Article Text
PDF
Abstract
Background Few studies have prospectively evaluated the association of leisure-time physical activity (LTPA) with mortality in Asians, who are more susceptible to insulin resistance than their Caucasian counterparts.
Methods Data from two large prospective cohort studies conducted in Shanghai were evaluated. After excluding participants who had a history of cancer, coronary heart disease or stroke at baseline, or who died within the first 3 years after study enrolment, 53 839 men and 66 888 women, followed for an average of 9.2 and 14.7 years, respectively, remained for the study.
Results Compared with those who reported no exercise, a reduction in mortality with an HR of 0.86 (95% CI 0.80 to 0.93) was observed in those who regularly engage in moderate-intensity LTPA, even those who reported an LTPA level lower than the minimum amount recommended by the current physical activity guidelines (150 min or 7.5 metabolic equivalent hours per week). The association between moderate-intensity exercise and mortality followed a dose–response pattern until the amount of LTPA reached 3–5 times the recommended minimum level. A similar pattern of association was observed for cause-specific mortality due to cardiovascular disease (CVD), cancer or other causes.
Conclusion Regular participation in moderate-intensity LTPA was associated with reduced mortality, particularly CVD mortality, even when the LTPA was below the minimum level recommended by current guidelines. Increasing the amount of moderate-intensity LTPA was associated with further risk reduction up to a potential threshold of 3–5 times the recommended minimum.
- physical activity
- mortality
- cancer
- exercise
- cardiovascular disease
Statistics from Altmetric.com
Introduction
Regular leisure-time physical activity (LTPA) has been shown to be associated with a reduced risk of all-cause mortality in many studies, including several recent large prospective cohort studies.1–5 Both the 2008 Physical Activity Guidelines for Americans6 and the WHO 2010 Global Recommendations on Physical Activity for Health7 recommend at least 150 min per week of moderate-intensity LTPA for substantial health benefits. However, these guidelines were developed primarily based on research results from studies conducted in Europe and North America.
Obesity is a major risk factor for insulin resistance; Asians are more susceptible to insulin resistance than people of European ancestry.8 Physical activity has been shown to improve insulin sensitivity.9 It has been shown that at the same body mass index (BMI) level the risk for type 2 diabetes is higher in Asians than other ethnic groups.10 11 Therefore, it is important to quantify the association between LTPA and the risk of death in Asians because the amount and intensity of LTPA associated with mortality for Asians might be different from that observed in European-ancestry populations. Moreover, moderate-intensity LTPA is often the exclusive form of exercise practised by middle-aged and elderly adults, particularly in Asian populations. In addition, providing evidence of the potential benefit of LTPA could incent increased LTPA participation by Asians, who account for 60% of the world’s population.
In the current study, which includes 120 727 middle-aged and elderly Chinese adults with long-term follow-up, we examined the association of overall and cause-specific mortality with moderate-intensity LTPA across a wide range of activity levels. In addition to health benefits potentially conferred by a minimum amount of physical activity, we also attempted to determine at what point additional LTPA confers no additional benefit in reducing mortality risk.
Methods
Study population
Data used in this study were from two population-based cohort studies, the Shanghai Men’s Health Study (SMHS) and the Shanghai Women’s Health Study (SWHS). The design and methods of these studies have been previously described in detail elsewhere.12 13 In brief, through in-person interviews of residents living in selected urban communities in Shanghai, the SMHS recruited 61 480 men, 40–74 years old, between 2002 and 2006, with a response rate of 74.0%. The SWHS, between 1996 and 2000, recruited 74 941 women, 40–70 years old, with a response rate of 92.7%. Detailed baseline surveys were conducted using structured questionnaires to obtain information on demographic characteristics, lifestyle habits, personal and familial disease history, and physical activity level.12 13 Anthropometric measurements were also taken at baseline by trained interviewers. Written informed consent was obtained from all study participants.
Cohort members were followed up for cause-specific mortality and incidence of site-specific cancers and selected chronic diseases through a combination of in-person surveys and annual record linkages to the Shanghai Cancer Registry and the Shanghai Vital Statistics Registry. Follow-up for participants’ mortality data was virtually complete. The primary endpoint for this analysis was death from any cause that occurred after the baseline survey but before 31 December 2013. For each death, the cause was determined from the death certificate and coded by International Classification of Diseases, Ninth Revision.14 For cause-specific mortality analyses, we evaluated deaths from cardiovascular disease (CVD) (codes 390–459 and 798), cancer (codes 140–208) and all other causes. The mean follow-up time was 9.2 years (SD=1.8 years) for the SMHS and 14.7 years (SD=2.3 years) for the SWHS.
Measures
Participants were asked to provide information on exercise participation during the 5-year period preceding the baseline interview. Respondents who reported participating in regular exercise (defined as engaging in LTPA of any type and intensity at least once per week for three consecutive months) were asked to describe up to three activities in which they most frequently engaged. They were asked questions about the average amount of time spent per week on each activity and the number of years of participation. Information was also obtained from all participants on physical activity essential to daily living, including household activities, and walking and cycling for transportation.
The reproducibility and validity of the physical activity questionnaire used in the SMHS and SWHS were evaluated in 196 participants and 200 participants, respectively, and found to be reasonably high.15 16 The test–retest correlation for exercise participation assessed using the physical activity questionnaire at dates 1 year apart was 0.68 for the SMHS. The 2-year test–retest correlation was 0.64 for the SWHS. The correlations of exercise reported on the second physical activity questionnaire with exercise reported on 7-day recalls and on activity logs were 0.62 and 0.71 for the SMHS, and 0.80 and 0.74 for the SWHS, respectively.
To assess exercise intensity, a standard metabolic equivalent (MET) score was assigned to each activity, based on the Compendium of Physical Activities developed by Ainsworth et al.17 18 We defined the intensity of each type of LTPA as moderate (3.0–5.9 METs) or vigorous (6.0 or more METs).6 The most frequently reported LTPA were tai chi (4.0 METs), dancing (4.5 METs) and walking (3.0 METs), which are all moderate-intensity exercises. Each participant was then categorised into one of the following four groups: no regular exercise (referent), moderate-only exercise, vigorous-only exercise, or both vigorous and moderate exercise.
Statistical analyses
Of the 136 421 participants from the SMHS and the SWHS, some were excluded from the analysis for the following conditions: due to loss to follow-up shortly after the baseline survey (6 men, 3 women); and a diagnosis or self-report at baseline of cancer (1953 men, 1598 women), coronary heart disease (3308 men, 5531 women), stroke (2322 men, 883 women) or no information collected on smoking (1 man). We further excluded subjects with less than 3 years of follow-up (1102 men, 753 women). After applying the exclusion criteria (not mutually exclusive), 120 727 participants (53 839 men, 66 888 women) remained for the primary analysis. Nearly 90% of the participants who reported any exercise in our study engaged in only moderate-intensity exercise. For ease of comparison, we further excluded participants who reported any vigorous-intensity LTPA (3467 men, 1121 women) from analyses quantifying the association of the amount of moderate-intensity LTPA with risk of death.
The amount of LTPA was quantified for each reported activity by multiplying the MET score by the time spent on the activity each week, giving MET-hours/week. Summing the MET-hours/week for each activity for a participant produced a value for total MET-hours/week of LTPA. This value was used to assign the participant to one of these six categories: no regular exercise (referent), exercise not meeting the minimum exercise guideline level (0.1 to <7.5 MET-hours/week), 1–2 times the minimum guideline level (7.5 to <15.0 MET-hours/week), 2–3 times (15.0 to <22.5 MET-hours/week), 3–5 times (22.5 to <37.5 MET-hours/week), and greater than or equal to 5 times the minimum guideline level (≥37.5 MET-hours/week). In addition, we conducted analyses stratified by age at baseline (40–59 years, ≥60 years), presence of comorbidity of hypertension or diabetes at baseline survey (yes, no), BMI at baseline (normal weight, overweight) and smoking status (never, ever) to evaluate potential modifying effects of these variables.
Multivariate-adjusted HRs and 95% CIs for the association of LTPA with mortality were estimated using Cox proportional hazards models with age as the time scale. The entry time was defined as age at the baseline interview, and the exit time was defined as age at death, age at the last follow-up or on 31 December 2013, whichever occurred first. Statistical models were stratified by year of birth (categorised into seven 5-year groups) and adjusted for the following covariates: education (elementary school or less, middle school, high school graduate, some college or higher); household income (low, middle, high); marital status (currently married, single/separated/divorced/widowed); smoking (men: never smoked, <10 pack-years, 10.0–19.9 pack-years, 20.0–39.9 pack-years, ≥40 pack-years; women: ever smoked cigarettes regularly (yes, no); the adjustment categories of smoking in men and women were different because less than 3% of the female participants in our data ever smoked, whereas two-thirds of the male participants were former or current smokers); ever consumed alcohol regularly (yes, no); non-LTPA daily activity (quartiles); and ever diagnosed with hypertension (yes, no), diabetes (yes, no), chronic hepatitis (yes, no) or chronic respiratory diseases (yes, no).
We checked the proportional hazards assumption graphically by comparing log-log survival curves and by testing the significance of interaction terms for exposures with the follow-up time. We found no evidence of departure from the assumption. Tests of linear trend were performed by treating a categorical variable as ordinal, and the interaction between two variables was estimated by comparing the −2 log likelihood of models with or without interaction terms. Statistical analyses were performed using SAS v 9.4. All statistical tests were based on two-sided probability.
Results
Selected characteristics for the cohorts by regular exercise status (any or none) are presented in table 1. Both male and female participants who reported regular exercise were older (P<0.001). Because of a notable difference in age between exercisers and non-exercisers, we report age-adjusted differences for all other variables except for age. Some differences, generally small, between exercisers and non-exercisers were found in waist-to-hip ratio; education; income; lifestyle, including smoking and alcohol consumption; comorbidity for hypertension and diabetes; and non-LTPA daily activities, for both men and women. Regular exercisers were more likely to have smaller waist-to-hip ratio, higher education and income, and were less likely to regularly smoke and consume alcohol. Men who exercised regularly also had a higher amount of non-LTPA daily activities, while the opposite association was observed for women. BMI for male exercisers was higher than for non-exercisers, but no difference was found between exercising and non-exercising women. We observed the difference in marital status and comorbidity for chronic liver disease and respiratory disease between exercisers and non-exercisers in women, but not in men. Exercisers tend to be more likely to have a prior diagnosis of a chronic disease, except chronic liver disease for both sexes, and respiratory disease for men.
Selected baseline characteristics* of study participants by regular exercise status, Shanghai cohort studies, 1996–2013
HRs* and 95% CIs of mortality associated with exercise, Shanghai cohort studies, 1996–2013
Of the 120 727 eligible subjects included in this analysis, 7549 deaths were reported, including 2197 deaths due to CVD, 3512 deaths due to cancer and 1840 due to other causes. Compared with men who reported no regular LTPA, men who engaged regularly in moderate-intensity-only LTPA had decreased mortality due to all causes, CVD and cancer (all-cause: HR=0.82, 95% CI 0.76 to 0.89; CVD: HR=0.77, 95% CI 0.66 to 0.90; and cancer: HR=0.83, 95% CI 0.74 to 0.93) (table 2). The risk of death due to other causes for men who engaged regularly in moderate-intensity-only LTPA was also reduced, although the P value for their HR estimates was greater than 0.05. Similar but slightly weaker associations were found for women, although not all HR estimates were less than 0.05. A much smaller number of study participants ever engaged in vigorous-intensity LTPA than moderate-intensity-only LTPA. One in three participants in these Shanghai cohorts reported to have some LTPA, while only 6.4% of the male and 1.7% of the female subjects ever participated regularly in vigorous-intensity LTPA. Because of the small sample size, the estimates of risk for vigorous-intensity LTPA participation were unstable. Nevertheless, it did not appear that those who regularly participated in vigorous-intensity LTPA experienced a lower risk of death than those who participated in moderate-intensity only LTPA (data not shown). The pattern of the association was similar between men and women (P for interaction >0.05).
Subsequent analyses were performed for participants in moderate-intensity-only LTPA. Compared with those who reported no regular exercise, regular participation in moderate-intensity LTPA below the minimum recommended physical activity level (<7.5 MET-hours/week) was associated with a decreased all-cause mortality risk (HR=0.86, 95% CI 0.80 to 0.93) and decreased CVD mortality (HR=0.75, 95% CI 0.65 to 0.86) in the combined analysis of data from men and women (table 3). The risk was reduced in a dose–response manner with increasing levels of LTPA, up to 22.5–37.4 MET-hours/week, 3–5 times the minimum guideline level. Exercise above 5 times the minimum guideline level was not associated with further reduction in risk, although the risk of death remained lower than for those who reported no regular participation in LTPA. A linear trend was found for both all-cause and CVD mortality with levels of exercise (Ptrend <0.001).
HRs* and 95% CIs of mortality associated with moderate-intensity exercise, Shanghai cohort studies, 1996–2013
Although an inverse association between LTPA level and risk of cancer death was observed in the combined analysis of men and women (Ptrend <0.05), the reduction in risk was only apparent in LTPA groups above the minimum guideline level. Similar to the association observed for all-cause and CVD mortality, the lowest risk for cancer death was observed among those who participated in LTPA at 3–5 times the minimum guideline level. For other cause mortality, an inverse association with LTPA level was observed in the combined analysis of men and women (Ptrend <0.05).
No apparent modifying effects on the association between moderate-intensity LTPA and mortality were found for age, comorbidity status for hypertension or diabetes, BMI at baseline, non-LTPA, or smoking status (table 4). The inverse association between LTPA and all-cause mortality was more apparent in older (≥60 years) than younger subjects, although the P value for the test for interaction was greater than 0.05. In virtually all subgroups, the lowest risk of death due to any cause was observed among those who exercised at 3–5 times of the minimum guideline level.
Stratified HRs* and 95% CIs of mortality associated with moderate-intensity exercise, Shanghai cohort studies, 1996–2013
Discussion
In this large prospective study of middle-aged and elderly Chinese men and women, we found that even a small amount of moderate-intensity LTPA, including levels less than the minimum amount of physical activity recommended by the 2008 Physical Activity Guidelines for Americans6 and the WHO 2010 Global Recommendations on Physical Activity for Health,7 was associated with decreased risk of death due to all causes, CVD or other causes. The lowest risk of death was found among those who exercised at 3–5 times the minimum guideline level for moderate-intensity LTPA; no further reduction in risk was noted with exercise above this level. Although an inverse association between LTPA levels and risk of cancer death was also observed, the association with cancer death risk was weaker than that observed for CVD mortality; a reduction in cancer mortality was apparent only at or above the minimally recommended physical activity level.
Although LTPA has been widely studied in relation to disease mortality, only a few previous studies were conducted in Asian populations.5 19–22 Despite variation in the measurement for LTPA, all these studies found that exercise is associated with reduced disease mortality. Specifically, Wen et al 5 reported a profound reduction in mortality and prolonged life expectancy associated with moderate-intensity exercise in Taiwan. The findings from the Taiwan study are similar to ours, although that study had a much younger study population (mostly <40 years old) of above-average socioeconomic status. Ueshima et al 19 and Inoue et al,22 separately, reported reduced mortality from all-causes and cancer in association with daily total physical activity in cohorts in Japan. However, these two Japanese studies did not quantify the impact of LTPA on mortality by the amount of LTPA as recommended in the physical activity guidelines.
The exercise pattern in China is quite different compared with some other populations. A large majority of study participants engaged in moderate-intensity LTPA only, and thus this study provides a unique opportunity to evaluate this important form of exercise practised exclusively in many middle-aged and elderly adults. We reported previously that retirement is highly correlated with LTPA participation.23 Retirees make up a higher percentage of the older age groups than the younger age groups, which explains the observation that exercise groups are considerably older compared with the non-exercise groups.
In our study, we found an inverse association between LTPA and disease mortality regardless of sex, age, comorbidity status for hypertension or diabetes, BMI at baseline, non-LTPA, or smoking status. In particular, we showed that subjects who had a diagnosis of hypertension or diabetes at baseline also experienced a reduction in disease mortality by regularly participating in LTPA, even an amount less than the recommended minimum physical activity level. This finding was supported by several previous studies, in which subjects with baseline comorbidity conditions, such as cancer and heart disease, showed benefit from regular exercise after the diagnosis of these diseases.24 25
Various mechanisms have been proposed to explain the health benefit of LTPA. LTPA participants are less likely to be obese, which reduces mortality due to both CVD and cancer via complex pathways including inflammation and insulin sensitivity regulation. Further, a reduction in adipose tissue decreases oestrogen biosynthesis in postmenopausal women, which contributes to a reduced risk of breast and other oestrogen-related cancers. On the other hand, LTPA has been shown to improve lipid lipoprotein profiles, reduce blood pressure and coagulation, and improve endothelial functions, which are associated with reduced CVD risk and mortality.26 These mechanisms are complex and further investigations are needed.
The sample size of this study is large, which enabled detailed analysis of the association of LTPA across a wide range of activity levels. A comprehensive assessment of major disease risk factors was conducted at baseline, allowing us to carefully adjust for the potential influence of confounding factors. The follow-up rate is high, enhancing the validity of study findings. A major limitation of this study is that the physical activity information was self-reported, and thus there might be some errors in reported intensity and duration of the exercise. These errors, however, are likely to be non-differential, which typically attenuates the true association. Furthermore, we have shown that the questionnaire used in our study has a reasonably high level of validity and reproducibility using different methods in assessing LTPA levels, especially assessing the totality of LTPA of the questionnaires.15 16 We asked participants to report up to three items of sustained exercises they conducted at least once a week for over 3 months, which might not capture all the LTPA they engaged in. Although some may engage in multiple activities for exercise, we found more than 95% of participants in our study engaged in two or fewer activities, among those who reported any exercise. Moreover, we collected more detailed information than the conventional method, and our method yielded higher validity and reproducibility compared with the method including a long list of activities.15 16 Some subjects may have made changes in their exercise level due to a disease diagnosis, which may have led to bias due to reverse causation. To minimise this bias, we excluded participants with less than 3 years of follow-up. Furthermore, we adjusted for baseline major comorbidity conditions, and the results were essentially unchanged.
In summary, results from our study, along with those reported from previous studies, suggest that middle-aged and elderly adults could benefit from LTPA at the minimum level recommended by the current physical activity guidelines. The association between LTPA and disease mortality is unlikely to be linear, and an inflection point may be near the level of 3–5 times the minimum level of physical activity recommended by the current guidelines. Our study provides additional support to promote physical activity to reduce mortality in Asian populations.
What is already known on this subject
Leisure-time physical activity (LTPA) has been consistently linked with reduced all-cause and cardiovascular mortality.
Current physical activity guidelines recommend the minimum level of physical activity for substantial health benefit. It is less clear whether there is any potential health benefit for LTPA at the amount below the currently recommended minimum level and possible non-linear association of mortality with very high level of LTPA.
Few studies have quantified the association of LTPA with total and cause-specific mortality in East Asians.
What this study adds
Analyses of data from large prospective cohorts in middle-aged and elderly Chinese show that moderate-intensity LTPA is associated with reduced risk of death due to all causes combined, cardiovascular diseases, cancer and other diseases.
A reduced mortality was found even among those who exercised at the amount below the minimum level recommended by the current physical activity guidelines. The lowest risk of death was found among those who exercised at 3–5 times the minimum guideline level for moderate-intensity LTPA; no further reduction in risk was noted with exercise above this level.
Acknowledgments
We thank Dr Hui Cai and Dr Yong Cui for their assistance in analysing the data, and Ms Kimberly A Kreth for her assistance in preparing the manuscript. We also would like to thank the staff members and the study participants of the Shanghai Men’s Health Study and the Shanghai Women’s Health Study for their contributions to this research.
References
Footnotes
Contributors YL and WZ designed the research. WW, Y-TG, H-LL, GY, Y-BX, X-OS and WZ collected the data. YL and WW analysed the data. YL and WZ wrote the paper and had primary responsibility for final content. All authors interpreted the data, read and approved the final manuscript.
Funding This work was supported by grants from the US National Institutes of Health (R01 CA082729, UM1 CA173640 to X-OS) and (R37 CA070867, UM1 CA182910 to WZ). The study sponsor had no role in study design; collection, analysis and interpretation of the data; writing of report; and decision to submit the report for publication.
Competing interests None declared.
Ethics approval Approval for human subjects research in these studies was granted by the institutional review boards of Shanghai Cancer Institute and Vanderbilt University Medical Center (00340 and 000598).
Provenance and peer review Not commissioned; externally peer reviewed.