/80 to 89, 140 to 159/90 to 99, ≥160/100 mm Hg), body mass index (categories: <18.5, 18.5 to 24.9, 25.0 to 29.9, ≥30.0 kg/m2), fasting total cholesterol (categories: <4.14, 4.14 to 5.16, 5.17 to 6.19, ≥6.20 mmol/L), fasting glucose (categories: <7.0, ≥ 7.0 mmol/L), alcohol consumption (categories: <30, 30 to 104, 105 to 209, 210 to 419, ≥420 g/week), regular exercise (binary), income (quarters), and area of residence (categories: capital, large city, small city, rural).
Proportionality assumptions were assessed by graphic methods and by exploring statistical interactions with the time scale. Weibull regression analysis results were compared for the models in which the proportionality assumption was not upheld and in all cases the effect estimates and confidence intervals were similar. Thus, only the results from the Cox proportional hazards models are presented. To identify any effect of preexisting occult disease, analyses were repeated after excluding events in the first 5 years of follow-up. The population proportional attributable risk (population etiologic fraction), which provides the proportion of the overall risk of the outcome in the population attributable to the exposure (current smoking), assuming that our association is causal in this population, was used as our measure of absolute effect.16 The population proportional attributable risk was estimated using the formula: PPAR=prevalenceexposure(HR-1)/1+prevalenceexposure(HR-1), where prevalenceexposure is the prevalence of the exposure (current smoking) and HR the hazard ratio from the multivariable model for any current smoking.18 Stratified Cox proportional hazards analysis according to cholesterol level and body mass index category were undertaken to determine whether the effect of smoking on cardiovascular disease outcomes was weaker in those of lower cholesterol or lower body mass index. All the analyses were performed using PC/SAS statistical package (SAS Institute Inc, Cary, NC).
Mean cholesterol level in this study population was 4.9 (SD: 1.0) mmol/L. Over the follow-up period, a total of 9475 men experienced at least one stroke (any subtype), 4768 (505 fatal) experienced at least one ischemic stroke event, 2380 (895 fatal) at least one intracerebral hemorrhage, and 786 (227 fatal) at least one subarachnoid hemorrhage; 3329 (1005 fatal) men experienced at least one MI and 269 (40 fatal) an aortic aneurysm. The associations of smoking with outcomes were essentially the same when we looked at fatal outcomes only and nonfatal outcomes only and therefore all results are presented for fatal and nonfatal outcomes combined. Table 1 shows baseline distributions of cardiovascular risk factors by smoking status. In total, 59% of the men were current smokers with 22% being past smokers. Men who smoked had lower blood pressure and body mass index but were less likely to undertake regular exercise and were more likely to consume large amounts of alcohol and to be in the lowest income categories.
Table 2 shows the multivariable associations of baseline smoking status with each cardiovascular outcome. There were strong linear trends of increased risk of ischemic stroke, subarachnoid hemorrhage, MI, and aortic aneurysm with greater amount of cigarette smoking. These associations remained with adjustment for baseline cardiovascular risk factors and were unchanged when the first 5 years of events were excluded from the analyses (Table 2). Smoking was not associated with intracerebral hemorrhage in either simple age-adjusted models or multivariable adjusted models, and there was statistical evidence that the effect of smoking on other stroke subtypes (ischemic stroke and subarachnoid stroke) and MI was stronger than the effect on intracerebral hemorrhage (all probability values >0.01).
Among current smokers, duration of smoking was positively associated with risk of MI with those who had smoked for longest having increased risk compared with those who had smoked for 10 years or less (see Figure). However, duration of smoking was not related to ischemic stroke, subarachnoid hemorrhage, or aortic aneurysm with hazard ratios being essentially the same in each duration category for these outcomes (data not shown).
Assuming a causal effect of smoking, these suggest that it is responsible for 26% of ischemic strokes, 35% of subarachnoid hemorrhage, 37% of MI, and 22% of aortic aneurysms in this population (Table 3). Overall, smoking was responsible for 20% of all of the cardiovascular disease outcomes examined in this study.
Table 4 shows the multivariable adjusted associations of smoking status with each outcome in strata of high and low total cholesterol and high and low body mass index. The increased risk of ischemic stroke, subarachnoid hemorrhage, MI, and aortic aneurysm associated with smoking was of a similar magnitude in men whose total cholesterol was below the median for this study population as in those with higher levels and was also of a similar magnitude in those men with a body mass index of less than 25 kg/m2 as in those who were overweight or obese. There was some evidence that the effect of smoking on intracerebral hemorrhage differed by body mass index with a modest increase in risk of this outcome associated with current smoking in those with a body mass index lower than 25 kg/m2 and no effect of smoking on those with a higher body mass index (P for interaction=0.003).
In this very large study of Korean men, we have shown important increases in the risk of ischemic stroke, subarachnoid hemorrhage, MI, and aortic aneurysm related to current smoking. Our results suggest that smoking is responsible for 20% of all cardiovascular disease events in this population. Mean cholesterol in this study population was 4.9 mmol/L, which compares with 5.5 mmol for US and UK men of a similar age (www.heartstats.org). In a previous publication from the same cohort, we demonstrated that, despite overall low levels of cholesterol in this population, total cholesterol was positively and linearly associated with coronary heart disease and ischemic stroke but had an inverse association with hemorrhagic stroke.19 On further inspection, the inverse association with hemorrhagic stroke was only apparent in individuals with hypertension and appeared to be mediated by heavy drinking.19 Overall, despite very different risk factor distributions, associations between established cardiovascular disease risk factors and disease events appear to operate similarly in Eastern populations to how they operate in Western populations. Consequently, public health interventions aimed at preventing cardiovascular disease that have been demonstrated to work in Western population, including those that prevent tobacco consumption and support quitting in those who already smoke,20 should be implemented in East Asian populations.
It has been suggested that smoking may not be an important risk factor for CHD in populations with low cholesterol levels.4–7 Our results clearly show that this is not the case. Even smoking at a relatively low level of one to 9 cigarettes per day was associated with a doubling of the risk of MI compared with never smokers in this study. Furthermore, the effect of smoking was similar for CHD, and other cardiovascular disease outcomes, in those men with total cholesterol levels below the median for this population compared with those with higher levels. It is also noteworthy that our participants are relatively young and in this study, like in Western populations, longer duration of smoking was associated with greater MI risk. If this cohort of men continues to smoke, the lifetime population attributable risk of CHD will be very large.
Our findings are consistent with the recent publication from the INTERHEART study, a large retrospective case–control study from 52 countries, which found that current smoking was associated with a 3-fold increase in the odds of MI in the whole study population.21 This effect was similar across all regions, including China/Hong Kong. Our findings are also consistent with findings from the prospective Asia Pacific Cohort Studies Collaboration, which found a similar magnitude of effect of smoking on risk of CHD in East Asian populations compared with individuals from Australia or New Zealand.8
Consistent with our study, a systematic review of studies conducted on Western populations concluded that smoking was not associated with intracerebral hemorrhage.11 A large case–control study (N=331 cases) that was published since that systematic review also found no association between smoking and intracerebral hemorrhage (OR: 1.07; 95% CI: 0.63 to 1.81).22 By contrast, researchers in the United States found positive associations of smoking with intracerebral hemorrhage in 2 cohort studies of women and men, with the associations with intracerebral hemorrhage being of a similar magnitude to those with subarachnoid hemorrhage and ischemic stroke.23,24 Our findings suggest no overall association of smoking with intracerebral hemorrhage. We examined associations of smoking with each outcome by strata of cholesterol and body mass index. The motivation for these stratified analyses was to further explore the suggestion that in populations with low overall cardiovascular risk, specifically those with low cholesterol levels and related to this low body mass index, smoking was not an important risk factor. Overall, our findings strongly contradict this suggestion in that within the whole cohort, smoking is strongly related to atherosclerotic cardiovascular diseases, and even within this overall low risk group, these associations are maintained in those with the very lowest cholesterol and body mass index (those below the median) for atherosclerotic outcomes. The finding of a positive association of smoking with intracerebral hemorrhage in those with low body mass index was a surprising finding; therefore, this subgroup analysis may be a chance finding and requires replication in other studies before one could conclude that it is real.
Study Strengths and Limitations
The major strengths of this study are its very large size, prospective design, and ability to compare effects for different stroke subtypes and different cardiovascular disease end points. Due to the large sample size, we were able to examine effects with precision even with the first 5 years of events removed to ensure that our results were not affected by changes in behavior in relation to early symptoms of disease. We were also able to examine whether the effects varied by total cholesterol and by body mass index. Smoking was self-reported, as it is in most other epidemiological studies, and we have no verification of the accuracy of this report in this study population. If there was some underreporting of current smoking or amount of smoking, then our results may be an underestimate of the true effect. Our outcomes were obtained from routine hospital discharge diagnoses or death certificates and there may be some inaccuracy, particularly by stroke subtype.12 However, previous studies provided supporting evidence that the accuracy of diagnosis of stroke subtypes and MI was quite acceptable. Furthermore, inaccuracy of diagnosis is less likely to differ across the different smoking groups.
Our findings demonstrate that smoking is just as important a risk factor for cardiovascular disease in populations with low cholesterol levels as it is in Western populations. Demonstrating this fact in large prospective studies is likely to be important for stimulating programs aimed at reducing smoking prevalence in low- and middle-income countries such as Korea.25,26
Sources of Funding
This study was supported by the Ministry of Health and Welfare Korea (#01-PJ1-PG1-01CH10-0007), the Samsung Biomedical Research Institute (#SBRI C-A7-416-1), and HL45522 from the National Heart, Lung and Blood Institute. D.A.L. is funded by a UK Department of Health Career Scientist Award.
- Received May 23, 2007.
- Revision received July 19, 2007.
- Accepted August 9, 2007.
Doll R, Peto R, Boreham J, Sutherland I. Mortality in relation to smoking: 50 years’ observations on male British doctors. BMJ. 2004; 328: 1519.
Parish S, Collins R, Peto R, Youngman L, Barton J, Jayne K, Clarke R, Appleby P, Lyon V, Cederholm-Williams S, et al. Cigarette smoking, tar yields, and non-fatal myocardial infarction: 14 000 cases and 32 000 controls in the United Kingdom. The International Studies of Infarct Survival (ISIS) Collaborators. BMJ. 1995; 311: 471–477.
Robertson TL, Kato H, Gordon T, Kagan A, Rhoads GG, Land CE, Worth RM, Belsky JL, Dock DS, Miyanishi M, Kawamoto S. Epidemiologic studies of coronary heart disease and stroke in Japanese men living in Japan, Hawaii and California Coronary heart disease risk factors in Japan and Hawaii. Am J Cardiol. 1977; 39: 244–249.
Asia Pacific Cohort Studies Collaboration. Smoking, quitting, and the risk of cardiovascular disease among women and men in the Asia–Pacific region. Int J Epidemiol. 2005; 34: 1036–1045.
World Health Organization. The Tobacco Atlas. Geneva: WHO; 2002.
Thrift AG, Donnan GA, McNeil JJ. Epidemiology of intracerebral hemorrhage. Epidemiol Rev. 1995; 17: 361–381.
Shinton R, Beevers G. Meta-analysis of relation between cigarette smoking and stroke. BMJ. 1989; 298: 789–794.
Iso H, Date C, Yamamoto A, Toyoshima H, Watanabe Y, Kikuchi S, Koizumi A, Wada Y, Kondo T, Inaba Y, Tamakoshi A; JACC Study Group. Smoking cessation and mortality from cardiovascular disease among Japanese men and women: the JACC Study. Am J Epidemiol. 2005; 161: 170–179.
Song YM, Sung J, Kim JS. Which cholesterol level is related to the lowest mortality in a population with low mean cholesterol level: a 6.4-year follow-up study of 482 472 Korean men. J Epidemiol. 2000; 151: 739–747.
Park JK, Kim KS, Kim CB, Lee TY, Lee KS, Lee DH, Lee S, Jee SH, Suh I, Koh K-W, Ryu SY, Park K-h, Park W, Wang S, Lee H, Chae Y, Hong H, Suh JS. The accuracy of ICD codes for cerebrovascular diseases in medical insurance claims. Korean J Prev Med. 2000; 33: 76–82.
Kirkwood BR, Sterne JAC. Measures of association and impact. In: Kirkwood BR, Sterne JAC, eds. Essential Medical Statistics. Oxford: Blackwell Science Ltd; 2003: 447–457.
Ebrahim S, Sung J, Song Y-M, Ferrer R, Lawlor DA, Davey Smith G. Serum cholesterol, haemorrhagic and ischaemic stroke, and myocardial infarction: the Korean National Health Service prospective cohort study. BMJ. 2006; 333: 22–26.
Shibuya K, Ciecierski C, Guindon E, Bettcher DW, Evans DB, Murray CJ. WHO framework convention on tobacco control: development of an evidence based global public health treaty. BMJ. 2003; 327: 154–157.
Teo KK, Ounpuu S, Hawken S, Pandey MR, Valentin V, Hunt D, Diaz R, Rashed W, Freeman R, Jiang L, Zhang X, Yusuf S; INTERHEART Study Investigators. Tobacco use and risk of myocardial infarction in 52 countries in the INTERHEART study: a case–control study. Lancet. 2006; 368: 647–658.
Kurth T, Kase CS, Berger K, Gaziano JM, Cook NR, Buring JE. Smoking and risk of hemorrhagic stroke in women. Stroke. 2003; 34: 2792–2795.
Kurth T, Kase CS, Berger K, Schaeffner ES, Buring JE, Gaziano JM. Smoking and the risk of hemorrhagic stroke in men. Stroke. 2003; 34: 1151–1155.
Chen Z, Peto R. Stopping smoking works. University of Oxford Annual Review 1998. Available at: www.ox.ac.uk/publicrelations/pubs/annualreview/ar98/amoking.shtml. Accessed September 3, 2006.
Ebrahim S, Smeeth L. Non-communicable diseases in low and middle-income countries: a priority or a distraction? Int J Epidemiol. 2005; 34: 961–966.