Exercise and longevity

Studies of exercise and lifespan usually report a hazard ratio (the death rate in one group relative to another; 0.80 means 20% lower). In the pooled National Cancer Institute Cohort Consortium analysis, compared with adults reporting no leisure-time activity, the hazard ratio was 0.80 (95% CI 0.78 to 0.82) below the guideline minimum, 0.69 (0.67 to 0.70) at one to two times the minimum, 0.63 (0.62 to 0.65) at two to three times, and 0.61 (0.59 to 0.62) at three to five times [1]. Most of the benefit therefore came with the first increments of activity; moving from the guideline minimum to three to five times the minimum added a further eight percentage points. Adults doing ten or more times the minimum had a hazard ratio of 0.69 (0.59 to 0.78), with no sign of harm at very high amounts [1].

The same pattern appears when activity is converted to years of life. Among 654,827 adults in the same consortium (82,465 deaths, median follow-up 10 years), up to 75 minutes of brisk walking a week was linked to 1.8 extra years of life expectancy after age 40 (95% CI 1.6 to 2.0), and 450 minutes or more to 4.5 extra years (4.3 to 4.7) [2]. Being active and of normal weight was linked to 7.2 more years than being inactive with a body mass index of 35 or more [2]. In a Taiwanese study of 416,175 adults followed for an average of 8 years, about 15 minutes a day (92 minutes a week) of exercise was linked to 14% lower mortality and three years longer life expectancy than inactivity, and each further 15 minutes a day to a further 4% reduction [9].

Studies that measure activity with body-worn accelerometers rather than questionnaires show the same shape with larger differences. In a pooled analysis of eight such studies (36,383 adults, median follow-up 5.8 years), the most active quarter of participants had a hazard ratio of 0.27 (0.23 to 0.32) compared with the least active quarter, and for light-intensity activity the top quarter had a hazard ratio of 0.38 [10]. In 15 studies of daily step counts (47,471 adults, median follow-up 7.1 years), the top quarter (median 10,901 steps a day) had a hazard ratio of 0.47 (0.39 to 0.57) compared with the bottom quarter (3,553 steps); the benefit levelled off at about 6,000 to 8,000 steps a day in adults aged 60 and over and at 8,000 to 10,000 in younger adults [11].

Exercise research usually separates aerobic (cardio) activity, muscle-strengthening activity, balance and mobility work, and everyday movement that is done for reasons other than training. Each has its own body of evidence, and the evidence differs in kind: mortality studies for aerobic and strength activity, falls trials for balance work, and device-measured activity studies for everyday movement.

Aerobic fitness has the strongest link to mortality. It is usually measured as VO2 max (the most oxygen the body can use during all-out exercise) or as the peak workload on a treadmill test in METs (multiples of the energy used sitting at rest). Across 33 studies, each 1-MET increase in fitness was linked to 13% lower all-cause mortality (risk ratio 0.87, 95% CI 0.84 to 0.90), and people with low fitness (under 7.9 METs) had 70% higher mortality than people with high fitness [3]. In 122,007 patients referred for treadmill testing, the fittest group had a hazard ratio of 0.20 compared with the least fit, with no upper limit of benefit observed [14]. Among 750,302 US veterans, the lowest mortality was seen at about 14 METs in both men and women, and the least fit had about four times the mortality of the most fit [12]. The VO2 max, Zone 2 training, high-intensity interval training and walking and steps articles cover aerobic fitness in detail.

Muscle-strengthening activity is linked to lower mortality independently of aerobic activity. A pooled analysis of 16 studies found 10 to 17% lower risk of death, cardiovascular disease, cancer and diabetes, with a J-shaped curve for mortality: the largest reduction at about 30 to 60 minutes a week, and an unclear effect at higher amounts [4]. People who combined muscle-strengthening and aerobic activity had lower all-cause, cardiovascular and cancer mortality than people who did neither [4]. See resistance training, muscle mass, grip strength and sarcopenia.

Balance and functional exercise is studied mainly for falls rather than for mortality. A Cochrane review of 108 randomised trials (23,407 people aged 60 and over, average age 76) found that exercise reduced the rate of falls by 23%, balance and functional exercises by 24%, programmes combining several types of exercise by 34%, and tai chi by 19% [15]. See balance training, flexibility and mobility, functional fitness, fall prevention and yoga and tai chi.

The total weekly amount of activity appears to count for more than the number of sessions. In 11 rounds of the Health Survey for England and the Scottish Health Survey (63,591 adults aged 40 and over, 8,802 deaths over 561,159 person-years), compared with inactive adults, the hazard ratio for all-cause mortality was 0.70 (95% CI 0.60 to 0.82) in "weekend warriors" who met the guideline in one or two sessions a week and 0.65 (0.58 to 0.73) in regularly active adults who spread it over three or more sessions [7]. The confidence intervals overlapped.

Even adults below the guideline who exercised once or twice a week had a hazard ratio of 0.66 (0.62 to 0.72) compared with inactive adults [7]. Cardiovascular mortality followed the same pattern, at about 0.60 in all three active groups. Activity in these surveys was self-reported.

Long daily sitting is linked to higher mortality, but the size of the link depends on how active a person is otherwise. A harmonised analysis of 13 studies with data on 1,005,791 adults found mortality rates 12 to 59% higher in the two least active quarters than in the reference group, which sat under 4 hours a day and was in the most active quarter [13]. Adults who sat more than 8 hours a day and were in the least active quarter had a hazard ratio of 1.59 (95% CI 1.52 to 1.66).

In the most active quarter, which corresponded to about 60 to 75 minutes of moderate activity a day, sitting more than 8 hours a day was not linked to higher mortality (hazard ratio 1.04, 0.99 to 1.10) [13]. Television viewing behaved differently: watching 3 or more hours a day was linked to higher mortality at all activity levels except the most active quarter, where the risk rose only at 5 or more hours (1.16, 1.05 to 1.28). Adults who sat the least but were also the least active still had a hazard ratio of 1.27 (1.22 to 1.31) [13].

Device-measured studies found a similar gradient: the most sedentary quarter of participants had a hazard ratio of 2.63 (1.94 to 3.56) compared with the least sedentary [10]. The 2020 World Health Organization guidelines recommend reducing sitting time at all ages but state that the evidence was insufficient to set a specific limit [5]. See sedentary time for detail.

Lower mortality is also seen in people who become active in middle age. The NIH-AARP Diet and Health Study asked 315,059 adults aged 50 to 71 to recall their leisure-time activity at ages 15 to 18, 19 to 29, 35 to 39 and 40 to 61, and followed them to 2011, during which 71,377 died [8]. Compared with adults who were inactive throughout adulthood, those who maintained high activity had a hazard ratio of 0.64 (95% CI 0.60 to 0.68), and those who were inactive for most of adult life but increased activity at 40 to 61 had a hazard ratio of 0.65 (0.62 to 0.68). Recalled activity decades later is imprecise, which limits these estimates.

Fitness remains linked to survival in older age groups. Among US veterans, the graded link between treadmill fitness and mortality held in people in their seventies and eighties [12], and in the treadmill-test study of 122,007 patients the benefit of the highest fitness level over a high level was present in patients aged 70 and over (hazard ratio 0.71, 0.52 to 0.98) [14]. Exercise programmes for adults aged 60 and over reduce falls, a major cause of injury and loss of independence at that age [15].

Randomised controlled trials show that exercise training changes several established cardiovascular and metabolic risk factors. A network meta-analysis of 270 trials (15,827 participants) found reductions in resting systolic and diastolic blood pressure of about 4.5/2.5 mm Hg with aerobic training, 4.6/3.0 mm Hg with dynamic resistance training, 6.0/2.5 mm Hg with combined training and 8.2/4.0 mm Hg with isometric training [16]. In the Diabetes Prevention Program, a lifestyle programme with goals of at least 150 minutes of physical activity a week and 7% weight loss reduced the incidence of type 2 diabetes by 58% (95% CI 48 to 66%) over an average of 2.8 years in 3,234 adults with raised blood glucose; the effect cannot be attributed to activity alone because diet and weight loss were part of the programme [17].

Regular exercise has anti-inflammatory effects, thought to act both through loss of visceral fat, which reduces the release of inflammatory adipokines, and through an anti-inflammatory response triggered by each bout of exercise [18]. Chronic low-grade inflammation is linked to cardiovascular disease, type 2 diabetes and other age-related conditions.

Exercise also affects the brain. In a randomised trial of 120 older adults, aerobic training increased the volume of the anterior hippocampus, a memory region that shrinks with age, by 2%, equivalent to reversing one to two years of age-related loss, and the increase was associated with higher serum levels of brain-derived neurotrophic factor (BDNF) [19]. Muscle-strengthening activity maintains muscle mass and strength, whose loss with age (sarcopenia) is linked to disability, falls and death.

Most estimates of the effect of exercise on lifespan come from observational studies, which can show association but not cause. The main source of bias is reverse causation: people who are becoming ill tend to reduce their activity, so inactive groups include more people who were already likely to die sooner. Studies reduce this by excluding early deaths and adjusting for existing disease, but some residual bias is likely. Active people also differ from inactive people in diet, smoking, income, education and inherited fitness, and statistical adjustment for these factors is incomplete.

Much of the older evidence relies on self-reported activity, which is imprecise; the life expectancy analysis named self-reported activity and body mass index as a limitation [2]. Device-measured studies avoid recall error but have shorter follow-up and smaller samples, mostly of middle-aged and older adults, and they found larger differences between the most and least active groups than questionnaire studies [10].

Randomised trials with death as the outcome are rare and hard to run. In Generation 100, 1,567 Norwegians aged 70 to 77 were randomised to two weekly sessions of high-intensity interval training, moderate continuous training, or national activity advice for five years [6]. All-cause mortality did not differ between the control group (4.7% died) and the combined exercise groups. The high-intensity group had a hazard ratio of 0.63 (95% CI 0.33 to 1.20) compared with controls, a difference that was not statistically significant. Most participants were healthy and already active at the start, and control participants did more of their own activity as high-intensity exercise than the moderate-training group did, which reduced the contrast between groups [6]. The consistency of the observational findings across countries and measurement methods, the graded relationship with amount of activity, and the trial evidence on risk factors are the main reasons exercise is still regarded as causally linked to longer life.