Article Text
PDF
Abstract
Objectives We aimed to investigate the dose–response associations of long-term leisure-time physical activity (LTPA) obtained from repeated measures with all-cause and cardiovascular disease (CVD) mortality outcomes in Taiwanese adults.
Methods We included 210 327 participants with self-reported LTPA at least in two medical examinations (867 968 data points) for up to 20 years (median, IQR: 4.8 years, 2.3–9.0). Dose–response relationships were modelled with restricted cubic spline functions and Cox regressions HRs (95% CIs) adjusted for main covariates.
Results During up to 23 years of follow-up (3 655 734 person-years), 10 539 participants died, of which 1919 of CVD. We observed an inverse, non-linear dose–response association between long-term LTPA and all-cause and CVD mortality. Compared with the referent (0 metabolic equivalent of task (MET) hours/week), insufficient (0.01–7.49 MET hours/week), recommended (7.50–15.00 MET hours/week) and additional (>15 MET hours/week) amounts of LTPA had a lower mortality risk of 0.74 (0.69–0.80), 0.64 (0.60–0.70) and 0.59 (0.54–0.64) for all-cause mortality and 0.68 (0.60–0.84), 0.56 (0.47–0.67) and 0.56 (0.47–0.68) for CVD mortality. When using only baseline measures of LTPA, the corresponding mortality risk was 0.88 (0.84–0.93), 0.83 (0.78–0.88) and 0.78 (0.73–0.83) for all-cause and 0.91 (0.81–1.02), 0.78 (0.68–0.89) and 0.80 (0.70–0.92) for CVD mortality.
Conclusion Long-term LTPA was associated with lower risks of all-cause and CVD mortality. The magnitude of risk reductions was larger when modelling repeated measures of LTPA compared with one measure of LTPA at baseline.
- Epidemiology
- Cohort Studies
- Physical activity
- Death
- Cardiovascular Diseases
Data availability statement
Data may be obtained from a third party and are not publicly available. The data of this study can be requested from the MJ Health Research Foundation (http://www.mjhrf.org).
Statistics from Altmetric.com
Introduction
The 2020 WHO guidelines on physical activity (PA) and sedentary behaviour have provided new evidence-based public health recommendations on the amounts of PA required for health benefits.1–3 These recommendations are based on the most up-to-date evidence linking PA with health outcomes, such as mortality, cardiovascular disease (CVD), diabetes and some cancers. However, the certainty of evidence supporting these recommendations was rated as moderate by WHO Guidelines Review Committee because there is still a possibility that the ‘true effect’ (ie, stronger) is substantially different from the estimated effect, and it has been recommended that further research on the precise dose–response associations between PA and health outcomes is needed.1 2 4
Numerous studies have observed an association between higher levels of PA with lower risk of mortality.1–10 However, much of the available evidence is based on a single baseline measure of PA in middle-aged and older populations, assuming stability of PA during the follow-up. Considering that PA changes across life stages11 and that changes in PA affect survival,12–15 repeated measures of PA in younger populations taking into account changes over time may provide more robust effect estimates as well as more precise dose–response association between PA and risk of death.16 17 Also, there is a paucity of data on the association between PA taken into account and its changes over time with risk of death in non-Caucasian populations, possibly limiting the generalisability of the current evidence.
Hence, the purpose of this study was to investigate the dose–response associations of long-term leisure-time PA obtained from up to 20 repeated measures with all-cause and CVD mortality in a cohort of more than 210 000 Taiwanese adults aged 18 years and older.
Methods
Study design and participants
This study analysed data from the Taiwan Mei Jau (MJ) cohort.18 Details of the MJ cohort and data collection procedures have been reported elsewhere.18 19 In brief, this is an open and dynamic cohort established by the MJ Health Management Institution, a private fee-for-service company, which offers comprehensive health medical screenings in Taiwan. The cost of each health examination was covered by the participants themselves or their employer’s health insurance. Standardised protocols included physical examinations, laboratory tests and self-administered questionnaires applied to all individuals, following the ISO 9001 for quality management. The cohort was established as part of the health screening data management process, handled by the MJ Health Research Foundation (www.mjhrf.org), and has enrolled more than 600 000 participants all over Taiwan since 1994. Although study participants are mostly young and middle-aged adults, some of their offspring, parents or relatives in other age strata have also been included in the cohort. In addition, all participants were encouraged to repeat the medical examinations every year or when they considered it necessary, and around one in two participants had additional medical examinations.
For the present analysis, we included 210 327 apparently healthy men and women (online supplemental figure S1), aged 18 years and above, with at least two data points on leisure-time PA between 1997 and 2016. All participants were free of cancer or CVD at the first medical examination, with a follow-up of at least 1 year between the last medical examination and the time when their vital status ascertained. Participants in the cohort signed a consent form authorising MJ Health Management Institution to manage the data generated from their medical examinations. The present study was approved by the institutional review boards of the MJ Health Management Institution, the National Health Research Institutes and the National Cheng Kung University.
Supplemental material
Dose–response associations between long-term leisure-time physical activity with total and cardiovascular mortality in 210 327 adults. Long-term leisure-time physical activity was calculated as the cumulative average with between 2 and 20 repeated measures from 1997 to 2016 and truncated to 40.00 metabolic equivalent of task (MET) hours/week. A total of 10 539 participants died, 1919 of cardiovascular disease. Analyses were adjusted for age (time scale), sex, educational level, marital status, smoking, alcohol consumption, body mass index groups, occupational physical activity, regular meal patterns, cancer, cardiovascular diseases, diabetes, hypertension, dyslipidaemia and ECG.
Leisure-time PA assessment
The time spent in leisure-time PA was assessed using the MJ PA Questionnaire in 1997 and subsequent medical examinations.11 20 Participants were asked to report the intensity, frequency and duration of leisure-time PA during the last month, with several examples of activity types given in four intensity categories: light moderate (eg, slow walking), moderate (eg, brisk walking), low vigorous (eg, jogging), or high vigorous (eg, running); a metabolic equivalent (MET; 3.5 mL/kg/min) value of 2.5, 4.5, 6.5 and 8.5 was assigned to each intensity, respectively. To calculate the total volume of PA (MET hours/week), the MET value for the reported intensity was multiplied by the frequency and duration. For participants who indicated activities in more than one intensity category, an averaged MET value was assigned. Since the PA questionnaire has been slightly modified over the years (online supplemental table S1), all data were harmonised before pooling. To best represent long-term leisure-time PA, we calculated the cumulative average (ie, average of repeated measures) with all available assessments (online supplemental table S2), which reduces random measurement error and minimise within-person variations over time.17 Long-term leisure-time PA obtained from the cumulative average was truncated at 40.00 MET hours/week (99th percentile in the first examination) to minimise the influence of outliers. Study participants were also categorised according to WHO PA guidelines as follows1 2; (1) ‘none’ (0 MET hours/week), (2) ‘insufficient’ (0.01–7.49 MET hours/week), (3) ‘recommended’ (7.50–15.00 MET hours/week, equivalent to 150–300 min/week in moderate intensity or 75–150 min/week in vigorous intensity) and (4) ‘additional’ (>15.00 MET hours/week, equivalent to >300 min/week in moderate intensity or >150 min/week in vigorous intensity).
Covariates
Information on covariates were assessed in each medical examination. Participants reported their age, educational attainment and marital status during the interview. Smoking, alcohol intake and occupational PA based on standard questions were also self-reported. An optimal regular meal pattern21 was defined as an affirmative answer to the following question: ‘Do you eat on time and in regular amounts?’; this question was shown to predict lower risk of biological cardiometabolic risk factors such as obesity, hypercholesterolemia, atherogenic dyslipidaemia, metabolic syndrome and type 2 diabetes.20 During the clinical examination, body weight and height were measured without shoes and wearing light indoor clothing, and body mass index was calculated as weight in kg divided by squared height in m. Hypertension was identified from the medical history or as systolic/diastolic blood pressure≥140/≥90 mm Hg.22 Dyslipidaemia was defined according to the National Cholesterol Education Program Adult Treatment Panel III criteria as total cholesterol≥240 mg/dL, low-density lipoprotein cholesterol≥160 mg/dL, high-density lipoprotein cholesterol<40 mg/dL or use of lipid-lowering drugs.23 Diabetes was defined as diagnosed diabetes or fasting blood glucose≥126 mg/dL. A standard 12-lead 5 min resting ECG test was performed to detect abnormal heart rate (eg, arrhythmia).
Outcomes
Deaths were identified through follow-up linkage to Taiwan’s National Death File registry until 31 May 2020. We used the International Classification of Diseases, 9th Revision (ICD-9), and International Classification of Diseases, 10th Revision (ICD-10) codes to classify the underlying causes of death. The primary health outcomes for the present study were all-cause and CVD mortality (ICD-9, 390–405, 410–414 and 420–440; ICD-10, 101–102, 105–115, 120–125, 127 and 130–170), since these are the most critical mortality outcomes prioritised by the WHO PA guidelines advisory group.1 4
Statistical analysis
We calculated the follow-up person-years from the date of the first examination and the date of death or the end of follow-up (31 May 2020), whichever came first. A time-varying covariate Cox proportional hazards regression model, with age as the underlying time scale, was used to calculate adjusted HRs and 95% CIs for long-term leisure-time PA and each study outcome. To examine the dose–response relationship between long-term leisure-time PA and all-cause and CVD mortality, we modelled curves from restricted cubic spline Cox regressions, with four knots (5th, 35th, 65th and 95th) distributed across the range of MET hours/week values. HRs and 95% CIs for all-cause and CVD mortality associated with long-term leisure-time PA categories (with none as the reference category) were calculated for the total sample and by subgroups according to age, gender, obesity (≥25 kg/m2 for this Asian population) and cardiometabolic chronic conditions (hypertension, dyslipidaemia and diabetes); the likelihood ratio test was used to compare the model with and without the interaction term. To examine the extent to which the dose–response associations differed using the baseline leisure-time PA measurement compared with the repeated measures, long-term modelling, we also calculated the HRs and 95%CIs of leisure-time PA at baseline with all-cause and CVD mortality.
All analyses were adjusted for the following covariates at baseline and updated at each medical examination; sex (male, female), educational attainment (middle school or below, high school, junior school, college or above), marital status (married or living with partner, single or widowed), smoking (never, former, current), alcohol consumption (never, former, current), regular meal patterns (suboptimal, optimal), occupational PA (sedentary, sedentary with occasional standing/walking, standing or walking, heavy labour), body mass index (<18.5, 18.5–22.9, 23.0–24.9, 25.0–26.9, 27.0–29.9, >30.0 kg/m2), cancer (yes, no), CVD (yes, no), hypertension (yes, no), dyslipidaemia (yes, no), diabetes (yes, no) and ECG (abnormal, normal). If necessary, we included an indicator for missing data in the regression models at baseline, although the most missing information for any covariate was 5%. To minimise missing values on covariates during follow-up, missing variables were replaced with the last value carried forward.
To minimise reverse causation and rule out pre-existing subclinical disease, we repeated the analysis after excluding (1) former/current smoker participants, (2) incident physician-diagnosed cancer or CVD reported in subsequent medical examinations and (3) those who died in the first 4 years of follow-up since the last medical examination. Finally, in sensitivity analyses, we examined the main associations by calculating the cumulative average by (1) weighting more the first measure obtained before the outcome (ie, the baseline examination), (2) weighting more the most recent medical examination before the outcome (ie, the last examination) or (3) using the last measure obtained (also called simple update).17 24 We tested the proportional hazards assumption by modelling the interaction of follow-up time with leisure-time PA and observed no significant deviations. Analyses were conducted with the use of STATA software, V.14.1. Statistical tests were two sided with alpha set at 0.05.
Patient and public involvement
No patients were directly involved in designing the research question or in conducting the research. No patients were asked for advice on interpretation or writing up of the results. There were no plans to involve patients or the relevant patient community in the dissemination of study findings currently.
Results
We included 210 327 participants (50.6% female), aged 18–97 years (mean age; 38.6 years; SD, 12.1) (online supplemental table S3), providing 867 968 data points (98.5% from 881 340 repeated measures included in the present analysis) between 1997 and 2016 (median, IQR: 4.8 years, 2.3–9.0) for estimating long-term leisure-time PA (online supplemental table S4). The median of long-term leisure-time PA was 3.8 (IQR, 1.3–9.0) MET hours/week, and the prevalence of participants with none, insufficient, recommended and additional amounts of leisure-time PA was 6.7%, 62.7%, 19.0% and 11.6%, respectively. Overall, participants with more medical examinations had similar characteristics at baseline than participants with more examinations during the follow-up (online supplemental table S5). Table 1 shows the characteristics of the total sample, by long-term leisure-time PA categories.
Study population characteristics at baseline by categories of long-term leisure-time physical activity
During up to 23 years of follow-up (total person-years, 3 655 734), 10 539 participants died, 1919 of CVD. We observed an inverse, non-linear dose–response association between long-term leisure-time PA and all-cause and CVD mortality (figure 1). The shape of the dose–response association curves for all-cause and CVD mortality showed a pronounced risk reduction at about 5–10 MET hours/week, with a risk reduction greater in magnitude for CVD deaths. At higher levels of leisure-time PA, a linear lower risk was observed, being slightly more decrescent for all-cause mortality.
Compared with the referent (ie, no leisure-time PA), insufficient, recommended and additional amounts of leisure-time PA were associated with lower all-cause and CVD mortality risk (figure 2 and online supplemental table S6). The risk reductions were 26%, 36% and 41% for all-cause mortality by increasing categories of leisure-time PA, as compared with no leisure-time PA; the corresponding risk reductions for CVD mortality were 32%, 44% and 44% for insufficient, recommended and additional amounts of leisure-time PA, respectively. Doing additional leisure-time PA appeared associated with a somewhat further reduced risk for all-cause mortality but not for CVD mortality. The risk reductions of doing insufficient amounts of leisure-time PA were substantial, and even doing about half of the recommended amount was associated with a markedly reduced risk compared with doing no leisure-time PA (online supplemental table S7). Without exceptions, the associations between long-term leisure-time PA with all-cause and CVD mortality were consistent (P for interactions>0.1 for all) when stratified by age and sex, and among participants with obesity and cardiometabolic health conditions (table 2).
Associations between long-term leisure-time physical activity categories with total and cardiovascular mortality in 210 327 adults. Long-term leisure-time physical activity was calculated as the cumulative average (see methods section) with between 2 and 20 repeated measures from 1997 to 2016. Insufficient: 0.01–7.49 metabolic equivalent of task (MET) hours/week. Recommended: 7.50–15.00 MET hours/week. Additional:>15.00 MET hours/week. Analyses were adjusted for age (time scale), sex, educational level, marital status, smoking, alcohol consumption, body mass index groups, occupational physical activity, regular meal patterns, diabetes, hypertension, dyslipidaemia and ECG. n for each increasing category is 14 042, 131 838, 40 065 and 24 382. Deaths for all/cardiovascular cases for each increasing category are 930/178, 5148/896, 2617/476 and 1844/369.
Associations between long-term leisure-time physical activity categories with all-cause and cardiovascular mortality by age, gender, obesity and cardiometabolic chronic conditions at baseline
Leisure-time PA at baseline (first examination) showed a similar shape of distribution to long-term data, but with an important difference in frequency density (figure 3). The association between leisure-time PA and risk of death was greater in magnitude (19%–32% and 34%–43% lower risk for any cause and CVD deaths, respectively) when using repeated measures of PA compared with the associations between baseline PA (single point) and risk of death (figure 4 and online supplemental table S8).
Descriptive data on leisure-time physical activity at baseline (first examination, blue) and long-term (cumulative average, red) in the present study. IQR, 25th–75th. Baseline leisure-time physical activity was also truncated to 40.00 metabolic equivalent of task (MET) hours/week (99th percentile of distribution) for this analysis.
Dose–response associations between long-term (red) and baseline (blue) leisure-time physical activity with total and cardiovascular mortality in 210 327 adults. Long-term leisure-time physical activity was calculated as the cumulative average (see methods section) with between 2 and 20 repeated measures from 1997 to 2016 and truncated to 40.00 metabolic equivalent of task (MET) hours/week. A total of 10 539 participants died, 1919 of cardiovascular disease. Values are HRs and 95% CI lines were removed for simplicity purposes. Analyses were adjusted for age (time scale), sex, educational level, marital status, smoking, alcohol consumption, body mass index groups, occupational physical activity, regular meal patterns, cancer, cardiovascular diseases, diabetes, hypertension, dyslipidaemia and ECG.
Sensitivity analyses suggested that our results were attenuated by only a small amount after accounting for reverse causation when excluding smokers, those with underlying diseases, and early deaths within 4 years (online supplemental table S9). Also, we found similar associations using different methods of calculating the cumulative average. However, when modelling a single measure of PA in the last examination, the associations with mortality outcomes were greater in magnitude than those previously observed when modelling PA at baseline as the single measure exposure (online supplemental table s10).
Discussion
Findings into context
In this large Taiwanese cohort of more than 210 000 participants, long-term leisure-time PA calculated from up to 20 repeated measures showed a clear non-linear dose–response association with long-term all-cause and CVD mortality. Participating in leisure-time PA less than the recommended levels (ie, <7.5 MET hours/week) reduced total and CVD mortality risk by around 30% compared with the inactive referent. Recommended amounts of leisure-time PA were associated with reductions in all-cause and CVD mortality risks of 36% and 44%, respectively. Engagement in additional amounts of leisure-time PA than those recommended was also associated with a lower risk of 41% and 44% for deaths from all causes and CVD mortality, respectively, and almost similar in magnitude when compared with the mortality risk reductions observed in those doing the recommended amounts.
Conceivably, to date, the most comprehensive work exploring the dose–response association between PA and mortality was the pooled analysis conducted by Lear et al 7 including 661 137 participants (median age 62 years) from six population-based prospective cohorts, and they found that compared with no leisure-time PA (0 MET hours/week), the HR (95% CI) was 20%, 31%, 37%, 39%, 39% and 31% for amounts of 0.1 to <7.5, 7.5 to <15.0, 15.0 to <22.5, 22.5 to <40.0, 40.0 to <75.0 and ≥75.0 MET hours/week, respectively. Other large cohort studies and pooled analyses have also examined dose–response associations between PA and mortality, but the differences in reference and comparison groups have made comparisons difficult. For example, Liu et al 8 in the Prospective Urban Rural Epidemiologic study examined whether total PA (including leisure-time and other domains) was associated with lower risk of mortality in 130 843 participants (mean age 50 years) from 17 high-income, middle-income and low-income countries, and compared with not meeting the recommended amount of PA, those doing between one and five times the recommended amount, and more than five times the recommended amount, the HR (95% CI) mortality risk reduction was 20% and 35%, respectively. In a pooled analysis of 467 729 East Asian adults (mean age 55 years),9 compared with those who reported no or almost no leisure-time PA, there was a 14%–15% reducing risk of all-cause mortality among those with higher amounts of leisure-time PA. Bauman et al 10 including 416 175 individuals (mean age 42 years) from this Taiwan MJ cohort showed that compared with those doing <3.75 MET hours/week, the HR (95% CI) was 0.86 for 3.75–7.49 MET hours/week, 0.80 for 7.50–16.49 MET hours/week, 0.71 for 16.50–25.49 MET hours/week and 0.65 for ≥25.50 MET hours/week.
All these previous studies, however, examined associations with mortality risk using a single data point of PA assessed at baseline, whereas we used repeated measures of the exposure, which likely provides more robust estimates and consider changes in PA behaviour over time. The Nurses’ Health Study I and II and the Health Professionals Follow-up Study included updated information on leisure-time PA every 2 years, allowing for modelling long-term PA, and some work with these cohorts has shown the associations of long-term PA on mortality outcomes.17 25–29 Among women from the Nurses’ Health Study I, the all-cause relative risks (RR) (95% CI) were 0.82 (0.76 to 0.89), 0.75 (0.69 to 0.81), 0.74 (0.68 to 0.81) and 0.71 (0.61 to 0.82) for amounts of <1 hour/week, 1–1.9 hours/week, 2–3.9 hours/week, 4–6.9 hours/week and ≥7 hours/week in leisure-time PA, respectively; corresponding estimates for cardiovascular mortality were 0.80 (0.68 to 0.96), 0.74 (0.62 to 0.88), 0.62 (0.50 to 0.77) and 0.69 (0.49 to 0.97). Among diabetic men from the Health Professionals Follow-up Study, the RR for all-cause mortality across fifths of leisure-time PA were 1.0, 0.80, 0.57, 0.58 and 0.58. Overall, these studies found strong inverse dose–response associations, but comparisons with the present study should be interpreted with caution owing to differences in cohort designs, and the specific characteristics in these cohorts (eg, selected populations, high socioeconomic status, Caucasian).17 25–29 However, these previous studies and our present work that have included repeated measures suggest that the association between PA and mortality are bigger than most of the evidence has shown to date using on a single self-reported measure of PA.16
Possible explanations of our results
PA promotes acute physiological responses by activating a number of molecular pathways in whole organ systems and thereby reducing the risk of developing chronic diseases such as CVD, cancer and others aging-related diseases that poorer survival outcomes.30–33 PA assessed with repeated measures likely better captures the effect of exercise adaptations on health since we found stronger associations between PA and from repeated measures of PA compared with one single measure assessed at baseline. This might explain the large effect size observed for the association between long-term PA and CVD mortality and other CVD-related mortality outcomes because the benefits of regular PA for the cardiovascular system seems to be more pronounced than for other organ systems.30 31 The stronger association observed compared with studies using a single measure of PA may also be a result of the reduction in random measurement errors, which bias associations towards the null.
Public health and research implications
Our results provide additional support for PA guidelines suggesting that any PA is better than none (ie, ‘every move counts’), and are in agreement with previous observations suggesting a marked risk reduction even if not meeting the current recommendations compared with those doing no PA.1 2 We also found that doing additional amounts of leisure-time PA beyond the recommended levels provides small but significant additional reductions in mortality risks; this supports the recommendation that doing more than 300 min at moderate intensity or 150 min at vigorous intensity has small but additional health benefits (ie, reduces mortality risk), and consequently is considered a conditional recommendation (ie, the balance of benefits to harms is small).
Although randomised controlled trials are the ‘gold standard’ research design to provide the highest levels of certainty in evidence, trials are most feasibly conducted in ideal and often unrealistic conditions with highly motivated and homogenous participants, which provide results with high internal validity but limited external validity. In addition, trials that aim to examine mortality as the primary outcome require inclusion of at least three arms to explore a dose–response and an unrealistic effort to maintain compliance in a supervised weekly training over many years.34 Consequently, large-scale observational studies including repeated measures, such as the present study, are important for confirming current PA guidelines and providing new data to inform future guidelines.
Strengths and limitations
Limitations of this work should be considered. First, we relied on self-reported PA measures. Although repeated measures minimise measurement error and within-individual variability over time, the PA questionnaire was slightly modified over time and some issues that usually reduce the validity of self-report (eg, misclassifications, social desirability, and recall bias, etc) remain. Second, selection bias might be also introduced since according our study design participants had a different number of medical examinations (eg, the healthy participant effect); however, our analyses showed that participants with fewer medical examinations had similar patterns of leisure-time PA over time and comparable sociodemographic characteristics, lifestyle behaviours and health risk factors at baseline as those participants with more medical examinations. These potential reporting and selection biases would likely have led to an underestimation of the observed dose–response effect sizes of PA on mortality outcomes in the present study. Third, although the analyses were adjusted for key confounders, the potential for residual confounding remains due to unmeasured confounders and the very low-resolution measurement of some confounders (eg, diet and alcohol). Fourth, this was an observational study that neither can demonstrate causality, nor rule out effects of unmeasured or residual confounding; however, our sensitivity analyses confirmed our main findings.
On the other hand, we only examined leisure-time PA, therefore, we could not evaluate associations with total PA and/or other domain-specific PA (ie, transportation, work, household). Also, information on sedentary behaviours was not collected in this cohort and therefore potential moderating effects on these results could not be investigated.5 35 Future research using repeated objective measures, such as accelerometers, might be critical to providing evidence on dose–response association of PA on health outcomes.36 Finally, despite that the MJ cohort may be the most representative of the general population of Taiwan to date, the final analytic sample may not be representative since our sample comprised of participants with additional medical examinations, who tend to be from higher socioeconomic background.
Conclusions
Based on a large Taiwanese cohort of relatively young adults, we found non-linear, inverse dose–response associations between repeated measures of leisure-time PA and all-cause and CVD mortality, and the magnitude of risk reductions, mainly for CVD deaths, were larger than when considering only one measure of PA at baseline. Doing some PA, even less than the currently recommended levels, was associated with substantially lower mortality risk compared with the inactive referent. Doing the recommended and additional amounts of leisure-time PA over a long term was associated with even further risk reductions. Hence, these findings provide more certain evidence for most of the recommendations and good practice statements included in the recent WHO PA guidelines.
Key messages
What is already known on this topic?
Evidence on the association between leisure-time physical activity (LTPA) and mortality is mainly limited by relying on a single measure of that assumes the stability of this behaviour during follow-up.
What this study adds?
We observed an inverse, non-linear dose–response association between long-term LTPA obtained from repeated measures from at least two medical examinations for up to 20 years (median, IQR: 4.8 years, 2.3–9.0) and all-cause and cardiovascular mortality.
Achieving the recommended amount or even less than recommended amount of LTPA over a long term lowered the risk of all-cause and cardiovascular mortality, whereas doing more activity beyond the recommendations was associated with a slight further reduction in all-cause mortality risk.
The association between LTPA and risk of death was greater in magnitude when using repeated measures of LTPA compared with the associations between a single baseline measure of LTPA and risk of death.
How this might influence research, policy or practice?
These results provide support and higher certainty of evidence for most of the recommendations and good practice statements included in the recently updated WHO guidelines and to inform future guidelines.
Data availability statement
Data may be obtained from a third party and are not publicly available. The data of this study can be requested from the MJ Health Research Foundation (http://www.mjhrf.org).
Ethics statements
Patient consent for publication
Ethics approval
This study involves human participants and was approved by the Institutional Review Boards of the Mei Jau Health Management Institution, the National Health Research Institutes and at National Cheng Kung University (A-ER-108-081). Participants gave informed consent to participate in the study before taking part.
Acknowledgments
The authors thank Mei Jau (MJ) Health Research Foundation for the authorisation of using MJ health data.
References
Supplementary materials
Supplementary Data
This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
Supplementary Data
This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
Supplementary Data
This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
Footnotes
DD, IL and UE are joint senior authors.
Twitter @DrMelodyDing
Correction notice This article has been corrected since it published Online First. The abstract has been corrected and affiliations updated.
Contributors DMG conceived and designed the study. DMG, FR-A and TY acquired the data. DMG and VC-S analysed and interpreted the data. DMG and FR-A obtained the funding. DMG drafted the first version of this article. All authors critically revised and edited the article. DD, I-ML and UE supervised this work. DMG was the guarantor.
Funding This work was supported by FIS grants (State Secretary of R+D+I and FEDER/FSE; 16/609, 16/1512, 18/287, 19/319, 20/00657). DMG was supported by a ‘Ramon y Cajal’ contract (RYC2016-20546) and DD by a Heart Foundation Australia Future Leader Fellowship (no. 101234) while contributing to this work. A preliminary version of this work was awarded the 2020 National Sports Medicine Award (University of Oviedo, Oviedo, Spain).
Disclaimer The funders had no role in study design, data collection, data analysis, data interpretation or writing of the report. Any interpretation or conclusion related to this article does not represent the views of MJ Health Research Foundation.
Competing interests None declared.
Patient and public involvement Patients and/or the public were not involved in the design, or conduct, or reporting, or dissemination plans of this research.
Provenance and peer review Not commissioned; externally peer reviewed.
Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.