Article Text

Download PDFPDF

Physical activity and mortality: what is the dose response and how big is the effect?
  1. Ulf Ekelund1,2,
  2. Knut Eirik Dalene1,
  3. Jakob Tarp1,
  4. I-Min Lee3,4
  1. 1 Department of Sport Medicine, Norwegian School of Sport Sciences, Oslo, Norway
  2. 2 Department of Chronic Diseases and Ageing, Norwegian Institute of Public Health, Oslo, Norway
  3. 3 Department of Medicine, Brigham and Women’s Hospital and Harvard Medical School, Boston, Massachusetts, USA
  4. 4 4Department of Epidemiology, Harvard TH Chan School of Public Health, Boston, Massachusetts, USA
  1. Correspondence to Professor Ulf Ekelund, Department of Sport Medicine, Norwegian School of Sport Sciences, Oslo 0806, Norway; Ulf.Ekelund{at}

Statistics from

More than 60 years of research have consistently shown that physical activity beneficially affects several health outcomes and reduces the risk of premature mortality. The Harvard Alumni Study, one of the first studies to suggest a dose–response association between physical activity and health, examined the association between ‘total’ physical activity (defined as strenuous sports, walking and stair climbing) and participation in strenuous sports and the risk of heart attack in about 17 000 men.1 ‘Total’ physical activity was categorised into groups based on energy expenditure in physical activity per week. The following were the important observations: (1) there was a substantial risk reduction when comparing the ‘inactive’ reference group expending <500 kcal per week in ‘total’ physical activity with the second group expending 500–999 kcal per week; (2) there was graded dose–response association, with declining risk up to about 3000 kcal per week of ‘total’ physical activity, beyond which the benefits plateaued; and (3) energy expenditure from strenuous sports appeared to have greater protective effect than ‘total’ volume of activity. However, the latter observation may partly be explained by more accurate reporting of sporting activities compared with physical activities of daily living of lower intensity (eg, stair climbing and walking). The dose–response curve from the landmark paper by Paffenbarger et al 1 was subsequently confirmed after accounting for confounding,2 and it has been replicated in women and ethnically diverse populations from low-income and middle-income countries.3–5

These previous studies assessed physical activity using different self-report instruments, which may be prone to cognitive biases (eg, recall bias) and which may result in overestimation of physical activity and underestimation of sedentary time. Due to such measurement error associated with self-report physical activity instruments, the magnitude of associations with health outcomes may be underestimated. Also, self-report instruments rarely provide data on light-intensity physical activity (LPA), which has not been reported well.

Our 2019 study in the BMJ

We therefore conducted a study aimed at examining the dose–response associations between total physical activity, sedentary time and intensities of physical activity assessed by accelerometry.6 Following a systematic review, we identified and included eight studies of more than 36 000 middle-aged and older men and women followed for almost 6 years on average, during which more than 2100 died. Each included study harmonised their individual-level data according to a standardised protocol, and data were thereafter meta-analysed.

Our results suggest that total physical activity and all intensities of physical activity were associated with lower risk of death, and sedentary time associated with higher risk of death, in a non-linear dose–response manner. The risk of death was about 50% lower comparing the second quarter with the least active referent group for total physical activity. This suggests that all physical activities regardless of intensity are beneficial in reducing the risk of premature death. Similarly, risk reduction ranged from 35% to 45% when comparing the second quarter for LPA, ‘low’ LPA, ‘high’ LPA and moderate-to-vigorous-intensity physical activity (MVPA) with the referent, suggesting that even small increases in physical activity of any intensity substantially reduce the risk of premature death. Public health efforts may therefore emphasise the importance of increasing physical activity levels in those who are least active and in whom the benefits appear the greatest.

The maximal risk reduction for all-cause mortality was about 60% for MVPA, observed at about 24 min of MVPA per day. Similarly, LPA, ‘low’ light-intensity and ‘high’ light-intensity physical activities were strongly and inversely related to the risk of death. Therefore, longevity may be promoted by undertaking physical activities across the full intensity spectrum, including unstructured activities of light intensity. For example, a similar risk reduction of about 60% can be achieved when performing about 80 min of ‘high’ LPA compared with 24 min per day of MVPA, whereas many hours of ‘low’ LPA appear necessary to achieve this risk reduction. Future physical activity guideline development may want to consider this flexibility.

Risk reduction was substantially greater in magnitude when using accelerometer compared with self-report,3–5 and we observed a maximal risk reduction at much lower levels of MVPA. Figure 1 schematically shows the difference in risk reduction for a given amount of MVPA (minutes per week) between three influential studies assessing physical activity by self-report3–5 and accelerometry-measured physical activity.6 The shape of the dose–response curves is similar, whereas risk reduction for a similar amount of MVPA is about three times greater in magnitude when physical activity is assessed by accelerometry. This is likely explained by (1) the different precisions of measurement, and (2) that accelerometry captures all activities during the time the device is worn, while self-report studies assessed recreational or leisure-time MVPA.3–5

Figure 1

Schematic comparison of the prospective associations between self-reported (dashed lines) and accelerometer-assessed (solid line) moderate-to-vigorous-intensity physical activity (MVPA) and all-cause mortality (multivariable adjusted models). Self-reported physical activity (PA): (A) Lear et al 4: n=130 843, mean follow-up 6.9 years, baseline age 35–70 years, 5334 deaths; (B) Wen et al 3: n=416 175, mean follow-up 8.1 years, baseline age 20+ years, 12 118 deaths; (C) Arem et al 5 (meta-analysis): n=661 137, mean follow-up 14.2 years, baseline age 21–98 years, 116 686 deaths. Accelerometer-assessed physical activity: (D) Ekelund et al 6 (meta-analysis): n=36 383, median follow-up 5.8 years, baseline age 40+ years, 2149 deaths. *The reference groups across the self-report studies differ and are defined as follows: ‘no recreational physical activity’ (ie, walking and MVPA)4; ‘inactive’ leisure-time physical activity (ie, <3.75 MET-hours/week of walking, moderate-intensity and vigorous-intensity activity3); and ‘no’ leisure-time MVPA.5

Current physical activity guidelines are based on an absolute definition of MVPA (ie, >3 Metabolic Equivalent of Task (METs)). Our findings and those of others7 suggest that LPA (<3 METs) reduces the risk of death in middle-aged and older populations. However, physical activity defined as LPA of 2–3 METs may be equivalent to MVPA (in relative terms) in elderly and in those with low cardiorespiratory fitness. Thus, more research is needed on the importance of relative intensity versus absolute intensity in reducing the risk of death and improving health.

What does the future hold?

Our study6 points towards both challenges and opportunities. Here we highlight six of the most relevant ‘take-home’ messages for researchers, clinicians and public health workers.

  1. Reduce sedentary time and subsequently increase total physical activity in small doses. This is likely achievable for the majority of people. It may be a more appropriate goal for those currently physically inactive than aiming immediately for the current guidelines of 150 min of MVPA per week.

  2. Public health guidelines can expand to also include physical activity of light intensity. Different intensities can be combined in numerous ways to achieve health benefits.

  3. Device-based measures of physical activity in large population-based samples are already substantial and will continue to progress. Data from these devices will greatly extend our knowledge on the dose–response association between physical activity and health outcomes.

  4. Longitudinal studies that assess physical activity and sedentary time repeatedly in younger populations are needed to understand how changes in physical activity may affect health in middle and older ages.

  5. A research priority should be to examine the dose–response associations using measurement methods free from cognitive biases in ethnically diverse population subgroups, including very old adults, people of different bodyweight and shape, and people with existing chronic conditions. Such studies will determine if tailored ‘subgroup-specific’ physical activity guidelines are necessary or useful.

  6. How device-measured physical activity and sedentary time combine to influence health is unknown. How different sedentary postures such as sitting or standing affect health is also unknown. This knowledge is important to inform tailored interventions as sedentary behaviours are highly prevalent in many societies.

While researchers are busy tackling these research priorities, it is wise for practitioners and members of the public to adhere to a message that is as simple as ‘sit less, move more and more often’.


We thank the reviewers for their valuable comments.



  • Twitter @Ulf_Ekelund, @knuteirik47

  • Contributors UE drafted the revised version of the editorial. All authors approved the final version. JT and KED produced the figure.

  • Funding The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors.

  • Competing interests None declared.

  • Patient consent for publication Not required.

  • Provenance and peer review Commissioned; externally peer reviewed.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.