Highlights
-
•
Both higher volume and intensity of physical activity are associated with longer life expectancy with no apparent threshold effect, especially in women, such that those with the highest physical activity volumes combined with the highest intensity profiles lived the longest.
-
•
Differences in physical activity volume and intensity equivalent to adding a daily 10-min brisk walk were associated with a longer life expectancy of 0.9 years in inactive women and 1.4 years in inactive men.
-
•
The findings support public health campaigns that focus on 10 min a day of brisk walking for inactive adults.
Keywords: Life expectancy, Physical activity intensity, Physical activity volume, Survival, Walking
Abstract
Background
There is a lack of research examining the interplay between objectively measured physical activity volume and intensity with life expectancy. The purpose of the study was to investigate the interplay between objectively measured PA volume and intensity profiles with modeled life expectancy in women and men within the UK Biobank cohort study and interpret findings in relation to brisk walking.
Methods
Individuals from UK Biobank with wrist-worn accelerometer data were included. The average acceleration and intensity gradient were extracted to describe the physical activity volume and intensity profile. Mortality data were obtained from national registries. Adjusted life expectancies were estimated using parametric flexible survival models.
Results
40,953 (57.1%) women (median age = 61.9 years) and 30,820 (42.9%) men (63.1 years) were included. Over a median follow-up of 6.9 years, there were 1719 (2.4%) deaths (733 in women; 986 in men). At 60 years, life expectancy was progressively longer for higher physical activity volume and intensity profiles, reaching 95.6 years in women and 94.5 years in men at the 90th centile for both volume and intensity, corresponding to 3.4 additional years (95% confidence interval (95%CI): 2.4–4.4) in women and 4.6 additional years (95%CI: 3.6–5.6) in men compared to those at the 10th centiles. An additional 10-min or 30-min daily brisk walk was associated with 0.9 (95%CI: 0.5–1.3) and 1.4 years (95%CI: 0.9–1.9) longer life expectancy, respectively, in inactive women; and 1.4 years (95%CI: 1.0–1.8) and 2.5 (95%CI: 1.9–3.1) in inactive men.
Conclusion
Higher physical activity volumes were associated with longer life expectancy, with a higher physical activity intensity profile further adding to a longer life. Adding as little as a 10-min brisk walk to daily activity patterns may result in a meaningful benefit to life expectancy.
Graphical abstract
1. Introduction
Physical activity (PA) promotion is a mainstay of public health.1 PA guidelines have coalesced around moderate-to-vigorous intensity PA (MVPA).2 More recent iterations also note potential health benefits for light-intensity or total PA,3,4 raising the possibility that health benefits may be gained through focusing on overall PA. Studies to date have shown that both PA volume and intensity are associated with health outcomes,5, 6, 7, 8, 9, 10 with some evidence that PA of moderate-to-vigorous intensity may result in the strongest health associations.6, 7, 8,10 However, most studies report associations with time spent in different intensities of PA without adequate adjustment for total volume.11 Studies that have directly investigated the interplay between PA volume and intensity have produced equivocal findings for whether intensity is independently associated with health outcomes,6,7,12,13 which is compounded by the fact that PA volume and intensity are linked given volume is a product of intensity by time. Thus, these constructs are not “independent”, and this results in complex interpretations when analyzed jointly.14 Innovations in accelerometer metrics have helped address this limitation. The average acceleration observed over each 24-h period assesses overall movement, and is commonly used as a proxy for the total volume of PA, while the intensity gradient has been developed to describe the distribution of intensity across the same period.15 The intensity gradient uses a single metric to describe how overall acceleration is accumulated over each day by assessing the relative contribution that higher intensity PA makes to overall PA levels. For example, 2 higher intensity PAs of the same volume but different intensities (e.g., 30 min of fast walking vs. 15 min of jogging) effect the intensity gradient differently (greater effect for jogging).15 Importantly, average accelerometer correlates highly with traditional measures of time spent in MVPA (r > 0.9), whereas the correlation between average accelerometer and the intensity gradient are lower (r: 0.4‒0.5).15 As such, the intensity gradient enables meaningful investigation of joint associations with total PA volume and health outcomes.
In addition, investigations of the association of accelerometer-derived measures of PA volume and intensity with health outcomes have to date been one-dimensional, relying on measures of relative risk. Measures of absolute risk, particularly life expectancy estimates, are increasingly recognized as key health indicators that can be easier to translate into clinical practice with meaningful public health messages and impact.16,17 A previous analysis investigated the association of self-reported MVPA and objectively measured total PA with life expectancy, showing high vs. low categories of PA were associated with a 3–5 year difference in life expectancy at 45 years of age.18 However, no previous studies have attempted to investigate the relative importance of PA volume vs. intensity with regard to life expectancy. Our aim was therefore to investigate the interplay between objectively measured PA volume and intensity profiles with modeled life expectancy in women and men within the UK Biobank cohort study and interpret findings in relation to brisk walking. We hypothesise that both PA volume and intensity will be associated with longer life expectancy.
2. Methods
2.1. Cohort definition
The UK Biobank prospective cohort study recruited more than 500,000 women and men across England, Wales, and Scotland between March 2006 and July 2010. PA was assessed in a random sub-cohort (excluding those in the North West region) who were invited for a follow-up visit between February 2013 and December 2015.19 The overall response rate in the UK Biobank was 5.5%.20 A sub-set of UK Biobank participants with valid email addresses (n = 236,519) were invited to wear an accelerometer during follow-up, of which 106,053 accepted,19 representing 1.2% of the overall individuals originally invited to take part in the UK Biobank. These participants were invited to wear the Axivity AX3 wrist-worn triaxial accelerometer (Axivity, Newcastle, UK) continuously for 7 days.19 From the initial sample of 502,366 individuals, we excluded women who were pregnant and participants with missing data on confounders or who self-reported a previous doctor's diagnosis of cancer or cardiovascular disease (CVD), leaving 399,291 participants, of which 78,763 had accelerometer data. Of these, 6990 were identified as having developed cancer or CVD between study entry and accelerometer assessment and so were excluded, leaving 71,773 participants. The cohort flowchart and details of the definitions of confounders, CVD, and cancer are reported in Supplementary Fig. 1. The median (interquartile range (IQR)) time between the baseline and accelerometer wear visit was 5.7 (4.9‒6.5) years.
Ethical approval for UK Biobank was obtained from the North West Centre for Research Ethics Committee (MREC, 11/NW/0382). In Scotland, UK Biobank has approval from the Community Health Index Advisory Group (CHIAG). The study complies with the Declaration of Helsinki and written consent was obtained from all participants.
2.2. Accelerometer data: volume and intensity
Accelerometer time series data (5-s epoch) were downloaded and converted to R-format for entry into R-package GGIR Version 1.10-7 (http://cran.r-project.org).21 Participants were excluded if their accelerometer files showed fewer than 3 days of valid wear (defined as >16 h/day) or wear-data were not present for each 15-min period of the 24-h cycle; detection of non-wear has been described previously.22,23 Participants were also excluded if they failed calibration (including those not calibrated on their own data, using a post-calibration error of greater than 0.01 g (10 mg)). Average acceleration and intensity gradient (IG) were extracted for each valid day and then averaged.15 Average acceleration, reported in milligravity (mg) units, was quantified by the Euclidean Norm Minus One (ENMO) method. Average acceleration quantifies the average intensity of PA over a given duration and was calculated across the full 24-h day.15 It is therefore reflective of the overall level, or volume, of PA undertaken across the 24-h day. The IG uses the natural log of the negative curvilinear relationship between the intensity of a PA and time accumulated at that intensity, whereby a higher (less negative) value indicates proportionally more time accumulated in higher intensity activities.15 The IG therefore describes the intensity profile for each individual. The metrics are further described in Supplementary Table 1.
2.3. Outcomes
UK Biobank data linked with National Health Service (NHS) England (England and Wales) and NHS Central Register (Scotland) provided information on date of death. The follow-up time started at the accelerometer-wearing visit and terminated at the occurrence of death or censoring (September 30, 2021 for England and Wales; October 31, 2021 for Scotland). For descriptive purposes, information on cause of death related to cancer (International Classification of Diseases (ICD)-10 codes C00 to C97) and CVD (ICD-10 codes I00 to I99) were extracted.
2.4. Confounding variables
Information was extracted on the following potential confounders or mediators: Sex, material deprivation, number of medications, current employment, number of self-reported non-cancer illnesses, long-standing illness, disability or infirmary, red meat, processed meat, fruit and vegetable score, alcohol intake, sleep duration, smoking status, and body mass index (BMI). Material deprivation was assessed by the Townsend score, a composite of unemployment, non-car ownership, non-home ownership, and household overcrowding using census data at the postcode level. Number of medications, non-cancer illnesses, long-standing illness, and disability or infirmary were assessed by a researcher-administered interview, with diabetes, chronic kidney disease, and arthritis extracted as non-cancer illnesses of interest. Red meat, processed meat, and fruit and vegetable score were assessed by a food frequency questionnaire and converted into continuous variables, as reported previously.24 The fruit and vegetable score was classified as healthy (≥5 units/day) or unhealthy (<5 units/day). Alcohol intake and smoking status were also assessed by self-report. Average sleep duration was assessed by questionnaire and expressed as low (<7 h), normal (7–8 h), or high (>8 h). Information on age and season (using 2 orthogonal sine functions, as previously described),7 collected at accelerometer wear visit, were also extracted.
2.5. Statistical analysis
Descriptive values are reported as median and IQR for continuous variables and number and percentage for categorical ones.
All analyses were stratified by sex, given patterns of PA and life expectancy were anticipated to differ by sex. We initially quantified the Bayesian Information Criterion (BIC) in 4 sex-specific Royston-Parmar parametric survival models: With a linear or nonlinear (spline with 3 internal knots at the 25th, 50th, and 75th centiles of variable distributions and 2 boundary knots at min and max) association for both intensity and volume; and each of these 2 models with or without an interaction between intensity and volume.25 In women, the lowest BIC (better model) was found when including a linear association for volume and intensity without an interaction; in men, it was found when including a linear association with interaction.
We then used these 2 models to quantify the sex-specific life expectancy by predicting the survival curve for each individual with age as time scale; employing this time scale allows an estimate of age-specific survival probability across varying ages. Conditional on surviving until 60 years (which is close to the population median age), individual survival curves were predicted for the observed values of confounders and for varying volume and intensity (10th, 25th, 50th, 75th, and 90th centiles) up to a time horizon of 100 years, consistent with the approach used by the Office for National Statistics.26 Using the G-computation approach (i.e., regression standardization),27,28 survival curves were then averaged across all individuals to obtain the adjusted (also known as standardized, marginal, or population-averaged) survival curve for each volume and intensity centile, the area under which represents the residual life expectancy;25,29 using the 10th centile of volume and intensity as reference, we also estimated the hazard ratios and the difference between the adjusted survival curves to obtain the difference in life expectancy. Confounders included in the models were season (accelerometer wear visit), Townsend deprivation index, number of medications, number of self-reported non-cancer illnesses, long-standing illness, disability or infirmary, current employment, red meat, processed meat, fruit and vegetable score, alcohol intake, sleep duration, and smoking status. Age was not included as a covariate because it was the time scale.
Additionally, a sensitivity analysis was undertaken adjusting for factors that could be hypothesized to act as both mediators and confounders (BMI, diabetes, chronic kidney disease, and arthritis); Supplementary Fig. 2 shows the hypothesized relationships between exposure, outcome, confounders, and mediators. Sensitivity analyses were also undertaken to investigate whether the pattern of results was maintained when conditioned on surviving until 50 or 70 years of age, instead of until 60 years, and when adjusting only for accelerometer wear visit season (base model).
In order to translate the main results into interpretable findings with meaning for public health, we modeled the association of adding a daily brisk walk (10 min and 30 min) to each participant's average acceleration and intensity gradient on the life expectancy for groups of individuals at the bottom 10th, 25th, and 50th centiles of the PA volume and intensity profiles. The acceleration value used as indicative of brisk walking was 250 mg.18 For volume, we assumed that the introduced activity would replace time spent at the participant's average acceleration as previously reported.15,30 Therefore, for a given activity, the new volume is calculated by: New volume = Baseline volume + ((duration of added activity/1440) × (acceleration associated with added activity – average acceleration)). For the intensity gradient, we took the original 5-s epoch-level profile and, at day level, substituted time spent at low intensity with an equal amount of time spent at the intensity of the introduced activity. The new intensity gradient was calculated at the day level, then averaged across days. The physical volume and intensity profile values for each individual before and after adding 10 min of brisk walking were used to predict the association with life expectancy using the model described above.
All analyses were conducted with ALICE High Performance Computing at the University of Leicester using Stata routines, stpm2, and standsurv commands in Stata/BE Version 18.0 (Stata Corp., College Station, TX, USA); results are reported with 95% confidence interval (95%CI) and graphs prepared in Stata and Inkscape Version 1.2.1 (https://inkscape.org/release/inkscape-1.4/windows/64-bit/msi/?redirected=1). The statistical code is publicly available at GitHub (https://github.com/frazac82/StatsCode_Vol-Int-LE).
2.6. Equity, diversity, and inclusion
The author team for this analysis included 12 men, 3 women, 3 ethnic minority individuals, and 4 early career researchers. Our analysis considered men and women separately.
3. Results
3.1. Cohort characteristics
The baseline characteristics of the 40,953 (57.1%) women (median age = 61.9 years) and 30,820 (42.9%) men (63.1 years) included in the analyses are shown in Table 1, with the stratification by volume and intensity provided in Supplementary Tables 2 and 3. The continuous distributions of intensity and volume are depicted in Supplementary Fig. 3.
Table 1.
Characteristics of the cohort and events during follow-up.
| Women | Men | Total | |
|---|---|---|---|
| No. of individuals | 40,953 | 30,820 | 71,773 |
| Age (year) | 61.9 (55.1–67.5) | 63.1 (55.7–68.3) | 62.4 (55.3–67.8) |
| Townsend deprivation index | –2.4 (–3.8 to –0.1) | –2.5 (–3.9 to –0.3) | –2.5 (–3.8 to –0.2) |
| Number of treatments/medications | 1.0 (0.0–3.0) | 1.0 (0.0–3.0) | 1.0 (0.0–3.0) |
| Current employment status | |||
| Doing unpaid or voluntary work | 308 (0.8) | 84 (0.3) | 392 (0.5) |
| Full- or part-time student | 137 (0.3) | 49 (0.2) | 186 (0.3) |
| In paid employment or self-employed | 25,985 (63.5) | 20,921 (67.9) | 46,906 (65.4) |
| Looking after home and/or family | 1986 (4.8) | 135 (0.4) | 2121 (3.0) |
| None of the above | 190 (0.5) | 119 (0.4) | 309 (0.4) |
| Retired | 11,524 (28.1) | 8627 (28.0) | 20,151 (28.1) |
| Unable to work because of sickness/disability | 513 (1.3) | 371 (1.2) | 884 (1.2) |
| Unemployed | 310 (0.8) | 514 (1.7) | 824 (1.1) |
| Number of self-reported non-cancer illnesses | 1.0 (0.0–2.0) | 1.0 (0.0–2.0) | 1.0 (0.0–2.0) |
| Long-standing illness, disability or infirmity | 9522 (23.3) | 8076 (26.2) | 17,598 (24.5) |
| Diabetes | 707 (1.7) | 1195 (3.9) | 1902 (2.7) |
| Chronic kidney disease | 37 (0.1) | 32 (0.1) | 69 (0.1) |
| Arthritis | 3892 (9.5) | 1891 (6.1) | 5783 (8.1) |
| Red meat intake (times/week) | 1.5 (1.5–2.5) | 2.0 (1.5–2.5) | 1.5 (1.5–2.5) |
| Processed meat intake (times/week) | 0.5 (0.5–1.0) | 1.0 (0.5–3.0) | 1.0 (0.5–3.0) |
| Fruit and vegetable score (units/day) | |||
| <5 | 5002 (12.2) | 6791 (22.0) | 11,793 (16.4) |
| ≥5 | 35,951 (87.8) | 24,029 (78.0) | 59,980 (83.6) |
| Alcohol intake (frequency) | |||
| Daily or almost daily | 7779 (19.0) | 8613 (27.9) | 16,392 (22.8) |
| 3 or 4 times a week | 10,009 (24.4) | 8996 (29.2) | 19,005 (26.5) |
| Once or twice a week | 10,561 (25.8) | 7652 (24.8) | 18,213 (25.4) |
| 1–3 times a month | 5225 (12.8) | 2599 (8.4) | 7824 (10.9) |
| Special occasions only | 4879 (11.9) | 1656 (5.4) | 6535 (9.1) |
| Never | 2500 (6.1) | 1304 (4.2) | 3804 (5.3) |
| Sleep (h/day) | |||
| <7 | 8606 (21.0) | 6897 (22.4) | 15,503 (21.6) |
| 7 to 8 | 29,764 (72.7) | 22,316 (72.4) | 52,080 (72.6) |
| >8 | 2583 (6.3) | 1607 (5.2) | 4190 (5.8) |
| Smoking status | |||
| Never | 25,347 (61.9) | 16,841 (54.6) | 42,188 (58.8) |
| Former | 13,252 (32.4) | 11,486 (37.3) | 24,738 (34.5) |
| Current | 2354 (5.7) | 2493 (8.1) | 4847 (6.8) |
| Body mass index (kg/m2) | 25.2 (22.8–28.4) | 26.6 (24.4–29.2) | 25.9 (23.5–28.8) |
| Valid days of accelerometer measures | 6.0 (6.0, 6.0) | 6.0 (6.0, 6.0) | 6.0 (6.0, 6.0) |
| Volume (overall mg) | 28.2 (23.6–33.6) | 27.3 (22.5–33.0) | 27.8 (23.1–33.3) |
| Intensity (IG) | –2.574 (–2.679 to –2.463) | –2.507 (–2.621 to –2.383) | –2.547 (–2.658 to –2.427) |
| Death | 733 (1.8) | 986 (3.2) | 1719 (2.4) |
| Follow-up (person-years) | 281,518 | 210,805 | 492,322 |
| Rate (per 1000 person-years) | 2.6 (2.4–2.8) | 4.7 (4.4–5.0) | 3.5 (3.3–3.7) |
| Cancer mortality | 456 (1.1) | 520 (1.7) | 976 (1.4) |
| Cardiovascular mortality | 116 (0.3) | 217 (0.7) | 333 (0.5) |
Notes: Shown are median (interquartile range) or number (column-wise percentage), except for rates that are reported with 95% confidence interval. Intensity gradient (IG) refers to the distribution of the intensity of physical activity where a higher (less negative) value indicates proportionally more time is accumulated in higher intensity activities; mg is average acceleration measured in milli-gravitational units, proxy for physical activity volume. Details on IG and mg are reported in Supplementary Table 1. A greater Townsend index score indicates a greater degree of deprivation. Red meat intake was calculated as the total weekly frequency of beef (UK data field (DF 1369); https://biobank.ndph.ox.ac.uk/ukb/search.cgi), lamb/mutton (DF 1379), and pork (DF 1389) meat intake. The DF 1349 reports information on processed meat intake. The fruit and vegetable score was calculated as the total daily frequency of fresh fruit (DF 1309), dried fruit (DF 1319), cooked vegetable (DF 1289), and salad/raw vegetable (DF 1299) intake. Information was collected at the baseline visit (March 2006–July 2010), except for physical activity volume and intensity as well as age (which were collected at the accelerometer wear visit, February 2013–December 2015).
Over a median (IQR) follow-up of 6.9 (6.4–7.4) years (492,322 person-years), 1719 (2.4%) deaths occurred: 733 in women and 986 in men. Cancer was the main cause of mortality, accounting for 456 (62.2%) deaths in women and 520 (52.7%) deaths in men. The crude mortality rates were 2.6 (95%CI: 2.4–2.8) and 4.7 (95%CI: 4.4–5.0) per 1000 person-years in women and men, respectively (Table 1).
3.2. PA and life expectancy
Fig. 1 shows the modeled life expectancy at 60 years of age and corresponding hazard ratios across the 10th, 25th, 50th, 75th, 90th centiles of volume and intensity profiles, stratified by sex; Fig. 2 shows the difference in modeled life expectancy compared to those with a low volume (10th centile) and low intensity (10th centile) profile.
Fig. 1.
Life expectancy at 60 years and corresponding hazard ratios. Estimates, conditional on survival until 60 years and for a maximum age of 100 years, are adjusted for season (accelerometer wear visit), Townsend deprivation index, number of medications, current employment, number of self-reported non-cancer illnesses, long-standing illness, red meat, processed meat, fruit and vegetable score, alcohol intake, sleep, and smoking status (baseline visit). 95%CI = 95% confidence interval; IG = intensity gradient. Ref = Reference for hazard ratio (hazard ratio = 1).
Fig. 2.
Life expectancy difference at 60 years. Estimates, conditional on survival until 60 years and for a maximum age of 100 years, are adjusted for season (accelerometer wear visit), Townsend deprivation index, number of medications, current employment, number of self-reported non-cancer illnesses, long-standing illness, red meat, processed meat, fruit and vegetable score, alcohol intake, sleep, and smoking status (baseline visit). 95%CI = 95% confidence interval; IG = intensity gradient. Reference: 10th centile volume and intensity.
In women, the highest PA volume and intensity profile was associated with a life expectancy of 95.6 (95%CI: 94.3–97.0) years (Fig. 1), corresponding to 3.4 additional years (95%CI: 2.4–4.4) and a hazard ratio of 0.48 (95%CI: 0.38‒0.60) compared to women with the lowest PA volume and intensity (Figs. 1 and 2). For any given volume, women that had a high compared to low intensity profile lived for ≥1.7 additional years (Fig. 2). In men, those with the highest PA volume and intensity profile had a life expectancy of 94.5 (95%CI: 93.2–96.0) years (Fig. 1), equating to 4.6 additional years (95%CI: 3.6–5.6) and a hazard ratio of 0.44 (95%CI: 0.37‒0.53) compared to men with the lowest PA volume and intensity (Figs. 1 and 2). Up to and including the median levels of PA volume, men with a high compared to low PA intensity profile lived for ≥1.7 additional years (Fig. 2). However, the association with intensity profile was attenuated at higher volumes.
Fig. 3 reports the association with modeled life expectancy of adding a 10- and 30-min daily brisk walk in participants at the bottom 10th, 25th, and 50th centiles of overall PA volume and intensity. Women starting with a low volume (10th centile) were estimated to live 0.9 years (95%CI: 0.5–1.3) longer for a daily 10-min brisk walk and 1.4 years (95%CI: 0.9–1.9) longer for a daily 30-min brisk walk. The corresponding estimates for men were 1.4 (95%CI: 1.0–1.8) and 2.5 (95%CI: 1.9–3.1). Slightly lower differences in life expectancy were modeled for women and men at the bottom 25th or 50th centiles of volume and intensity.
Fig. 3.
Association of brisk walking with life expectancy at 60 years. Estimates, conditional on survival until 60 years and for a maximum age of 100 years, are adjusted for season (accelerometer wear visit), Townsend deprivation index, number of medications, current employment, number of self-reported non-cancer illnesses, long-standing illness, red meat, processed meat, fruit and vegetable score, alcohol intake, sleep, and smoking status (baseline visit). Associations with years of life are shown for 10 (red) and 30 (blue) min of brisk walking per day in individuals at the bottom 10th, 25th, and 50th centiles of both volume and intensity. 95%CI = 95% confidence interval.
3.3. Sensitivity analyses
The pattern of results after additional adjustment for BMI, diabetes, chronic kidney disease, and arthritis was consistent with the main analysis (Supplementary Table 4).
The pattern of results was also maintained when life expectancy was estimated at 50 or 70 years, rather than 60 years. However, actual differences in life expectancy were marginally higher at 50 years (3.6 and 4.9 years in women and men, respectively, at the highest vs. lowest levels of volume and intensity) and lower at 70 years (3.1 and 4.1 years, respectively) (Supplementary Fig. 4).
The base model also revealed the same overall pattern of results at ages 50, 60, and 70 years, although actual differences in life expectancy were numerically higher compared to the estimates from the adjusted model (Supplementary Fig. 5).
4. Discussion
In a cohort of middle-aged adults, PA volumes accumulated through a high PA intensity profile were associated with the longest life expectancy. At 60 years of age, women and men at the highest decile for both PA volume and intensity profiles lived 3.4 and 4.6 additional years, respectively, compared to those at the lowest decile for both PA volume and intensity profiles.
4.1. Policy implications
Behaviorally, the concept of PA intensity is embedded within that of PA volume; hence, increasing moderate- or vigorous-intensity PA will result in a greater PA volume and a higher intensity profile for inactive populations. For example, adding a 10-min brisk walk to daily habitual PA patterns was modeled to result in a longer life expectancy of 0.9 years in the least active women and 1.4 years in the least active men. This supports the importance of promoting small bouts of MVPA in inactive adults, including public health campaigns that focus on 10 min of brisk walking a day.31 Our results are also consistent with recent research showing that even low levels of vigorous-intensity PA are associated with substantial reductions in the relative risk of morbidity and mortality.32
4.2. Comparison to the literature
While other studies have concluded that the lowest relative risk of all-cause mortality or CVD are associated with higher volumes of moderate- or vigorous-intensity PA,5, 6, 7, 8, 9, 10 there are notable differences of interpretation with our study. Previous research focusing on relative risk have consistently shown a threshold for PA volume with little additional benefit beyond median values.6,9 Furthermore, studies that have formally stratified by volume and intensity have concluded that intensity appears the more important determinant for mortality or cardiovascular outcomes.6,7 We did not find evidence of a meaningful interaction between volume and intensity in women, with both displaying an independent dose–response pattern of association with life expectancy. Conversely, while men with the highest PA volume lived the longest, there appeared little additional benefit from having a high intensity profile at high volume, whereas there was a notable association with intensity profile at lower PA volumes. This difference compared to previous literature could reflect the importance of stratifying by sex, which previous studies with objective measures of PA have largely avoided due to a low number of events; it may also emphasize differences in the measurement properties of our measure of intensity profile compared to those of other approaches. Of note, our results are consistent with those of previous studies using subjective measures of MVPA that were able to stratify by sex, where those with the highest levels of self-reported PA were also found to have the highest life expectancy in a dose-response association.33,34
The differences in life expectancy generated by engaging in modest differences in accelerometer-assessed PA volume and intensity are similar to those reported for physically active lifestyles generally. For example, self-reported levels of PA that are consistent with or greater than the PA guidelines (150 min of moderate PA per week) are associated with an estimated difference in life expectancy of between 1–4 years compared to inactive adults.35 It has also been shown that the association of PA with life expectancy is stronger with accelerometer-assessed compared to self-reported measures,18 supporting the importance of objectively quantifying the impact of PA on survival using device-based measurement.
Our observational findings are supported by evidence using markers of aging, including leucocyte telomere length, a widely used marker of biological age.36 Interventional research has shown exercise training can act to attenuate age-related telomere shortening.37 Emerging exercise studies in humans and obese animal models have also shown a reduction in cellular senescence.38 However, the interventional evidence to date does not support conclusions around the optimal volume or intensity of PA needed to reduce aging-related processes.
4.3. Strengths and limitations
Strengths of our analysis include the outcome of modeled life expectancy using objective, device-based measurement of PA volume and intensity in a large, contemporary dataset, and in a research area of importance to public health. Limitations include the possibility that UK Biobank participants are healthier than the general population39 as well the use of a single time-point assessment of PA ∼5 years after entry into the cohort, which may have further contributed to cohort selection biases, limited generalizability, and a greater risk of residual confounding. The healthy nature of the accelerometer cohort within the UK Biobank was emphasized in our analysis, showing life expectancies in women and men that were a decade older than the UK average.40 Despite this difference, predicted PA energy expenditure within UK Biobank participants is comparable to national estimates,6 although other studies have shown that exposure-outcome associations for self-reported PA in UK Biobank participants may be over-estimated.19 A 7-day window of accelerometer wear may not reflect longer-term PA patterns. However, it has previously been shown that 7 days of accelerometer-assessed MVPA and total PA have good reproducibility over time at a population level.41 Some common PAs, such as resistance exercise training or cycling, may not be appropriately captured by wrist accelerometery. Furthermore, context is not captured, with occupational PA potentially contributing to the difference in the associations between women and men at high volume and intensity.42 With a median follow-up of 6.9 years after the accelerometer wear period, our analyses relied on modeled life expectancy. Longer follow-up and a greater number of mortality events will allow for more precise estimates in the future. Furthermore, different methods for estimating life expectancy have different strengths and limitations and can produce different estimates.16 We also acknowledge the follow-up period included the coronavirus disease 2019 (COVID-19) pandemic; however, only 3.3% of mortality events within our cohort were recorded with COVID-19 as the underlying cause of death. Lastly, it is possible that unmeasured factors may have further confounded reported associations or that some included confounders may have acted as mediators. Nevertheless, the overall pattern of results remained consistent across different levels of adjustment.
5. Conclusion
Higher PA intensity was associated with notable differences in life expectancy across the spectrum of PA volume in women and at lower PA volumes in men. For inactive adults, even modest differences equivalent to a 10-min daily brisk walk were associated with a longer life expectancy. Further research using surrogate measures of healthy aging and longevity, such as markers of cellular senescence, are needed to determine whether these associations are likely to be causal.
Authors’ contributions
FZ and PCD formed the core working group and developed the research question, developed the analysis code, drafted the manuscript, contributed to the interpretation and revised the manuscript for important intellectual content; AVR formed the core working group and developed the research question, drafted the manuscript, contributed to the interpretation and revised the manuscript for important intellectual content; CR formed the core working group and developed the research question, independently replicated the results, contributed to the interpretation and revised the manuscript for important intellectual content; JH and JG formed the core working group and developed the research question, contributed to the interpretation and revised the manuscript for important intellectual content; BDM contributed to the interpretation and revised the manuscript for important intellectual content; AB, YC, CE, JAL, UE, MJD, and KK contributed to the interpretation and revised the manuscript for important intellectual content; TY formed the core working group and developed the research question, drafted the manuscript, contributed to the interpretation and revised the manuscript for important intellectual content, acts as the study guarantor. All authors have read and approved the final version of the manuscript, and agree with the order of presentation of the authors.
Competing interests
The authors declare that they have no competing interests.
Acknowledgments
Data availability
Research was conducted using the UK Biobank Resource under Application #33266. The UK Biobank resource can be accessed by researchers on application. Variables derived for this study have been returned to the UK Biobank for future applicants to request. No additional data are available.
Acknowledgments
We are grateful to the participants of the UK Biobank Study and to those who collected and manage the data. We would also like to acknowledge the contribution of the reviewers of this manuscript who helped improve the content and interpretation of the findings. The study was funded by the National Institute for Health Research (NIHR) Leicester Biomedical Research Centre (BRC) and the Applied Research Collaborations East Midlands (ARC-EM). The views expressed are those of the author(s) and not necessarily those of the NIHR or the Department of Health and Social Care. The study was also supported by a UKRI project grant (MR/T031816/1).
Footnotes
Peer review under responsibility of Shanghai University of Sport.
Supplementary materials associated with this article can be found in the online version at doi:10.1016/j.jshs.2024.100970.
Supplementary materials
References
- 1.Booth FW, Roberts CK, Laye MJ. Lack of exercise is a major cause of chronic diseases. Compr Physiol. 2012;2:1143–1211. doi: 10.1002/cphy.c110025. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Kahlmeier S, Wijnhoven TM, Alpiger P, Schweizer C, Breda J, Martin BW. National physical activity recommendations: Systematic overview and analysis of the situation in European countries. BMC Public Health. 2015;15:133. doi: 10.1186/s12889-015-1412-3. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Piercy KL, Troiano RP, Ballard RM, et al. The physical activity guidelines for Americans. JAMA. 2018;320:2020–2028. doi: 10.1001/jama.2018.14854. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Bull FC, Al-Ansari SS, Biddle S, et al. World Health Organization 2020 guidelines on physical activity and sedentary behaviour. Br J Sports Med. 2020;54:1451–1462. doi: 10.1136/bjsports-2020-102955. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Ramakrishnan R, He JR, Ponsonby AL, et al. Objectively measured physical activity and all cause mortality: A systematic review and meta-analysis. Prev Med. 2021;143 doi: 10.1016/j.ypmed.2020.106356. [DOI] [PubMed] [Google Scholar]
- 6.Strain T, Wijndaele K, Dempsey PC, et al. Wearable-device-measured physical activity and future health risk. Nat Med. 2020;26:1385–1391. doi: 10.1038/s41591-020-1012-3. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Dempsey PC, Rowlands AV, Strain T, et al. Physical activity volume, intensity, and incident cardiovascular disease. Eur Heart J. 2022;43:4789–4800. doi: 10.1093/eurheartj/ehac613. [DOI] [PubMed] [Google Scholar]
- 8.Dempsey PC, Musicha C, Rowlands AV, et al. Investigation of a UK Biobank cohort reveals causal associations of self-reported walking pace with telomere length. Commun Biol. 2022;5:381. doi: 10.1038/s42003-022-03323-x. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Ekelund U, Tarp J, Steene-Johannessen J, et al. Dose-response associations between accelerometry measured physical activity and sedentary time and all-cause mortality: Systematic review and harmonised meta-analysis. BMJ. 2019;366:l4570. doi: 10.1136/bmj.l4570. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Saint-Maurice PF, Troiano RP, Berrigan D, Kraus WE, Matthews CE. Volume of light versus moderate-to-vigorous physical activity: Similar benefits for all-cause mortality? J Am Heart Assoc. 2018;7 doi: 10.1161/JAHA.118.008815. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Brady R, Brown WJ, Hillsdon M, Mielke GI. Patterns of accelerometer‒measured physical activity and health outcomes in adults: A systematic review. Med Sci Sports Exerc. 2022;54:1155–1166. doi: 10.1249/MSS.0000000000002900. [DOI] [PubMed] [Google Scholar]
- 12.Paluch AE, Bajpai S, Bassett DR, et al. Daily steps and all-cause mortality: A meta-analysis of 15 international cohorts. Lancet Public Health. 2022;7:e219–e228. doi: 10.1016/S2468-2667(21)00302-9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Paluch AE, Bajpai S, Ballin M, et al. Prospective association of daily steps with cardiovascular disease: A harmonized meta-analysis. Circulation. 2023;147:122–131. doi: 10.1161/CIRCULATIONAHA.122.061288. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Matthews CE, Saint-Maurice PF. The hare and the tortoise: Physical activity intensity and scientific translation. Eur Heart J. 2022;43:4815–4816. doi: 10.1093/eurheartj/ehac626. [DOI] [PubMed] [Google Scholar]
- 15.Rowlands AV, Edwardson CL, Davies MJ, Khunti K, Harrington DM, Yates T. Beyond cut points: Accelerometer metrics that capture the physical activity profile. Med Sci Sports Exerc. 2018;50:1323–1332. doi: 10.1249/MSS.0000000000001561. [DOI] [PubMed] [Google Scholar]
- 16.Chudasama YV, Khunti K, Gillies CL, et al. Estimates of years of life lost depended on the method used: Tutorial and comparative investigation. J Clin Epidemiol. 2022;150:42–50. doi: 10.1016/j.jclinepi.2022.06.012. [DOI] [PubMed] [Google Scholar]
- 17.Spiegelhalter D. Risk and uncertainty communication. Annu Rev Stat Appl. 2017;4:31–60. [Google Scholar]
- 18.Chudasama YV, Khunti KK, Zaccardi F, et al. Physical activity, multimorbidity, and life expectancy: A UK Biobank longitudinal study. BMC Med. 2019;17:108. doi: 10.1186/s12916-019-1339-0. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Doherty A, Jackson D, Hammerla N, et al. Large scale population assessment of physical activity using wrist worn accelerometers: The UK biobank study. PLoS One. 2017;12 doi: 10.1371/journal.pone.0169649. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Batty GD, Gale CR, Kivimäki M, Deary IJ, Bell S. Comparison of risk factor associations in UK Biobank against representative, general population based studies with conventional response rates: Prospective cohort study and individual participant meta-analysis. BMJ. 2020;368:m131. doi: 10.1136/bmj.m131. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Migueles JH, Rowlands AV, Huber F, Sabia S, van Hees VT. GGIR: A research community–driven open source R package for generating physical activity and sleep outcomes from multi-day raw accelerometer data. J Meas Phys Behav. 2019;2:188–196. [Google Scholar]
- 22.Dawkins NP, Yates T, Razieh C, et al. Differences in accelerometer-measured patterns of physical activity and sleep/rest between ethnic groups and age: An analysis of UK Biobank. J Phys Act Health. 2022;19:37–46. doi: 10.1123/jpah.2021-0334. [DOI] [PubMed] [Google Scholar]
- 23.Van Hees VT, Gorzelniak L, Dean León EC, et al. Separating movement and gravity components in an acceleration signal and implications for the assessment of human daily physical activity. PLoS One. 2013;8:e61691. doi: 10.1371/journal.pone.0061691. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Zaccardi F, Franks PW, Dudbridge F, Davies MJ, Khunti K, Yates T. Mortality risk comparing walking pace to handgrip strength and a healthy lifestyle: A UK Biobank study. Eur J Prev Cardiol. 2021;28:704–712. doi: 10.1177/2047487319885041. [DOI] [PubMed] [Google Scholar]
- 25.Royston P, Lambert PC. Flexible Parametric Survival Analysis Using Stata: Beyond the Cox Model. College Station, TX: Stata Press; 2011.
- 26.Office for National Statistics. Guide to Interpreting Past and Projected Period and Cohort Life Tables. Available at:https://www.ons.gov.uk/peoplepopulationandcommunity/healthandsocialcare/healthandlifeexpectancies/methodologies/guidetointerpretingpastandprojectedperiodandcohortlifetables. [accessed 24.04.2024].
- 27.Vansteelandt S, Keiding N. Invited commentary: G-computation–Lost in translation? Am J Epidemiol. 2011;173:739–742. doi: 10.1093/aje/kwq474. [DOI] [PubMed] [Google Scholar]
- 28.Syriopoulou E, Wästerlid T, Lambert PC, Andersson TM. Standardised survival probabilities: A useful and informative tool for reporting regression models for survival data. Br J Cancer. 2022;127:1808–1815. doi: 10.1038/s41416-022-01949-6. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Lambert PC. standsurv. Available at: https://pclambert.net/software/standsurv/. [accessed 04.10.2022].
- 30.Schwendinger F., Wagner J., Knaier R., et al. Reference values for accelerometer metrics and associations with cardiorespiratory fitness: A prospective cohort study of healthy adults and patients with heart failure. Curr Issues Sport Sci. 2023;8:029. doi: 10.36950/2023.2ciss029. [DOI] [Google Scholar]
- 31.Brannan MGT, Foster CE, Timpson CM, et al. Active 10 - a new approach to increase physical activity in inactive people in England. Prog Cardiovasc Dis. 2019;62:135–139. doi: 10.1016/j.pcad.2019.02.001. [DOI] [PubMed] [Google Scholar]
- 32.Ahmadi MN, Clare PJ, Katzmarzyk PT, et al. Vigorous physical activity, incident heart disease, and cancer: How little is enough? Eur Heart J. 2022;43:4801–4814. doi: 10.1093/eurheartj/ehac572. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Li Y, Pan A, Wang DD, et al. Impact of healthy lifestyle factors on life expectancies in the US population. Circulation. 2018;138:345–355. doi: 10.1161/CIRCULATIONAHA.117.032047. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Li Y, Schoufour J, Wang DD, et al. Healthy lifestyle and life expectancy free of cancer, cardiovascular disease, and type 2 diabetes: Prospective cohort study. BMJ. 2020;368:l6669. doi: 10.1136/bmj.l6669. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.Reimers CD, Knapp G, Reimers AK. Does physical activity increase life expectancy? A review of the literature. J Aging Res. 2012;2012 doi: 10.1155/2012/243958. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36.Blackburn EH, Epel ES, Lin J. Human telomere biology: A contributory and interactive factor in aging, disease risks, and protection. Science. 2015;350:1193–1198. doi: 10.1126/science.aab3389. [DOI] [PubMed] [Google Scholar]
- 37.Song S, Lee E, Kim H. Does exercise affect telomere length? A systematic review and meta-analysis of randomized controlled trials. Medicina. 2022;58:242. doi: 10.3390/medicina58020242. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38.Chen XK, Yi ZN, Wong GT, et al. Is exercise a senolytic medicine? A systematic review. Aging Cell. 2021;20:e13294. doi: 10.1111/acel.13294. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39.Fry A, Littlejohns TJ, Sudlow C, et al. Comparison of sociodemographic and health-related characteristics of UK Biobank participants with those of the general population. Am J Epidemiol. 2017;186:1026–1034. doi: 10.1093/aje/kwx246. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 40.Office for National Statistics. National life tables–life expectancy in the UK: 2018 to 2020. Available at:https://www.ons.gov.uk/peoplepopulationandcommunity/birthsdeathsandmarriages/lifeexpectancies/bulletins/nationallifetablesunitedkingdom/2018to2020#:∼:text=1.-,Main%20points,period%20of%202015%20to%202017. [accessed 21.09.2023].
- 41.Keadle SK, Shiroma EJ, Kamada M, Matthews CE, Harris TB, Lee IM. Reproducibility of accelerometer‒assessed physical activity and sedentary time. Am J Prev Med. 2017;52:541–548. doi: 10.1016/j.amepre.2016.11.010. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 42.Cillekens B, Lang M, van Mechelen W, et al. How does occupational physical activity influence health? An umbrella review of 23 health outcomes across 158 observational studies. Br J Sports Med. 2020;54:1474–1481. doi: 10.1136/bjsports-2020-102587. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
Research was conducted using the UK Biobank Resource under Application #33266. The UK Biobank resource can be accessed by researchers on application. Variables derived for this study have been returned to the UK Biobank for future applicants to request. No additional data are available.