Both physical inactivity and cigarette smoking are major risk behaviours for coronary heart disease. The estimated population attributable risk for physical inactivity is about 33%1 and about 17% for smoking.2 Exercise and smoking cessation are important strategies to reduce the social burden of coronary heart disease.3 The reported benefits of exercise include lowering of blood pressure levels,4 increasing insulin sensitivity5 and high-density lipoprotein (HDL)-cholesterol levels6 and improvement of endothelial function.7 On the other hand, smoking is associated with the development of hypertension,8 impaired insulin sensitivity,9 reduction of HDL-cholesterol levels10 and impaired endothelial function.11 These effects of smoking may weaken the beneficial effects of exercise for prevention of coronary heart disease.

We reported previously the presence of an inverse association between sports participation and mortality from coronary heart disease among Japanese aged 40–79 years.12 In this study we examined whether smoking modifies the relationship between sports participation and mortality from coronary heart disease among Japanese men and women, using the same cohort with an extended follow-up.

METHODS

Study cohort and baseline questionnaire

The Japan Collaborative Cohort Study for Evaluation of Cancer Risk sponsored by Monbusho (JACC study) was undertaken from 1988 to 1990, when 110 792 people (46 465 men and 64 327 women) aged 40–79 years and living in 45 communities across Japan participated in municipal health screening examinations and completed self-administered questionnaires about their lifestyles and medical histories of previous cardiovascular disease and cancer.13 Informed consent was obtained from participants or the community representatives for conducting an epidemiological study before completing the questionnaire, based on guidelines of the Council for International Organisations of Medical Science.14 The ethics committee of the University of Tsukuba approved this study.

A total of 35 773 men and 45 344 women provided a valid response to questions about smoking: their smoking status (current, ex- or never smoking) and amount of cigarette smoking a day;15 and the average weekly time spent in athletic and sporting events: “What is the average amount of time you spend engaging in sports watch week?” with a list of possible answers of “<1 hour”, “1–2 hours”, “3–4 hours” and “⩾5 hours”.12 The validity of the estimated time spent participating in sports and leisure time physical activity was examined in a random sample of 739 men and 991 women from the baseline participants.16 The Spearman rank correlation for the reported time spent participating in sports with leisure-time physical activity reported during the past 12 months, and that estimated from a structured interview based on the Minnesota leisure-time physical activity questionnaire,17 was 0.53 in men and 0.58 in women. The reliability of engagement in sports was examined 12 months apart in a random sample of 416 men and 636 women. We obtained modest κ coefficients of 0.45 in men and 0.40 in women for time spent participating in sports (four categories).

We excluded 2002 men and 2283 women from the analysis because of a previous history of stroke, coronary heart disease, or cancer at the time of baseline inquiry. Therefore, a total of 76 832 participants (33 771 men and 43 061 women) were enrolled in the present study.

Mortality surveillance

For mortality surveillance, the investigators conducted a systematic review of death certificates, all of which were forwarded to the local public health centre in each community. It is believed that all deaths that occurred in the cohort were ascertained, except for subjects who died after they had moved from their original community, in which case the subject was treated as a censored case. Mortality data were sent centrally to the Ministry of Health and Welfare, and the underling causes of death were coded for the National Vital Statistics according to the International Classification of Disease (ICD), 10th revision. Cause-specific mortality was determined by the total deaths due to coronary heart disease (ICD, 10th revision, codes I20–I25) and myocardial infarction (I21). The mortality follow-up inquiry for this study was conducted until the end of 2003, and the average follow-up was 12.5 years.

Statistical analysis

Statistical analysis was based on cause-specific mortality rates. Person-years of follow-up were calculated from the date of the baseline questionnaire to the date of death, moving from the community, or the end of 2003, whichever occurred first. The hazard ratio of mortality from coronary heart disease was defined as the death rate for the participants within the four categories of time spent in sports participation divided by the corresponding rate among those who reported 1–2 hours/week average time in sports. Because the subjects in the lowest categories (<1 hour/week average time in sports) were more likely to have ill health and the goal of this study was to study physical activity in healthy subjects, the second lowest categories (1–2 hours/week average time) in adjusted means and proportions of selected cardiovascular risk factors at baseline inquiry were presented among the categories of sports participation, stratified by smoking status.

Differences in the mean values and proportions relative to those who reported 1–2 hours/week average time in sports were examined, using the t-test or χ² test. Multivariable hazard ratios and 95% confidence intervals were calculated after adjustment for sex, age and potential confounding factors by using the Cox proportional hazards model. These factors included body mass index (quintiles), history of hypertension (no versus yes), history of diabetes (no versus yes), alcohol intake category (never, ex-drinker and current drinker of ethanol at 1–22, 23–45, 46–68 and ⩾69 g/day), hours of sleep (<6.0, 6.0–6.9, 7.0–7.9, 8.0–8.9 and ⩾9.0 hours/day), age of completed education (<13, 13–15, 16–18 and >19 years old), job style (manual worker, office worker and non-worker), perceived mental stress (low, medium and high), frequency of fish intake (0, <1, 1–2, 3–4 and ⩾5 times/week), and the average daily time spent walking (<0.5 hour, 0.5 hour, 0.6–0.9 hour and ⩾1.0 hour).

Test for effect modification by smoking status was conducted with an interaction term generated by multiplying the sports participation categories by smoking status. Because there was a difference between men and women in the prevalence of smoking status, we also presented the data stratified by sex. To clarify an effect of smoking itself for mortality from coronary heart disease, we also calculated the hazard ratios according to combined variables of smoking status and time spent in sports, compared with never smokers with 1–2 hours/week of sports participation as reference. We tested the assumption of proportional hazards according to time spent in sports participation, by using both a time-dependent covariate method and a linear correlation test,18 and found no violation for proportionality.

All probability values for statistical significance were two-tailed, and all confidence intervals were estimated at the 95% level. All statistical analyses were conducted using SAS, version 9.13 (SAS Institute, Inc, Cary, NC, USA).

RESULTS

A total of 76 832 participants were followed up for an average of 12.7 years; 638 deaths from coronary heart disease (496 myocardial infarction) were identified.

Table 1 shows selected cardiovascular risk factors according to the four categories of time spent in sports stratified by smoking status. Compared with never smokers who reported 1–2 hours/week of time spent in sports (second lowest category), those who reported ⩾5 hours/week (highest category) were male, older and office workers, educated less, had a high dietary intake of ethanol, less likely to have a history of hypertension, and walked more. Compared with ex-smokers who reported 1–2 hours/week, those who reported ⩾5 hour/week were older, had a high dietary intake of fish and ethanol, slept more, and walked more. Compared with current smokers who reported 1–2 hours/week, current smokers who reported ⩾5 hours/week were older and more likely to be manual workers, have more sleep, and walk more. After sex stratification, these associations were not altered substantially.

Table 2 shows multivariable hazard ratios of mortality from coronary heart disease adjusted for age an sex according to time spent in sports. Compared with people who reported 1–2 hours/week of time spent in sports, those who reported ⩾5 hours/week had approximately 50–80% lower sex and age-adjusted mortality from coronary heart disease among never and ex-smokers, although there was no association among current smokers. Adjustment for known cardiovascular risk factors did not alter materially these associations. The multivariable hazard ratios (95% confidence intervals) for the ⩾5 hours/week versus 1–2 hours/week of time spent in sports for mortality were 0.44 (0.23 to 0.86), p = 0.02 among never smokers, 0.18 (0.05 to 0.60), p>0.01 among ex-smokers and 0.82 (0.47 to 1.40), p = 0.46 among current smokers for coronary heart disease, and 0.44 (0.21 to 0.90), p = 0.03 among never smokers, 0.16 (0.04 to 0.71), p = 0.02 among ex-smokers and 0.96 (0.53 to 1.75), p = 0.90 among current smokers for myocardial infarction. The multivariable hazard ratios among non-smokers (never and ex-smokers) were 0.35 (0.20 to 0.62), p<0.001 for coronary heart disease (for interaction between non-smokers and current smokers p = 0.08), and 0.35 (0.19 to 0.66), p = 0.001 for myocardial infarction (for interaction between non-smokers and current smokers p = 0.06) (data not shown in the table). These inverse associations for sports participation with mortality were not altered substantially when deaths occurring within 2 years after baseline were excluded (data not shown in the table).

When examined by sex, these associations were found to be similar between men and women. We could not have stable hazard ratios owing to the small sample size among never smoking men and ex- or current smoking women. However, ex-smoking men and never smoking women with ⩾5 hours/week of time spent in sports had lower mortality from coronary heart disease and myocardial infarction than those with 1–2 hours/week of time spent in sports. The multivariable hazard ratios (95% confidence intervals) for the ⩾5 hours/week versus 1–2 hours/week of time spent in sports for mortality were 0.17 (0.05 to 0.57), p<0.01 among ex-smoking men and 0.34 (0.15 to 0.78), p = 0.01 among never smoking women for coronary heart disease, and 0.15 (0.03 to 0.67), p = 0.01 among ex-smoking men and 0.35 (0.14 to 0.86), p = 0.02 among never smoking women for myocardial infarction.

To examine the effect of smoking itself for mortality from coronary heart disease, we showed the multivariable hazard ratios in combined variables of smoking status and sports participation (fig 1). Compared with never smokers with 1–2 hours/week of sports participation (as reference), current and ex-smokers had higher mortality from coronary heart disease, except for ex-smokers with ⩾5 hours/week of sports participation. The multivariable hazard ratios (95% confidence intervals) of mortality from coronary heart disease for current smokers were 2.25 (1.53 to 3.31), p<0.001 among people with <1 hour/week, 2.24 (1.39 to 3.61), p<0.001 among people with 1–2 hours/week, 1.65 (0.92 to 2.96), p = 0.10 among people with 3–4 hours/week, and 1.69 (0.96 to 2.96), p = 0.07 among people with ⩾5 hours/week of time spent in sports. The respective values for ex-smokers were 1.33 (0.86 to 2.05), p = 0.20, 1.57 (0.92 to 2.70), p = 0.10, 1.59 (0.83 to 3.06), p = 0.16, and 0.29 (0.09 to 0.95), p = 0.04.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1 Multivariable hazard ratios of mortality from coronary heart disease according to time of sports participation and smoking status, using never smokers with 1–2 hours/week of sports participation as reference group.

DISCUSSION

In the present large prospective study of Japanese men and women aged 40–79 years, we found that smoking modified the relationship between sports participation and mortality from coronary heart disease. Never and ex-smokers had about 50% and 80% lower mortality from coronary heart disease, although current smokers showed no association between sports participation and mortality from coronary heart disease. These relationships were not substantially altered after adjustment for known risk factors or sex stratification. To our knowledge, this is the first cohort study to show that smoking may reduce the beneficial effect of exercise for prevention of fatal coronary heart disease.

A previous study of health professional women in the United States showed that physical activity was inversely associated with the risk of coronary heart disease among current and ex-smokers, but not among never smokers.19 We showed an association between sports participation and mortality from coronary heart disease among ex- or never smokers, but not current smokers. Reasons for the different effect modification between the present and previous studies were uncertain, but it might be due to differences in population characteristics such as smoking status. The proportion of heavy smokers (⩾15 cigarettes a day) among current smokers was higher in the present study than in the previous study (77.5% vs 63%).

The plausible mechanisms for this interaction are as follows. First, the adverse effects of smoking may weaken the beneficial effects of exercise on the cardiovascular risk factors which affect risk of coronary heart disease,20^–22 because both smoking and physical inactivity similarly lead to higher blood pressure levels,4 8 impaired insulin sensitivity,5 9 lower HDL-cholesterol levels6 10 and impaired endothelial function.7 11 Second, current smokers are less likely than non-smokers to engage in exercise of sufficient intensity,23 which leads to the failure to prevent coronary heart disease.24 Smoking may also reduce the capacity for highly intense exercise and hence reduce the intensity of exercise necessary to prevent coronary heart disease.

The limitations of the present study include the absence of systematic information on preclinical disorders that prevented the participants from participating in exercise. This might have lead to a bias of cause–effect reversal, even though most of the subjects were apparently healthy. To avoid potential bias due to preclinical disorders and/or psychosocial distress, we chose the second lowest physical activity categories as a reference. This probably led to underestimation of hazard ratios. We also repeated the data analysis by excluding early deaths, which would also reduce the potential confounding effect of preclinical disorders.

Second, we had no data on history of dyslipidaemia (high total or low-density lipoprotein-cholesterol and low HDL-cholesterol levels), which is an established risk factor of coronary heart disease. However, dyslipidaemia might be a mediator but not a confounding factor for coronary heart disease, because smoking or physical inactivity lead to dyslipidaemia,6 10 which is a risk factor for coronary heart disease.21

In conclusion, this study provides epidemiological evidence that smoking may reduce the beneficial effect of sports participation for reduction of fatal coronary heart disease. Smoking itself also increased mortality from coronary heart disease, and there seems to be no effects of sports participation on prevention of coronary heart disease among smokers, unless they quit smoking. Smoking prevention or cessation would be of value to enhance a preventive effect on coronary heart disease.