How sure are we that physical activity makes us live longer?

The Physical Activity and Population Health BJSM Blog Series

By Sonia Cheng @soniawmcheng

The health benefits of regular exercise and a physically active lifestyle are indisputable – it plays a key role in improving cardiorespiratory fitness, maintaining physical function, and correcting biological risk factors such as hypertension and high cholesterol. But whether exercise also lengthens the lifespan is a far more difficult question to answer.

This issue of BJSM features some spirited discussion about whether exercise really is the secret to a long life in a review article1 and two editorials2,3.

We share some highlights from this discussion: The conflicting evidence between observational studies and randomised controlled trials about whether there is a causal association between higher physical activity levels and lower mortality risk; whether limitations in study design may have led to spurious conclusions; and how future research can better answer this question.

How certain are we about the link between exercise and longevity?

Observational studies have consistently shown a strong dose-response association between higher physical activity levels and lower mortality risk, across many different populations and after adjusting for known confounders. Based on epidemiological estimates, solving the global problem of physical inactivity may increase life expectancy by a mere 0.68 years, which is a magnitude of effect similar to the elimination of smoking or obesity4.

However, a causal relationship between mortality and physical activity in adulthood has yet to be confirmed in randomised controlled trials of initially healthy individuals5,6, nor in animals. Without definitive data from interventional studies, the gold standard for causality inference, can we really guide public health recommendations and policy with certainty?

What is the best available evidence?  

In his review article, Kujala highlights the discrepant findings between observational and interventional studies about the effects of physical activity on longevity.

The Prospective Urban Rural Epidemiologic (PURE) study cannot assert a causal relationship between physical activity and mortality, but the findings do support higher levels of physical activity for all individuals, irrespective of age, sex, or country of origin. Compared to those who did not meet the physical activity guidelines, lower mortality risk was observed in participants who reported moderate (HR 0.80, 95% CI 0.74 to 0.87) and high (HR 0.65, 95% CI 0.60 to 0.70) levels of physical activity (600 to <3000 MET-minutes/week and ≥3000 MET-minutes/week respectively).

In their editorial, Shiroma and Lee point out that although we have not seen a proven causal relationship between physical activity and longevity in randomised controlled trials such as LIFE and Look AHEAD, these null findings should be interpreted with caution. Neither trial was designed with mortality as a primary outcome, and therefore may not have been powerful enough to demonstrate effects on longevity.

Shiroma and Lee add that Look AHEAD did not exclusively target physical activity – the trial involved a combined physical activity and dietary intervention to achieve and maintain weight loss, and included the use of medications for diabetes and high cholesterol. It is possible that this medical management of cardiovascular risk factors reduced the observed effect of the intervention, and may explain the non-significant difference in cardiovascular disease between the intervention and control groups.

Is the evidence open to misinterpretation?

This question is much easier to answer – it’s a resounding yes. In their editorial, Wade and colleagues draw attention to the limitations of observational study designs which make it difficult to identify causal mechanisms between physical activity and mortality.

  1. The difficulty of taking into account all factors that may be associated with both physical activity and longevity.

The authors use Jerry Morris’ London bus driver/conductor study as a classic example. Bus conductors were found to have a lower risk of coronary heart disease compared to the drivers, suggesting that occupational physical activity may reduce coronary heart disease in middle-aged men. A follow-up study later showed that more obese men were preferentially becoming drivers at recruitment. Differences in baseline adiposity in the driver/conductor populations may have contributed to the difference found in coronary heart disease risk, regardless of physical activity levels.

To further complicate things, variables such as body mass index (BMI), hypertension, and cardiorespiratory fitness are usually treated as potential confounders in the association between physical activity and mortality and are included in the statistical model. Shiroma and Lee point out that lower BMI, lower rates of hypertension, and higher fitness levels are correlated with both physical activity (the “exposure”) and mortality (the “outcome”). The authors argue that these variables may be mediators in the causal pathway, rather than true confounders. Controlling for these variables may then reduce the strength of the association between physical activity and longevity.

  1. Reverse causation.

Wade and colleagues highlight how reverse causation – the possibility that “the outcome is responsible for variation in the exposure, rather than the direction of interrogation” – may also lead to misinterpretation of the evidence. Physical activity decreases the risk of being overweight/obese, for example, but it is equally plausible that being overweight/obese limits the ability to engage in physical activity in the first place.

Studies of older adults and those with multi-morbidity are at particular risk of reverse causation. Individuals who are well enough to exercise due to absence of chronic disease will seemingly have a lower risk of mortality compared to their more inactive peers – Kujala describes this as the “healthy exerciser bias”. It might appear that older people can lengthen their lives by increasing their physical activity levels, even though the evidence from interventional studies is lacking. Since older and chronic disease populations are at greater risk of injury if exercise is not carefully prescribed, it is important that causal links between physical activity and longevity are not misinterpreted.

  1. Selection bias influences participation rates in observational studies.

A similar issue is seen when comparing mortality risk between sedentary and physically demanding occupations, suggests Wade and colleagues. Good physical function and cardiorespiratory fitness is required to undertake physically demanding work, and thus there is a selection bias towards lower mortality risk amongst those employed in manual work. This can distort the observed associations between occupational physical activity and longevity.

  1. The gold standard for causal inference may not be as golden as we thought.

In their editorial Shiroma and Lee warn against ignoring all available data except those from randomised controlled trials, which would be “a disservice to the rich totality of evidence available”. The authors argue that the results of randomised controlled trials are only trustworthy if they are well-designed and conducted, and they too are vulnerable to limitations.

Selection bias can be introduced into interventional studies because of dropouts, and because those who volunteer for the study may have different characteristics to those from the population who were not selected or who were unable/unwilling to participate (due to poor health or lack of interest, for example).

A well-conducted trial must have high participant compliance with the physical activity intervention, and this is associated with huge effort and cost. It is also difficult to investigate specific doses of physical activity, since there are too many variations in type, frequency and duration of activity to feasibly test in a free-living population.

How can future research tell us if physical activity lengthens the lifespan?

We may lack conclusive data from interventional studies to determine whether physical activity really does add years to life, but the three BJSM articles provide great insight and direction for future researchers who are looking to find an answer

Kujala’s review article suggests that continuing to improve the design of observational and interventional studies, and further investigating the role of genetic factors in the ageing process7, may help to clarify if there is a causal link between physical activity and longevity.

The jury may still be out on whether or not physical activity lengthens the lifespan, but it is incontrovertible that regular exercise and a physically active lifestyle gives you quality of life – better health, function and independence into old age.


Sonia Cheng graduated from The University of Sydney with a Bachelor of Applied Sciences (Physiotherapy) (Honours Class I) in 2014. Sonia is currently employed as a physiotherapist with Royal Prince Alfred Hospital and Westmead Hospital in Sydney. 

We invite you to share and support the Physical Activity and Population Health BJSM Blog Series. Join the conversation on ‘how change happens’ at #PAblogBJSM and #brightspotsBJSM. We welcome guest blogger contributions. If you have any ideas please email


  1. Kujala UM. Is physical activity a cause of longevity? It is not as straightforward as some would believe. A critical analysis. Br J Sports Med. Published Online First: 15 March 2018. doi: 10.1136/bjsports-2017-098639.
  2. Wade KH, Richmond RC, Davey Smith G Physical activity and longevity: how to move closer to causal inference. Br J Sports Med. Published Online First: 15 March 2018. doi: 10.1136/bjsports-2017-098995.
  3. Shiroma EJ, Lee I Can we proceed with physical activity recommendations if (almost) no clinical trial data exist on mortality? Br J Sports Med. Published Online First: 15 March 2018. doi: 10.1136/bjsports-2018-099185.
  4. Lee IM, Shiroma EJ, Lobelo F, Puska P, Blair SN, Kazmarzyk PT. Effect of physical inactivity on major non-communicable diseases worldwide: an analysis of burden of disease and life expectancy. 2012;380:219-29.
  5. Kujala UM. Physical activity, genes, and lifetime predisposition to chronic diseases. Eur Rev Aging Phys Act. 2011;8:31-6.
  6. Karvinen S, Waller K, Silvennoinen M, et al. Physical activity in adulthood: genes and mortality. Scientific Reports. 2015;5:18259.
  7. Latvala A, Ollikainen M. Mendelian randomization in (epi)genetic epidemiology: an effective tool to be handled with care. Genome Biology. 2016;17:156.

(Visited 20,865 times, 1 visits today)