Association of physical activity intensity and bout length with mortality: An observational study of 79,503 UK Biobank participants

Louise A C Millard; Kate Tilling; Tom R Gaunt; David Carslake; Deborah A Lawlor

doi:10.1371/journal.pmed.1003757

. 2021 Sep 15;18(9):e1003757. doi: 10.1371/journal.pmed.1003757

Association of physical activity intensity and bout length with mortality: An observational study of 79,503 UK Biobank participants

Louise A C Millard ^1,^2,^*, Kate Tilling ^1,², Tom R Gaunt ^1,², David Carslake ^1,², Deborah A Lawlor ^1,²

Editor: Soren Brage³

PMCID: PMC8480840 PMID: 34525088

Abstract

Background

Spending more time active (and less sedentary) is associated with health benefits such as improved cardiovascular health and lower risk of all-cause mortality. It is unclear whether these associations differ depending on whether time spent sedentary or in moderate-vigorous physical activity (MVPA) is accumulated in long or short bouts. In this study, we used a novel method that accounts for substitution (i.e., more time in MVPA means less time sleeping, in light activity or sedentary) to examine whether length of sedentary and MVPA bouts associates with all-cause mortality.

Methods and findings

We used data on 79,503 adult participants from the population-based UK Biobank cohort, which recruited participants between 2006 and 2010 (mean age at accelerometer wear 62.1 years [SD = 7.9], 54.5% women; mean length of follow-up 5.1 years [SD = 0.73]). We derived (1) the total time participants spent in activity categories—sleep, sedentary, light activity, and MVPA—on average per day; (2) time spent in sedentary bouts of short (1 to 15 minutes), medium (16 to 40 minutes), and long (41+ minutes) duration; and (3) MVPA bouts of very short (1 to 9 minutes), short (10 to 15 minutes), medium (16 to 40 minutes), and long (41+ minutes) duration. We used Cox proportion hazards regression to estimate the association of spending 10 minutes more average daily time in one activity or bout length category, coupled with 10 minutes less time in another, with all-cause mortality. Those spending more time in MVPA had lower mortality risk, irrespective of whether this replaced time spent sleeping, sedentary, or in light activity, and these associations were of similar magnitude (e.g., hazard ratio [HR] 0.96 [95% CI: 0.94, 0.97; P < 0.001] per 10 minutes more MVPA, coupled with 10 minutes less light activity per day). Those spending more time sedentary had higher mortality risk if this replaced light activity (HR 1.02 [95% CI: 1.01, 1.02; P < 0.001] per 10 minutes more sedentary time, with 10 minutes less light activity per day) and an even higher risk if this replaced MVPA (HR 1.06 [95% CI: 1.05, 1.08; P < 0.001] per 10 minutes more sedentary time, with 10 minutes less MVPA per day). We found little evidence that mortality risk differed depending on the length of sedentary or MVPA bouts. Key limitations of our study are potential residual confounding, the limited length of follow-up, and use of a select sample of the United Kingdom population.

Conclusions

We have shown that time spent in MVPA was associated with lower mortality, irrespective of whether it replaced time spent sleeping, sedentary, or in light activity. Time spent sedentary was associated with higher mortality risk, particularly if it replaced MVPA. This emphasises the specific importance of MVPA. Our findings suggest that the impact of MVPA does not differ depending on whether it is obtained from several short bouts or fewer longer bouts, supporting the recent removal of the requirement that MVPA should be accumulated in bouts of 10 minutes or more from the UK and the United States policy. Further studies are needed to investigate causality and explore health outcomes beyond mortality.

Louise Millard and co-workers study associations between bouts of moderate-to-vigorous physical activity and mortality.

Author summary

Why was this study done?

Spending more time active and less time sedentary is associated with many health benefits.
It is currently not clear whether these benefits differ if time spent active or sedentary is accumulated in longer versus shorter bouts or depending on which type of activity they replace.
Recent policy changes in the UK and the US removed the recommendation that activity should be accumulated in bouts of 10 minutes of more.

What did the researchers do and find?

We used data on 79,503 UK-based adult participants and assessed whether mortality risk differed if people did longer versus shorter bouts of moderate-vigorous physical activity (MVPA) or sedentary behaviour.
We found that more time spent in MVPA was associated with lower mortality risk irrespective of whether it replaced light activity, sedentary time, or sleep; more time spent sedentary was associated with higher mortality risk, particularly if it replaced MVPA.
We found little evidence to suggest that mortality risk differs depending on the length of MVPA or sedentary bouts.

What do these findings mean?

Interventions that increase time spent in MVPA or reduce sedentary time (or both) could benefit health.
Any time spent in MVPA may be beneficial, even bouts of shorter (e.g., 1 minute) duration, supporting recent UK and US policy changes removing the requirement that MVPA should be accumulated in bouts of 10 minutes or more.

Introduction

Physical activity is associated with many health benefits such as better cardiovascular health and reduced risk of some cancers and type 2 diabetes [1]. A recent systematic review of prospective studies suggested that higher levels of physical activity at any intensity, and less time spent sedentary, are associated with a reduced risk of mortality [2].

Policy in the UK recommends that people accumulate 150 minutes each week in moderate physical activity or 75 minutes in vigorous activity [3], while policy in the US and the World Health Organization (WHO) guidelines have been recently updated to recommend ranges (150 to 300 minutes each week for moderate intensity and 75 to 150 minutes for vigorous intensity) rather than minimum amounts alone [4,5]. Until recently, the advice also stated that activity should be accumulated in bouts of 10 minutes or more, but this has now been removed from the UK, the US, and WHO guidelines [3–5]. These changes were based on evidence from cross-sectional, prospective cohort, and randomised trials. For example, the removal of minimum bout length from WHO guidelines was based on a systematic review [6] of 27 research studies: 13 cross-sectional studies, 2 prospective cohort studies, 1 nonrandomised trial, and 11 randomised trials. The trials had small sample sizes (all ≤255) and short-term follow-up (≤18 months). The largest sample size among the included prospective cohort and cross-sectional studies was 6,321.

Few prospective cohort studies have assessed how the duration of moderate-vigorous physical activity (MVPA) bouts relates to health. A meta-analysis of 29,734 children (4 to 18 years old) across 21 cohort studies found a similar benefit of MVPA on cardiometabolic risk factors across different bout durations [7]. In that study, an isotemporal approach was used to estimate associations of spending more time in one MVPA bout duration category coupled with less time in another MVPA bout duration category. They controlled for the overall time in MVPA to investigate its composition, but did not account for time spent in other activity categories such as sleeping or sedentary [7]. Of 3 studies in adults, 2 found no notable association of MVPA bout length with their respective outcomes: cardiovascular risk factors (N = 2,190) [8] and all-cause mortality (N = 4,840) [9]. The other (N = 3,250) reported smaller mean waist circumference and lower body mass index (BMI) in those who spent more time in MVPA bouts of 10+ minutes rather than shorter bouts [10]. None of these studies considered couplings of activity categories, thus did not examine whether results differed depending on the form of activity substituted for MVPA. They all grouped bouts ≥10 minutes together [7–10]. Other studies have used 2 summary variables to characterise MVPA bouts: (1) the number of bouts; and (2) the average time spent in bouts (in total) per day, but these do not describe the range of bout lengths a person undertakes or how often they undertake them [11–15]. We have found only 1 study that examined the importance of sedentary bout length (N = 7,985 adults) [16]. It found that higher percentage of total sedentary time in shorter sedentary bouts (< = 29 minutes) was associated with lower mortality, but overall time spent sedentary was not accounted for (S1 Fig).

The aim of our study is to examine whether mortality differs depending on time spent in different activity categories (e.g., including sleep and sedentary, not just being physically active) and whether time spent sedentary or in MVPA is accumulated in longer versus shorter bouts. We use a novel analytical approach that addresses limitations of previous studies to assess associations of overall time spent in different activity categories and bout length categories in terms of coupling more time spent in one category with less time in another category.

Methods

The analysis plan was developed by LACM, DAL, KT, and TRG prior to analyses beginning. We initially sought to investigate the impact of physical activity bout length on BMI, but changed this to all-cause mortality because BMI is measured prior to accelerometer wear in the UK Biobank. Following reviewers’ comments, we made one change to our main analysis: splitting our MVPA 1- to 15-minute bout length stratum into 2 categories: 1 to 9 minutes and 10 to 15 minutes, so that the different impact of <10- and > = 10-minute bouts could be directly assessed. We also added 2 additional sensitivity analysis: (1) using the isometric log ratio transformation as an alternative approach to analysing compositional data; and (2) repeating our main analyses excluding the first year and first 2 years of follow-up to explore whether undiagnosed prevalent disease might confound our results. This study is reported as per the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guideline (S1 STROBE Checklist).

Participants

We used data from the UK Biobank participants. UK Biobank is a large prospective cohort of approximately 500,000 adults (5% of those invited [17]) aged 40 to 69 years old at recruitment between 2006 and 2010 [18]. Written informed consent was obtained to collect and store data and bio samples, to link participants to health and administrative data, and for researchers to use these data for health research. UK Biobank received ethical approval from the UK National Health Service’s National Research Ethics Service (ref 11/NW/0382). This research was conducted under UK Biobank application number 17810.

In 2013, participants who had provided valid email addresses were invited to participate in the accelerometer substudy, apart from the participants of one assessment centre (3,797 participants; 0.76% of the cohort) who were excluded due to participant burden concerns as they had been invited to participate in pilot studies [19]. Between 2013 and 2015, participants were sent devices in order of acceptance. Of those invited, 106,053 agreed to participate, and 103,711 (44% of invited) provided some accelerometer data [19]. Our study includes 84,176 participants with at least 72 hours accelerometer wear time and no missing data for confounding factors used in our analyses (Fig 1, Section A in S1 Text).

Data collection

Physical activity measurement

Physical activity was measured using the Axivity AX3 wrist-worn tri-axial accelerometer and stored in gravity (g) units. The devices were configured to start recording at 10 AM 2 working days after dispatching to participants by post and to record tri-axial acceleration at a frequency of 100 Hz (dynamic range +/− 8 g) for 7 days. This results in over 181 million values per person (100 values per second across 7 days for each of the 3 axes).

All-cause mortality

Date of death was obtained from the NHS Information Centre for participants from England and Wales and the NHS Central Register for participants from Scotland. For each participant, follow-up was defined as starting at the age they stopped wearing the accelerometer. End of follow-up was defined as either participant’s age on December 31, 2019 for those in England, Wales, and Scotland (the end of our predefined follow-up period) or their age at death for those who died before the end of follow-up.

Potential confounders

We considered the following to be likely confounding factors (based on their known or plausible effects on physical activity and mortality): sex, age at the time of accelerometer wear, ethnicity, socioeconomic position, smoking, BMI, and general ill health (S2 Fig). Sex, ethnicity, and smoking status (never, current, or previous) were self-reported at the baseline assessment. We used education level, household total income, and Townsend deprivation index (a score representing the deprivation of the participant’s neighbourhood) to reflect participant’s socioeconomic position (see Section B in S1 Text for details of these measures). At the baseline assessment, weight was measured (to the nearest 100 g) in light clothing and unshod using a Tanita BC418MA body composition analyser and height to the nearest cm using a Seca 202 device. We used 3 indicator variables for existing cardiovascular diseases, cancer, and respiratory diseases prior to accelerometer wear as measures of baseline ill health.

The season in which participants wore the accelerometer, while not a confounder since it would not plausibly affect subsequent risk of death, may affect activity. We therefore derived 2 variables denoting the day of the year on which the accelerometer was worn using the cosine function approach, defined as $c_{1} = c o s (\frac{2 π d}{365})$ and $c_{2} = s i n (\frac{2 π d}{365})$ , where d is the day of the year accelerometer wear began [20]. We included c₁ and c₂ as covariates in our models to reduce the variation in activity exposure variables.

Accelerometer data preprocessing

We used the UK Biobank accelerometer analysis tool (available at https://github.com/activityMonitoring/biobankAccelerometerAnalysis/) [19,21,22] to preprocess the accelerometer data and derive summary activity variables for each 1-minute epoch in each participant’s accelerometer time series. The steps conducted by this tool include resampling x/y/z axes to 100 Hz, calibration to local gravity [23], noise and gravity removal, epoch generation—including both average vector magnitude and machine learning predictions of physical activity categories for each epoch—and nonwear detection. The machine learning model [21] predicts activity categories (sleep, sedentary, walking, light activity, and MVPA) from accelerometer data. It was trained using accelerometer data captured in free-living conditions and labelled with “ground truth” activities from accompanying videos and the Compendium of Physical Activities determined using a body-worn camera [24].

Statistical analyses

Dealing with missing accelerometer data

While the UK Biobank participants were asked to wear the accelerometer continuously for 7 days, 24% of our sample had some missing data. We identified periods of nonwear using the Biobank accelerometer analysis tool, defined as consecutive stationary episodes (where all 3 axes had a standard deviation of less than 13.0 milligravities [m-grav]) lasting for at least 60 minutes (the sensitivity of the accelerometer makes it possible to detect very small movements indicating it is being worn) [19]. We used 2 approaches to explore missing accelerometer data—a “complete days” approach and an “other day” imputation approach—that make different missingness assumptions. The complete days approach uses only days with complete accelerometer data in our analyses (referred to as “valid” days). The other day imputation approach involved finding all periods of accelerometer data on other days that are during the same time period and have no missing data (including from days with missing data at other times). One of these periods is then randomly chosen as the imputed sequence for the missing region. Imputed “valid” days are those with no missing data after this imputation. Details on missing data assumptions of these approaches are provided in Section C in S1 Text. We report results using the complete days data as our main results, and results using the imputed data are provided in the Supporting information.

Deriving physical activity bouts

We assigned each 1-minute epoch (interval) of accelerometer data to an activity category—either sleep, sedentary, light activity, or MVPA. The machine learning model (see “Accelerometer data preprocessing” section above) predicted the activity categories with varying levels of success. For example, while 91% of minutes spent sleeping were correctly classified as sleep, only 25% of light activity minutes were correctly classified as light activity. For this reason, we used a hybrid approach that first identified MVPA as minutes≥100 m-grav (a threshold used in previous research [25]) and then used the machine learning model to identify minutes of sleep and sedentary behaviour from those not already assigned to MVPA [21]. All other minutes not assigned to MVPA, sleep, or sedentary categories were assigned to the light activity category. For each participant, we identified contiguous sequences of 1-minute epochs with a given activity category; these are referred to as activity bouts and can be of any length so long as the participant remains in the same activity category.

As a sensitivity analysis, we used only the machine learning model to define all categories and refer to this as the ML-only approach. As well as categories of MVPA, sleep, and sedentary, this model predicts walking and light activity. There are 2 reasons we used the hybrid approach as our main analysis: (1) the degree of misclassification of the machine learning model for MVPA estimated in the study publishing this model [21] (e.g., only 58% of MVPA minutes were predicted correctly as MVPA); and (2) the activity categorisation in [21] included a separate walking category, whereas we sought to categorise brisk walking as moderate activity and slow walking as light activity, with no separate walking category.

Deriving summary variables reflecting time spent in activity categories, overall, and in bout length strata

For both approaches, we calculated, for each participant, the overall time they spent in each activity category, on average per day. Hence, for each participant, the total across all activity categories (for both the hybrid and ML-only approaches) equalled 1,440, the number of minutes in a day. To investigate whether the association of time spent in MVPA (or sedentary) with all-cause mortality changes depending on the time participants spend in bouts of different duration, we categorised them as short (1 to 15 minutes), medium (16 to 40 minutes), and long (41+ minutes). For MVPA, we further split the short category into 2 subcategories (1 to 9 minutes and 10 to 15 minutes) so that bouts <10 minutes and >10 minutes can be compared (at the request of reviewers). We derived the time spent in MVPA and sedentary bouts of each length, on average per day. Further details on our derivation of activity summary variables are provided in Section D in S1 Text.

Estimating the association of overall time spent in activity categories with all-cause mortality taking account of total time spent in that activity and coupling of spending more time in one activity category with less time in another category

We used Cox proportional hazards regression to test the association of each activity summary variable with all-cause mortality. All models were performed with age as the time variable. We tested each association before and after adjustment for potential confounders. Exact dates of birth were not available so ages were estimated assuming birth on July 1 in the reported year of birth.

It is possible that BMI and ill health subsequent to activity assessment could mediate the effect of activity on mortality. While we adjusted for BMI and ill health assessed at baseline (3 to 9 years prior to activity assessment), tracking of these factors across time (e.g., due to factors that affect BMI across the life course) means that BMI and ill health measured before activity are also proxies for these factors measured after activity (S2 Fig). Adjusting for proxies of mediating factors could attenuate our estimates towards the null [26]. We therefore performed a sensitivity analysis excluding BMI and ill health as covariates.

Within 1 day, there are 1,440 minutes so a greater amount of time spent in one activity category must be coupled with a lesser amount of time spent in one or more other activity categories. For this reason, we model associations in terms of couplings of activity categories, in a similar way to our previous activity bigrams approach [27]. We assign, in turn, one activity category as the baseline and estimate the hazard associated with spending 10 minutes less time in this baseline category when coupled with spending 10 minutes more time in a given comparison category, on average per day. Further details of this approach are provided in Section E in S1 Text.

Relating time spent in short, medium, and long bouts of MVPA or sedentary behaviour with all-cause mortality

As with our models for overall time, to estimate the association of time spent in sedentary and MVPA bouts of different length with all-cause mortality, we model these associations in terms of couplings of activity categories. Hence, unlike previous work [7–10,16], our models estimate associations for couplings between MVPA and sedentary bouts and also with other activity categories. Further details are provided in Section F in S1 Text.

We conducted sensitivity analyses for all models, starting follow-up 1 year and 2 years after the start of accelerometer wear, to investigate whether our results may be biased by participants reducing their activity due to existing ill health. We also conducted sensitivity analyses using the isometric log ratio transformation, which accounts for the compositional nature of time spent in activity categories [28].

For all models, we generated Schoenfeld residuals and estimated the correlation between log-transformed survival time and the scaled Schoenfeld residuals to test the proportional hazards assumption.

Analyses were performed in R version 3.5.1, Matlab r2015a or Stata version 15, and all of our analysis code are available at https://github.com/MRCIEU/UKBActivityBoutLength/. Git tag v0.2 corresponds to the version of the analyses presented here.

Results

Of the 84,176 eligible participants, 79,503 and 82,277 were included in our complete days and other day imputed samples, respectively (Fig 1). Other day imputation greatly increased the number of valid days (e.g., 96% and 24% of participants had 7 valid days in the imputed and complete days data; S3 Fig). Descriptive statistics of our main (complete days) sample are provided in Table 1. The total person years at risk in the complete days analysis was 405,438, and 1,615 participants died, giving a mortality rate of 5.10 per 1,000 person years (equivalent numbers for other day imputed sample were 419,520 and 1,688, with a mortality rate of 5.10 per 1,000 person years). Participants who were included in our analyses, compared with those who were invited to wear an accelerometer but did not respond, did not accept or had missing accelerometer of confounder data, were younger, more likely to be white, more educated, and living in an area with less social deprivation, had a higher income, lower BMI, were less likely to have ever smoked, less likely to have a circulatory disease or cancer, were more likely to have worn the accelerometer in winter, and were less likely to die during the follow-up period (S1 Table).

Table 1. Summary statistics of the UK Biobank participants in our sample.

	Mean (SD) or N (%)^a
Age in years at assessment centre (years)	55.89 (7.83)
Sex—% male	36,196 (45.53)
Ethnicity—white	77,145 (97.03)
Black or Black British	654 (0.82)
Asian or Asian British	686 (0.86)
Other	1,018 (1.28)
Smoking status—% ever	34,270 (43.11)
Income (pounds)—less than 18,000	11,557 (14.54)
18,000 to 30,999	19,197 (24.15)
31,000 to 51,999	22,864 (28.76)
52,000 to 100,000	20,063 (25.24)
>100,000	5,822 (7.32)
BMI (kg/m²)	26.71 (4.52)
Respiratory disease diagnosis	32,837 (41.30)
Circulatory disease diagnosis	24,723 (31.10)
Cancer diagnosis	11,157 (14.03)
Education—none of the below	5,931 (7.46)
College or university degree	35,871 (45.12)
A levels/AS levels or equivalent	28,748 (36.16)
O levels/GCSEs or equivalent	42,622 (53.61)
CSEs or equivalent	9,978 (12.55)
NVQ or HND or HNC or equivalent	14,603 (18.37)
Other professional qualifications (e.g., nursing and teaching)	27,986 (35.20)
Townsend deprivation index	−1.71 (2.82)
Death occurred	1,615 (2.03)
Season—winter (December to February)	17,372 (21.85)
Autumn (September to November)	23,243 (29.24)
Spring (March to May)	17,697 (22.26)
Summer (June to August)	21,191 (26.65)

Open in a new tab

^a Mean (SD) for continuous and percentage for binary variables. N = 79,503.

A level, Advanced level; AS level, Advanced Subsidiary level; BMI, body mass index; CI, confidence interval; CSE, Certificate of Secondary Education; GCSE, General Certificate of Secondary Education; HND, Higher National Diploma; HNC, Higher National Certificate; NVQ, National Vocational Qualification; SD, standard deviation.

S2 Table shows the distributions of the average number of minutes per day in each activity category. Sedentary time was more often accumulated in bouts of longer duration, and MVPA was more often accumulated in shorter bouts. Participants with more time sedentary on average spent more time in long sedentary bouts and less time in short or medium sedentary bouts (S3 Table). Participants who spent more time in MVPA on average spent more time in all categories of MVPA bout length, but particularly shorter bouts. On average, participants appeared to spend more time in MVPA using the hybrid approach compared with the ML-only approach. Patterns of correlation of overall time sedentary or in MVPA with bout length categories were similar for the hybrid and ML-only approaches. Correlations between the same characteristics derived using the hybrid and ML-only approaches were variable. Overall time sedentary was very strongly correlated (Pearson rho >0.99), with reasonable correlation for the sedentary bout length categories (Pearson rho of 0.73, 0.76, and 0.96 for short, medium, and long sedentary bouts, respectively) and lower correlations for MVPA (e.g., Pearson rho = 0.45 for overall time spent in MVPA and 0.26 for long MVPA bouts).

Associations of overall time spent in activity categories, with all-cause mortality

Associations of time spent in activity categories are shown in Fig 2. Overall, time spent in the different activity categories relates differently to mortality. Spending more time in MVPA was associated with lower mortality when coupled with less time spent sleeping, sedentary, or in light activity, and these associations were of a similar magnitude (e.g., hazard ratio [HR] 0.94 [95% CI: 0.93, 0.95; P < 0.001] and 0.96 [95% CI: 0.94, 0.97; P < 0.001] for 10 minutes more MVPA coupled with 10 minutes less time spent sedentary and in light activity, respectively). Those spending more time sedentary had higher mortality risk if this replaced light activity (HR 1.02 per 10 minutes more sedentary time, with 10 minutes less light activity per day [95% CI: 1.01, 1.02]; P < 0.001) and an even higher risk if this replaced MVPA (HR 1.06 per 10 minutes more sedentary time, with 10 minutes less MVPA per day [95% CI: 1.05, 1.08]; P < 0.001). Results of sensitivity analyses using the ML-only approach were largely consistent, although there were some differences (e.g., spending more time in light activity coupled with less time sleeping or sedentary were consistent with the null; S4 Table, S4 Fig). Results attenuated towards the null when starting follow-up 1 year and 2 years after accelerometer wear (S5 Fig).

Associations of MVPA and sedentary bout length with all-cause mortality

We found little evidence to suggest that associations differed across MVPA bout lengths (Fig 3A, S5 Table). For example, our estimate of association for spending 10 minutes less time in the shortest MVPA bouts (<10-minute duration) coupled with spending 10 minutes more time in long MVPA bouts (40+ minutes duration), with all-cause mortality, was consistent with the null (HR 1.01 [95% CI: 0.93, 1.10; P = 0.740]). We also found little evidence that associations differed across sedentary bout lengths (Fig 3B, S6 Table). For example, our estimate of association for spending 10 minutes less time in short sedentary bouts (<16 minutes duration) coupled with spending 10 minutes more time in long sedentary bouts (40+ minutes duration), with all-cause mortality, was consistent with the null (HR 1.03 [95% CI: 0.99, 1.06; P = 0.120]). Sensitivity analyses using the ML-only approach showed some differences compared with the hybrid approach (S7 Fig). Most notably, they suggest that spending less time in shorter sedentary bouts coupled with spending more time in longer sedentary bouts, associates with a lower all-cause mortality. Results starting follow-up 1 and 2 years after accelerometer wear were consistent with our main analysis (S8 Fig).

Results of sensitivity analyses using “other day” imputed data were broadly consistent with the results of our main analyses using the complete days data (S4, S6, S7, and S9 Figs, S4–S6 Tables). Results of sensitivity analyses excluding BMI and ill health as covariates were comparable to our main analyses (S4–S6 Tables). Results using isometric log ratio transformed activity variables were consistent with our main analyses (S10–S12 Figs).

We found little evidence of violation of the proportional hazards assumption across all Cox regression models (S4–S6 Tables).

Discussion

In this study, we found that time spent in MVPA was associated with lower mortality, irrespective of whether it was coupled with less time spent sleeping, sedentary, or in light activity and irrespective of whether it was obtained from several short bouts or fewer longer bouts. We also found that time spent sedentary was associated with higher mortality if it was coupled with less time in light activity (but to a lesser extent than if it was coupled with less time in MVPA). These findings emphasise the specific importance of MVPA. They also support recent changes to policy in the UK and the US and WHO guidelines that have removed the suggestion that MVPA should be accumulated in bouts of at least 10 minutes [3–5]. Those policy changes were made on the basis of cross-sectional, prospective cohort, and randomised controlled trial evidence, but those studies were small (e.g., in the systematic review on which this change in WHO guidelines was based, the largest observational study had 6,321 participants and the largest trial had 255 participants [6]).Our results do not support the specific promotion of accumulating MVPA in several smaller bouts but rather suggest that accumulating MVPA in any bout length could reduce risk of premature mortality. Similarly, they also suggest that replacing sedentary periods of any length with light activity, and, to a greater extent, with MVPA, could be beneficial. This is an important public health message as it allows people with different preferences and lifestyles to improve health through accumulating activity in different ways.

Importantly, the methods that we have used here address limitations of other studies that appear not to have controlled for overall time spent across all bout lengths of a given activity category [16], considered that greater amounts of one activity should be coupled with lesser amounts of another [8,10,16] or assessed each coupling combination [7,8,10,16]. We provide all of our code (https://github.com/MRCIEU/UKBActivityBoutLength/) so that others can use this method for exploring other outcomes, or risk factors for different patterns of activity, and examine associations in other studies with similar accelerometer data.

To our knowledge, there is only one existing study that assessed the association of MVPA bout length with mortality; it was considerably smaller than our study (N = 4,840), and, consistent with our findings, found no strong evidence of association between MVPA bout length and mortality [9]. Our findings contrast with those of a previous study that analysed sedentary bout length and concluded that longer versus shorter sedentary bouts (defined on the basis of the percent of all time spent sedentary) were associated with a higher risk of premature mortality [16]. We hypothesise that their results may be explained by an effect of total time spent sedentary on all-cause mortality, which was not taken into account in that study.

Study strengths and limitations

Strengths of this study include the large sample size and use of accelerometer data rather than self-report to measure activity and the prospective nature of the study. We have developed and used a method that appropriately accounts for coupling of activities. We have appropriately explored associations of total time spent in MVPA and sedentary with mortality, including whether this differs by bout length and depending on what alternative activity it is coupled with. This was possible because of our use of accelerometer data and would not be possible using the UK Biobank self-reported activity data. The UK Biobank self-reported data (or most other self-reported data) on activity bouts cannot be analysed in a compositional way because they do not include time spent in bouts of different length of each activity category (only average time spent in bouts for some activities or the number of days the participant did at least 10 minutes of moderate or vigorous activity). We undertook sensitivity analyses to assess missing accelerometer data assumptions. The code for generating our variables is freely available so can be used by others to explore associations with other health outcomes in the UK Biobank and in other studies with similar activity data.

Our study has a number of limitations. We used a previously published machine learning model to predict activity categories, and so it is possible that misclassifications of those predictions biased our estimates of association. For example, the model uses some orientation specific movement variables, and it is possible that the accelerometer orientation varied between participants. However, our main analysis used a hybrid approach where MVPA was identified using a threshold (>100 m-grav), since prediction accuracy for MVPA from the machine learning model was particularly low. This also has the benefit that average activity (denoted using the average vector magnitude) used to define MVPA in our hybrid approach is orientation independent. We also conducted sensitivity analyses using the machine learning predictions only (ML-only). These results were largely consistent for associations with overall time spent in each activity category, but showed some differences for our bout length results that may be due to biases in the types of activities assigned as MVPA by the ML-only approach compared with the hybrid approach. Further work is needed to compare the types of misclassifications of the hybrid and ML-only approaches.

Participants tended to spend relatively little time in MVPA overall and have MVPA bouts of short duration (the most common bout length was 1 minute, which was the shortest possible bout length in our data) so these estimates were imprecise. Further studies are needed in larger samples (e.g., when larger cohort studies are created) and with more precise measures of MVPA activity bouts (e.g., through more accurate prediction of MVPA using machine learning) to further explore these associations. We chose to use the same bout length strata for MVPA and sedentary behaviour for consistency, but we may have had more statistical power by defining strata according to the distribution of bout lengths for each category (e.g., participants spent more time in longer (versus shorter) sedentary bouts and more time in shorter (versus longer) MVPA bouts). We used 1-minute epochs to derive activity bouts (e.g., a 10-minute bout is a set of 10 adjacent 1-minute bouts), but using a different epoch definition may affect the values of derived bout variables and hence our results [29].

While we accounted for known, measured confounders, our analyses may be biased by residual confounding. It is possible that adjustment for other confounders might attenuate results (e.g., of overall time spent in MVPA) to the null. For example, it is possible that having mobility limitations, or little access to green space or facilities to be physically active, might be related to less time spent in light activity or MVPA and more sedentary behaviour and also to increased risk of mortality during follow-up. Adjustment for 3 different measures of socioeconomic position, including an area-based measure and BMI, is likely to have controlled for some of the potential confounding by these and therefore potentially reduced residual confounding [30]. Residual confounding could also occur due to undiagnosed underlying chronic disease, which could result in being less active and more sedentary, and be associated with increased mortality, particularly in the early years of follow-up. To explore this, we conducted sensitivity analyses starting follow-up 1 and 2 years after accelerometer wear. Results from these analyses showed some attenuation towards the null for our overall time spent in activity categories, which may suggest that our results are biased by confounding with existing ill health, but might also be explained by any true effect of activity on mortality being short term. Longer follow-up time would allow further sensitivity analyses starting follow-up 5 years after accelerometer wear. This, and repeat assessments of physical activity, would help to ensure that associations are not due to confounding via existing ill health and to explore the impact of changes in activity levels and whether any beneficial effect of activity might be short term.

Our use of time spent in each activity category and in activity bout length strata does not account for variability of activity levels within each of these. For example, participants spending more time in short MVPA bouts may have higher activity intensity levels within these compared to those spending more time in longer MVPA bouts. It also does not account for energy expenditure. Other recent work assessing the association of physical activity estimated energy expenditure (PAEE) with mortality, also in the UK Biobank, found that higher overall PAEE was associated with lower mortality and that associations were stronger with an increasing time spent in MVPA [31].

UK Biobank is a highly selected sample of the UK population with a response rate of 5.5% [32], and evidence suggests that those who volunteered are more affluent and healthy than those who did not [17]. The participants who were included here were also a more affluent and healthier group than the UK Biobank participants who were not included. This “selection” may mean our estimates are biased (see Section G in S1 Text for further discussion of this). Most of the participants in the UK Biobank are of white European origin, and our results may not generalise to other populations.

To conclude, we have used a novel approach to assess whether time spent in different activity types, and in short, medium, or long bouts of MVPA and sedentary behaviour, are associated with all-cause mortality. Our study confirms a strong association between active time and lower mortality, particularly for MVPA compared with light activity. We found little evidence that associations with time spent in MVPA or sedentary differ according to bout length. These results support the recent decision to amend the UK and the US physical activity guidelines to remove the advice that MVPA should be accumulated in bouts of 10 minutes or more [3,4]. Further work is needed to replicate our results in independent data and to investigate causality. Finally, our results highlight the importance of the isotemporal “coupling of time” perspective and suggest that this should be commonplace in any activity analyses, as public health advice based on increasing time spent in a given activity type is misleading without accompanying details of the activities from which this time should be taken.

Supporting information

S1 STROBE Checklist. STROBE, STrengthening the Reporting of OBservational studies in Epidemiology.

(DOC)

Click here for additional data file.^{(89KB, doc)}

S1 Fig. Directed acyclic graph illustrating path between bout length and survival through a common cause of bout length and total time spent sedentary.

(PDF)

Click here for additional data file.^{(59.8KB, pdf)}

S2 Fig. Directed acyclic graph illustrating hypothesised confounding factors and potential mediators.

(PDF)

Click here for additional data file.^{(78.9KB, pdf)}

S3 Fig. Distributions of the number of valid days of accelerometer data for participants included in our samples.

(PDF)

Click here for additional data file.^{(49.5KB, pdf)}

S4 Fig. Association of lower amounts of time spent in baseline activity category coupled with higher amounts of time in comparison category, with all-cause mortality using “other day” imputed accelerometer data.

(PDF)

Click here for additional data file.^{(88.5KB, pdf)}

S5 Fig. Results of sensitivity analysis using activity predictions only: Associations of lower amounts of time spent in baseline activity category coupled with higher amounts of time in comparison category, with all-cause mortality.

(PDF)

Click here for additional data file.^{(81.2KB, pdf)}

S6 Fig. Results of sensitivity analysis starting follow-up 1 and 2 years after accelerometer wear.

(PDF)

Click here for additional data file.^{(72.1KB, pdf)}

S7 Fig. Results of sensitivity analysis using ML-only approach: Association of time spent in MVPA and sedentary bouts of a given length, with all-cause mortality.

MVPA, moderate-vigorous physical activity.

(PDF)

Click here for additional data file.^{(114KB, pdf)}

S8 Fig. Results of sensitivity analysis starting follow-up 1 and 2 years after accelerometer wear: Association of time spent in MVPA and sedentary bouts of a given length, with all-cause mortality.

MVPA, moderate-vigorous physical activity.

(PDF)

Click here for additional data file.^{(102KB, pdf)}

S9 Fig. Association of time spent in MVPA and sedentary bouts of a given length, with all-cause mortality using hybrid approach and “other day” imputed data.

MVPA, moderate-vigorous physical activity.

(PDF)

Click here for additional data file.^{(87.1KB, pdf)}

S10 Fig. Results of sensitivity analysis using isometric log ratio transformed activity variables: Association of time spent in activity categories, with all-cause mortality.

(PDF)

Click here for additional data file.^{(74.3KB, pdf)}

S11 Fig. Results of sensitivity analysis using isometric log ratio transformed activity variables: Association of time spent in MVPA bouts of a given length, with all-cause mortality.

MVPA, moderate-vigorous physical activity.

(PDF)

Click here for additional data file.^{(108.4KB, pdf)}

S12 Fig. Results of sensitivity analysis using isometric log ratio transformed activity variables: Association of time spent in sedentary bouts of a given length, with all-cause mortality.

(PDF)

Click here for additional data file.^{(76.9KB, pdf)}

S13 Fig. Directed acyclic graph illustrating the potential for collider bias due to selection into the study sample.

(PDF)

Click here for additional data file.^{(59.3KB, pdf)}

S1 Table. Summary statistics of the UK Biobank participants who are in our sample, compared with those who were invited to wear an accelerometer but are not in our sample, for complete days version.

(DOCX)

Click here for additional data file.^{(43KB, docx)}

S2 Table. Summary of time spent in activity classifications on average per day.

(DOCX)

Click here for additional data file.^{(41.3KB, docx)}

S3 Table. Correlations between activity summary variables of main analysis (hybrid approach) and ML-only sensitivity approach.

(DOCX)

Click here for additional data file.^{(52.9KB, docx)}

S4 Table. Associations of transferring time between overall activity categories, with all-cause mortality.

(DOCX)

Click here for additional data file.^{(56.1KB, docx)}

S5 Table. Associations of transferring time from MVPA bout length category to other activity category, with all-cause mortality.

MVPA, moderate-vigorous physical activity.

(DOCX)

Click here for additional data file.^{(60.2KB, docx)}

S6 Table. Associations of transferring time from sedentary bout length categories to other activity category, with all-cause mortality.

(DOCX)

Click here for additional data file.^{(49.5KB, docx)}

S1 Text

Section A: Description of participant flow. Section B: Potential confounders. Section C: Missing data assumptions and imputation of accelerometer data. Section D: Deriving activity summary variables. Section E: Estimating the association of less time in a given activity category, when coupled with more time in another category. Section F: Estimating the association of less time in a given MVPA bout length stratum, when coupled with more time in another activity category/MVPA bout length stratum. Section G: Possible bias due to conditioning on a collider. MVPA, moderate-vigorous physical activity.

(DOCX)

Click here for additional data file.^{(58KB, docx)}

Acknowledgments

This research has been conducted using the UK Biobank Resource.

Abbreviations

BMI: body mass index
HR: hazard ratio
MVPA: moderate-vigorous physical activity
PAEE: physical activity estimated energy expenditure
STROBE: STrengthening the Reporting of OBservational studies in Epidemiology
WHO: World Health Organization

Data Availability

The data underlying the results presented in the study are available from UK Biobank [www.ukbiobank.ac.uk] for researchers who meet the criteria for access to these confidential data. Analysis code is available at [https://github.com/MRCIEU/UKBActivityBoutLength/].

Funding Statement

This work was supported by the University of Bristol (https://www.bristol.ac.uk/) and UK Medical Research Council (https://mrc.ukri.org/) [grant numbers MC_UU_00011/1, MC_UU_00011/3, MC_UU_00011/4 and MC_UU_00011/6]. LACM is funded by a University of Bristol Vice-Chancellor’s Fellowship. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

References

1.Lee I-M, Shiroma EJ, Lobelo F, Puska P, Blair SN, Katzmarzyk PT, et al. Impact of Physical Inactivity on the World’s Major Non-Communicable Diseases. Lancet. 2012;380:219–29. doi: 10.1016/S0140-6736(12)61031-9 [DOI] [PMC free article] [PubMed] [Google Scholar]
2.Ekelund U, Tarp J, Steene-Johannessen J, Hansen BH, Jefferis B, Fagerland MW, 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]
3.Physical activity guidelines: UK Chief Medical Officers’ report. 2019. Available from: https://www.gov.uk/government/publications/physical-activity-guidelines-uk-chief-medical-officers-report.
4.Piercy KL, Troiano RP, Ballard RM, Carlson SA, Fulton JE, Galuska DA, et al. The Physical Activity Guidelines for Americans. JAMA. 2018;320:2020–8. doi: 10.1001/jama.2018.14854 [DOI] [PMC free article] [PubMed] [Google Scholar]
5.Bull FC, Al-Ansari SS, Biddle S, Borodulin K, Buman MP, Cardon G, et al. World Health Organization 2020 guidelines on physical activity and sedentary behaviour. Br J Sports Med. 2020;54:1451–62. doi: 10.1136/bjsports-2020-102955 [DOI] [PMC free article] [PubMed] [Google Scholar]
6.Jakicic JM, Kraus WE, Powell KE, Campbell WW, Janz KF, Troiano RP, et al. Association between Bout Duration of Physical Activity and Health: Systematic Review. Med Sci Sports Exerc. 2019;51:1213–9. doi: 10.1249/MSS.0000000000001933 [DOI] [PMC free article] [PubMed] [Google Scholar]
7.Tarp J, Child A, Whitet T, Westgate K, Bugge A, Grøntved A, et al. Physical activity intensity, bout-duration, and cardiometabolic risk markers in children and adolescents. Int J Obes (Lond). 2018;42:1639–50. doi: 10.1038/s41366-018-0152-8 [DOI] [PMC free article] [PubMed] [Google Scholar]
8.Glazer NL, Lyass A, Esliger DW, Blease SJ, Freedson PS, Massaro JM, et al. Sustained and shorter bouts of physical activity are related to cardiovascular health. Med Sci Sports Exerc. 2013;45:109–15. doi: 10.1249/MSS.0b013e31826beae5 [DOI] [PMC free article] [PubMed] [Google Scholar]
9.Saint-Maurice PF, Troiano RP, Matthews CE, Kraus WE. Moderate-to-Vigorous Physical Activity and All-Cause Mortality: Do Bouts Matter? J Am Heart Assoc. 2018;7:e007678. doi: 10.1161/JAHA.117.007678 [DOI] [PMC free article] [PubMed] [Google Scholar]
10.Strath SJ, Holleman RG, Ronis DL, Swartz AM, Richardson CR. Objective physical activity accumulation in bouts and nonbouts and relation to markers of obesity in US adults. Prev Chronic Dis. 2008;5:A131. [PMC free article] [PubMed] [Google Scholar]
11.Brooke HL, Atkin AJ, Corder K, Brage S, van Sluijs EM. Frequency and duration of physical activity bouts in school-aged children: A comparison within and between days. Prev Med Rep. 2016;4:585–90. doi: 10.1016/j.pmedr.2016.10.007 [DOI] [PMC free article] [PubMed] [Google Scholar]
12.Masse L, Fuemmeler B, Anderson C, Matthews CE, Trost SG, Catellier DJ, et al. Accelerometer data reduction: a comparison of four reduction algorithms on select outcome variables. Med Sci. 2005;37:S544–54. doi: 10.1249/01.mss.0000185674.09066.8a [DOI] [PubMed] [Google Scholar]
13.Rowlands A, Pilgrim E, Eston R. Patterns of habitual activity across weekdays and weekend days in 9–11-year-old children. Prev Med. 2008;46:317–24. doi: 10.1016/j.ypmed.2007.11.004 [DOI] [PubMed] [Google Scholar]
14.Trost S, Pate R, Sallis J, Freedson P, Taylor WC, Dowda M, et al. Age and gender differences in objectively measured physical activity in youth. Med Sci. 2002;34:350–5. doi: 10.1097/00005768-200202000-00025 [DOI] [PubMed] [Google Scholar]
15.Riddoch CJ, Mattocks C, Deere K, Saunders J, Kirkby J, Tilling K, et al. Objective measurement of levels and patterns of physical activity. Arch Dis Child. 2007;92:963–9. doi: 10.1136/adc.2006.112136 [DOI] [PMC free article] [PubMed] [Google Scholar]
16.Diaz KM, Howard VJ, Hutto B, Colabianchi N, Vena JE, Safford MM, et al. Patterns of Sedentary Behavior and Mortality in U.S. Middle-Aged and Older Adults: A National Cohort Study. Ann Intern Med. 2017;167:465–75. doi: 10.7326/M17-0212 [DOI] [PMC free article] [PubMed] [Google Scholar]
17.Fry A, Littlejohns TJ, Sudlow C, Doherty N, Adamska L, Sprosen T, et al. Comparison of Sociodemographic and Health-Related Characteristics of UK Biobank Participants With Those of the General Population. Am J Epidemiol. 2017;186:1026–34. doi: 10.1093/aje/kwx246 [DOI] [PMC free article] [PubMed] [Google Scholar]
18.Sudlow C, Gallacher J, Allen N, Beral V, Burton P, Danesh J, et al. UK Biobank: An Open Access Resource for Identifying the Causes of a Wide Range of Complex Diseases of Middle and Old Age. PLoS Med. 2015;12:e1001779. doi: 10.1371/journal.pmed.1001779 [DOI] [PMC free article] [PubMed] [Google Scholar]
19.Doherty A, Jackson D, Hammerla N, Plötz T, Olivier P, Granat MH, et al. Large Scale Population Assessment of Physical Activity Using Wrist Worn Accelerometers: The UK Biobank Study. PLoS ONE. 2017;12:e0169649. doi: 10.1371/journal.pone.0169649 [DOI] [PMC free article] [PubMed] [Google Scholar]
20.Stolwijk AM, Straatman H, Zielhuis GA. Studying seasonality by using sine and cosine functions in regression analysis. J Epidemiol Community Health. 1999;53:235–8. doi: 10.1136/jech.53.4.235 [DOI] [PMC free article] [PubMed] [Google Scholar]
21.Doherty A, Smith-Byrne K, Ferreira T, Holmes MV, Holmes C, Pulit SL, et al. GWAS identifies 14 loci for device-measured physical activity and sleep duration. Nat Commun. 2018;9:5257. doi: 10.1038/s41467-018-07743-4 [DOI] [PMC free article] [PubMed] [Google Scholar]
22.Willetts M, Hollowell S, Aslett L, Holmes C, Doherty A. Statistical machine learning of sleep and physical activity phenotypes from sensor data in 96,220 UK Biobank participants. Sci Rep. 2018;8:7961. doi: 10.1038/s41598-018-26174-1 [DOI] [PMC free article] [PubMed] [Google Scholar]
23.van Hees VT, Fang Z, Langford J, Assah F, Mohammad A, da Silva ICM, et al. Autocalibration of accelerometer data for free-living physical activity assessment using local gravity and temperature: an evaluation on four continents. J Appl Physiol. 2014;117:738–44. doi: 10.1152/japplphysiol.00421.2014 [DOI] [PMC free article] [PubMed] [Google Scholar]
24.Ainsworth BE, Haskell WL, Herrmann SD, Meckes N, Bassett DRJR, Tudor-Locke C, et al. 2011 Compendium of Physical Activities: A Second Update of Codes and MET Values. Med Sci Sport Exerc. 2011;43:1575–81. doi: 10.1249/MSS.0b013e31821ece12 [DOI] [PubMed] [Google Scholar]
25.Bell JA, Hamer M, van Hees VT, Singh-Manoux A, Kivimaki M, Sabia S. Healthy obesity and objective physical activity. Am J Clin Nutr. 2015;102:268–75. doi: 10.3945/ajcn.115.110924 [DOI] [PMC free article] [PubMed] [Google Scholar]
26.Groenwold RHH, Palmer TM, Tilling K. To adjust or not to adjust? When a “confounder” is only measured after exposure. Epidemiology. 2021;32:194–201. doi: 10.1097/EDE.0000000000001312 [DOI] [PMC free article] [PubMed] [Google Scholar]
27.Millard LAC, Tilling K, Lawlor DA, Flach PA, Gaunt TR. Physical activity phenotyping with activity bigrams, and their association with BMI. Int J Epidemiol. 2017;46. doi: 10.1093/ije/dyx093 [DOI] [PMC free article] [PubMed] [Google Scholar]
28.McGregor DE, Palarea-Albaladejo J, Dall PM, Hron K, Chastin S. Cox regression survival analysis with compositional covariates: Application to modelling mortality risk from 24-h physical activity patterns. Stat Methods Med Res. 2020;29:1447–65. doi: 10.1177/0962280219864125 [DOI] [PubMed] [Google Scholar]
29.Orme M, Wijndaele K, Sharp SJ, Westgate K, Ekelund U, Brage S. Combined influence of epoch length, cut-point and bout duration on accelerometry-derived physical activity. Int J Behav Nutr Phys Act. 2014;11:34. doi: 10.1186/1479-5868-11-34 [DOI] [PMC free article] [PubMed] [Google Scholar]
30.Groenwold RHH, Sterne JAC, Lawlor DA, Moons KGM, Hoes AW, Tilling K. Sensitivity analysis for the effects of multiple unmeasured confounders. Ann Epidemiol. 2016;26:605–11. doi: 10.1016/j.annepidem.2016.07.009 [DOI] [PubMed] [Google Scholar]
31.Strain T, Wijndaele K, Dempsey PC, Sharp SJ, Pearce M, Jeon J, et al. Wearable-device-measured physical activity and future health risk. Nat Med. 2020;26:1385–91. doi: 10.1038/s41591-020-1012-3 [DOI] [PMC free article] [PubMed] [Google Scholar]
32.Swanson JM. The UK Biobank and selection bias. Lancet. 2012;380:110. doi: 10.1016/S0140-6736(12)61179-9 [DOI] [PubMed] [Google Scholar]

PLoS Med. doi: 10.1371/journal.pmed.1003757.r001

Decision Letter 0

Adya Misra

27 May 2020

Dear Dr Millard,

Thank you for submitting your manuscript entitled "Association of changing physical activity intensity and bout length with mortality: a study of 79,507 participants in UK Biobank" for consideration by PLOS Medicine.

Your manuscript has now been evaluated by the PLOS Medicine editorial staff [as well as by an academic editor with relevant expertise] and I am writing to let you know that we would like to send your submission out for external peer review.

However, before we can send your manuscript to reviewers, we need you to complete your submission by providing the metadata that is required for full assessment. To this end, please login to Editorial Manager where you will find the paper in the 'Submissions Needing Revisions' folder on your homepage. Please click 'Revise Submission' from the Action Links and complete all additional questions in the submission questionnaire.

Please re-submit your manuscript within two working days, i.e. by .

Once your full submission is complete, your paper will undergo a series of checks in preparation for peer review. Once your manuscript has passed all checks it will be sent out for review.

Feel free to email us at plosmedicine@plos.org if you have any queries relating to your submission.

Kind regards,

Adya Misra, PhD,

Senior Editor

PLOS Medicine

PLoS Med. doi: 10.1371/journal.pmed.1003757.r002

Decision Letter 1

Adya Misra

20 Aug 2020

Dear Dr. Millard,

Thank you very much for submitting your manuscript "Association of changing physical activity intensity and bout length with mortality: a study of 79,507 participants in UK Biobank" (PMEDICINE-D-20-02260R1) for consideration at PLOS Medicine.

Your paper was evaluated by a senior editor and discussed among all the editors here. It was also discussed with an academic editor with relevant expertise, and sent to independent reviewers, including a statistical reviewer. The reviews are appended at the bottom of this email and any accompanying reviewer attachments can be seen via the link below:

[LINK]

In light of these reviews, I am afraid that we will not be able to accept the manuscript for publication in the journal in its current form, but we would like to consider a revised version that addresses the reviewers' and editors' comments. Obviously we cannot make any decision about publication until we have seen the revised manuscript and your response, and we plan to seek re-review by one or more of the reviewers.

In revising the manuscript for further consideration, your revisions should address the specific points made by each reviewer and the editors. Please also check the guidelines for revised papers at http://journals.plos.org/plosmedicine/s/revising-your-manuscript for any that apply to your paper. In your rebuttal letter you should indicate your response to the reviewers' and editors' comments, the changes you have made in the manuscript, and include either an excerpt of the revised text or the location (eg: page and line number) where each change can be found. Please submit a clean version of the paper as the main article file; a version with changes marked should be uploaded as a marked up manuscript.

In addition, we request that you upload any figures associated with your paper as individual TIF or EPS files with 300dpi resolution at resubmission; please read our figure guidelines for more information on our requirements: http://journals.plos.org/plosmedicine/s/figures. While revising your submission, please upload your figure files to the PACE digital diagnostic tool, https://pacev2.apexcovantage.com/. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email us at PLOSMedicine@plos.org.

We expect to receive your revised manuscript by Sep 10 2020 11:59PM. Please email us (plosmedicine@plos.org) if you have any questions or concerns.

***Please note while forming your response, if your article is accepted, you may have the opportunity to make the peer review history publicly available. The record will include editor decision letters (with reviews) and your responses to reviewer comments. If eligible, we will contact you to opt in or out.***

We ask every co-author listed on the manuscript to fill in a contributing author statement, making sure to declare all competing interests. If any of the co-authors have not filled in the statement, we will remind them to do so when the paper is revised. If all statements are not completed in a timely fashion this could hold up the re-review process. If new competing interests are declared later in the revision process, this may also hold up the submission. Should there be a problem getting one of your co-authors to fill in a statement we will be in contact. YOU MUST NOT ADD OR REMOVE AUTHORS UNLESS YOU HAVE ALERTED THE EDITOR HANDLING THE MANUSCRIPT TO THE CHANGE AND THEY SPECIFICALLY HAVE AGREED TO IT. You can see our competing interests policy here: http://journals.plos.org/plosmedicine/s/competing-interests.

Please use the following link to submit the revised manuscript:

https://www.editorialmanager.com/pmedicine/

Your article can be found in the "Submissions Needing Revision" folder.

To enhance the reproducibility of your results, we recommend that you deposit your laboratory protocols in protocols.io, where a protocol can be assigned its own identifier (DOI) such that it can be cited independently in the future. For instructions see http://journals.plos.org/plosmedicine/s/submission-guidelines#loc-methods.

Please ensure that the paper adheres to the PLOS Data Availability Policy (see http://journals.plos.org/plosmedicine/s/data-availability), which requires that all data underlying the study's findings be provided in a repository or as Supporting Information. For data residing with a third party, authors are required to provide instructions with contact information for obtaining the data. PLOS journals do not allow statements supported by "data not shown" or "unpublished results." For such statements, authors must provide supporting data or cite public sources that include it.

We look forward to receiving your revised manuscript.

Sincerely,

Adya Misra, PhD

Senior Editor

PLOS Medicine

plosmedicine.org

-----------------------------------------------------------

Requests from the editors:

Title

Please revise your title according to PLOS Medicine's style. Your title must be nondeclarative and not a question. It should begin with main concept if possible. "Effect of" should be used only if causality can be inferred, i.e., for an RCT. Please place the study design ("A randomized controlled trial," "A retrospective study," "A modelling study," etc.) in the subtitle (ie, after a colon).

Abstract

* Please ensure that all numbers presented in the abstract are present and identical to numbers presented in the main manuscript text. * Please include the study design, population and setting, number of participants, years during which the study took place, length of follow up, and main outcome measures. * Please quantify the main results (with 95% CIs and p values).

Please rephrase “we found little evidence…” to be more specific. The same goes for the sentence in the main conclusions

Limitations of this observational study must be stated more explicitly and please provide 2-3 limitations

“supporting recent policy changes in some countries” is a bit vague and we suggest removing this or rephrasing to be more specific.

Author Summary

At this stage, we ask that you include a short, non-technical Author Summary of your research to make findings accessible to a wide audience that includes both scientists and non-scientists. The Author Summary should immediately follow the Abstract in your revised manuscript. This text is subject to editorial change and should be distinct from the scientific abstract. Please see our author guidelines for more information: https://journals.plos.org/plosmedicine/s/revising-your-manuscript#loc-author-summary

Throughout: please use square brackets for references and these should be placed after all punctuation. Please format the bibliography using Vancouver style

Introduction

I would suggest removing this sentence as it appears to be based on conjecture “All those studies grouped bouts ≥10 mins together, probably because they lacked power to explore more bouts length categories”

Suggest rephrasing this “have used two statistics…”

“Our study is the largest observational study of activity bouts to date and is the first to assess associations of MVPA” please add “to our knowledge” to temper assertions of primacy

Methods

The link in “Analyses were performed in R version 3.5.1, Matlab r2015a or Stata version 15, and all of our analysis code is available at [https://github.com/MRCIEU/UKBActivityBoutLength” is not working, please provide a working link here and in the discussion. This link should also be provided in the data availability statement.

Results

Please provide p values along with 95% CI as needed. Please note all p-values should be exact, unless p<0.001

Discussion

Please temper associations of primacy by adding “to our knowledge” for example in Line 309

Please ensure that the study is reported according to the STROBE guideline, and include the completed STROBE checklist as Supporting Information. When completing the checklist, please use section and paragraph numbers, rather than page numbers. Please add the following statement, or similar, to the Methods: "This study is reported as per the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guideline (S1 Checklist)."

Did your study have a prospective protocol or analysis plan? Please state this (either way) early in the Methods section.

a) If a prospective analysis plan (from your funding proposal, IRB or other ethics committee submission, study protocol, or other planning document written before analyzing the data) was used in designing the study, please include the relevant prospectively written document with your revised manuscript as a Supporting Information file to be published alongside your study, and cite it in the Methods section. A legend for this file should be included at the end of your manuscript.

b) If no such document exists, please make sure that the Methods section transparently describes when analyses were planned, and when/why any data-driven changes to analyses took place.

c) In either case, changes in the analysis-- including those made in response to peer review comments-- should be identified as such in the Methods section of the paper, with rationale.

Comments from the reviewers:

Reviewer #1: To the Authors

You have examined the role of substituting time spent asleep, sedentary, in light and moderate-to-vigorous intensity activity (MVPA) on the associations with mortality, including examining the role of bout length of MVPA (1-15min, 16-40min, and 41+ min bouts). The substitution of time across intensities has been addressed in recent meta-analyses with more deaths and is less novel but the influence of bout length is a matter of much current debate, since some health authorities have recently changed their public health recommendations for MVPA to occur in at least 10-min bouts; it therefore seems rather odd that this paper has defined short bout durations across this divide of 10-min - this is where the uncertainty still lies and the current analysis is a lost opportunity for addressing this issue; it is vital to re-parameterise the bouts to explore the <10-min bout duration range and then compare that to the 10-min and above estimates. The standard time format of derived activity measures in UK Biobank is 5-sec bouts so there is ample opportunity to address this issue.

Confounding by prevalent disease has not been adequately accounted for and the compositional nature of the time-bound data should be ideally complimented with analyses which account for that in a different way.

Specific comments

Number of diseases at pre-baseline is a poor way to account for confounding by prevalent disease; you should use binary indicators of each disease as they have different impact on both activity and mortality. Moreover, as the accelerometer measurement was done about 5 years after the main study baseline, you need to use the routine health data, eg Hospital Episode Statistics which is available in UK Biobank, to reclassify disease status at the accelerometer baseline used in this analysis.

It may be worth pointing out more clearly in the confounding section that you are using age at the time of accelerometer wear (and not at main study baseline).

Seasonal adjustment of accelerometer wear is better achieved using cosine functions as previous Biobank analyses have done.

Did you calibrate the accelerometer readings to local gravity, eg using the approach described by van Hees? Raw data from this sensor is uncalibrated and calibration will greatly reduce noise and can be performed on almost all individual accelerometer files. Your flowchart suggests that you may have performed this but then again maybe not as you do not loose any data in this step.

Your hybrid approach of vector magnitude-based MVPA classification, combined with separation of the non-MVPA component into the other categories by the activity type classification method seems very appealing - do you have any evidence of its validity, though? Note also that the activity classification uses a number of accelerometer features, some of which are orientation-specific; there are two main orientations of the accelerometer measurements in UK Biobank (as two versions of the monitor was used) and possibly an additional actual orientations owing to participants turning the puck inside the wristband upside down or reverse-lateral and wearing it on left or right wrist (all of which makes it challenging to reference accelerometer measurements to the anatomical coordinates). Suggest use version of activity classification without orientation-sensitive signal features.

You collapsed the time-series to 1-min resolution of activity types before any further analysis? That is fine but you should justify this, particularly given your research question here! Note, epoch resolution and bout quantification interacts, see for example paper by Orme et al. Also note that working with 1-min data would mean that your intensity distribution estimates are different to those reported using the derived 5-sec movement intensity data available in the data showcase - again, this is fine but should be highlighted when comparing to other work.

It is fine to use 10-min increments as the unit of time substitution in the analysis of association and your definitions of sedentary bout exposures are fine too. However, you should revisit the categorization of MVPA bout durations for defining these exposures; you must have one category boundary at 10-min (below and above), and it would make sense to have at least a few in the low range where all the variance is! Given you epoch decision and current guideline debate, I would suggest one category should be 1-min, then perhaps 2-9 min, then 10-15min and the remaining ones you already have (although I would probably collapse those if it was me). Your downstream analytical approach requires these to be mutually exclusive but it would actually be better to think about them as nested variables; at least 1-min duration, at least 5-min duration, at least 10-min duration, and so on - please reflect and respond.

You have conducted a time compositional analysis of a bounded phenomenon (24 hours in a day), which makes assumptions about linearity in a constrained variance space; consider supplementing your time substitution analysis with a compositional data analysis or multivariate pattern analysis.

Mortality results have been updated in 2020 and you should therefore rerun your analysis for greater power. Suggest include sensitivity analysis excluding those who died within the first year or two to account for potential reverse causality / confounding by ill-health (I think the 5 years you are mentioning in passing in the discussion is a bit extreme; this has been suggested by those who do not understand regression dilution).

Your findings do NOT directly support the abolition of the 10-min MVPA bout requirement as you have not compared associations for <10-min bout MVPA with those for 10-min or longer bouts!

Previous work on time substitution in Biobank showed broadly similar mortality associations using the self-reported variables - maybe worth comparing your results to?

Reviewer #2: This is a nice study examining associations between physical activity and mortality. The study includes a large sample of UK adults (~80k), measured with accelerometry. This is among the largest studies that include accelerometry in a sample of adults. I do have a few concerns, perhaps the most relevant is the short time of follow-up.

Major:

1 - Participants were assessed in 2013-2015 and have follow-up through 2018. Follow-up time is expected to range from 3-5 years which can be a concern. It is not clear how limited follow-up might have impacted the results but it would be helpful to see sensitivity analyses for England/Wales vs Scotland, which seem to have different follow-up times and also after excluding first 2-3 years of follow-up. It would also help if the authors include a sensitivity analysis excluding individuals with poor health or with major chronic conditions.

2 - The study also includes a rather limited set of confounders. Important confounders such as diet and physical mobility limitations were not included. These are important confounders that can add beyond the ones that were included (eg, BMI). This issue of unmeasured confounding is problematic, particularly given the relatively weak associations that were found for MVPA-mortality.

Minor:

1 - Include follow-up information earlier in the paper (in the all-cause mortality section).

2 - Is there any proxy for mobility limitation that could be used as an additional confounder?

3 - About the MVPA bouts: I was expecting to see bouts of 1-10 min; 11+, etc. The intro (appropriately) address the discussion of 10+ bouts but the categorization used in the paper (15+) is not aligned with this bout criteria.

Also, after looking at the bout distributions, looks like there is not much data for longer bouts. The 1-15min bouts is a pretty long bout - it represents the majority of total MVPA (almost 90% by looking at table S2). This category includes non-bouted and bouted PA (to a max of 15min) given that the lowest resolution is 1-min epochs. Non-bouted and bouted activity should be separated here. It would seem more reasonable to use 1-4, 5-9, 10+, etc. or 1-9, 10+. Also, were interruptions allowed when creating these bouts?

The longer bouts (16min +) is likely skewed and makes me doubt there is enough data here to test these associations. How many people reported PA in each of the bout levels? Showing these distributions and amount of '0s" will tell how much data there is for this particular analysis.

Finally, still related to bouts - looks there is pretty much no data on 41+ bouts of PA. Pease show the distribution for this as well.

4 - Include an explanation for using age as time metric as oppose to time of follow-up. I wondered if truncating birth at July 1st results in losing the benefit of using age as time metric.

5- Please include a Table 1 with information on demographics and follow-up time. This in the suppl material but it would be helpful to see this in the main paper. Also, Table S1 shows there were 538 events which conflicts with ~700 reported in the main text. Please clarify.

6 - In the discussion section, references in support of excluding the bout criteria (as per old guidelines recommendation) can't be supported here as bouts were not generated to test this assumption. See earlier comment and refine bout classifications.

7 - The following statement shows up a couple times in the paper but its not accurate:

"There are no previous studies assessing the association of MVPA bout length with mortality"

Here is a study that should be cited/discussed for this purpose:

Moderate‐to‐Vigorous Physical Activity and All‐Cause Mortality: Do Bouts Matter?

Saint-Maurice PF et al. JAHA 2018

8 - I mentioned this before but suggest that the discussion on short follow-up be expanded as a critical limitation. I also suggest additional sensitivity analyses. Eg, excluding people with chronic conditions.

The following paper might be helpful for this purpose:

https://pubmed.ncbi.nlm.nih.gov/32472927/

Given the weak associations with MVPA, a small degree of confounding is likely to attenuate these effects to become null. This should be part of the discussion here as well.

Reviewer #3: I confine my remarks to statistical aspects of this paper. These were very well done, but I have a few questions and comments before I can recommend publication.

First, I wish more of the technical details were in the main paper, but I suppose that is a matter for the editors; it isn't a mistake to put it in a supplement.

Line 160-161 I am curious about how sleep was distinguished from sedentary activity. Even if the technical details are given in an appendix, I think a plain language summary of this would be useful; after all, people do move in their sleep and some people lie still (but awake) for some time before sleeping. In fact, a plain language summary of all the levels would be useful.

Line 165-168

1. What was the "gold standard"? That is, you say some periods were misclassified. But how was the right classification determined? And, if the right category could be determined, why wasn't it used? It seems to me there are going to be some murky periods no matter what you do. (e.g. time spent just before sleep; time spent going to the bathroom in the middle of the night; activity that is on the border between moderate and light; and so on.

2. Given that you state some of the misclassification (58% of MPVA) I'd suggest giving all the results, for both models.

Also, as a statistician, I'm just curious why three different programs were used --- it's fine to do so and all the programs are good ones, I'm just wondering

Overall a very interesting analysis!

Peter Flom

Any attachments provided with reviews can be seen via the following link:

[LINK]

PLoS Med. 2021 Sep 15;18(9):e1003757. doi: 10.1371/journal.pmed.1003757.r003

Author response to Decision Letter 1

29 Oct 2020

Attachment

Submitted filename: reviewer-responses20201028.docx

Click here for additional data file.^{(64.9KB, docx)}

PLoS Med. doi: 10.1371/journal.pmed.1003757.r004

Decision Letter 2

Emma Veitch

3 Dec 2020

Dear Dr. Millard,

Thank you very much for submitting your revised manuscript "Association of changing physical activity intensity and bout length with mortality: a prospective study of 79,503 participants in UK Biobank" (PMEDICINE-D-20-02260R2) for consideration at PLOS Medicine.

The revisions were seen again by the previous reviewers, whose comments are enclosed below; I hope you find them constructive. Two now recommend acceptance but one reviewer (r#1), although acknowledging the improvements made, does still ask for further clarifications on the methods and reporting. At this stage we'd ask that you respond to those points in a further revision before we can make a final decision on the paper. Please note this reviewer says there are some errors in internal links in the document; I checked this and although these ("error!" links) appear in the compiled pdf, they seem to be OK in the submitted Word file. The editors are fine for the authors to include a mention that preplanned analyses were modified due to input obtained during peer review.

The reviews are appended at the bottom of this email and any accompanying reviewer attachments can be seen via the link below:

[LINK]

We expect to receive your revised manuscript by Dec 24 2020 11:59PM. Please email us (plosmedicine@plos.org) if you have any questions or concerns.

Please use the following link to submit the revised manuscript:

https://www.editorialmanager.com/pmedicine/

Your article can be found in the "Submissions Needing Revision" folder.

To enhance the reproducibility of your results, we recommend that you deposit your laboratory protocols in protocols.io, where a protocol can be assigned its own identifier (DOI) such that it can be cited independently in the future. For instructions see http://journals.plos.org/plosmedicine/s/submission-guidelines#loc-methods.

We look forward to receiving your revised manuscript.

Sincerely,

Emma Veitch, PhD

PLOS Medicine

On behalf of Adya Misra, PhD, Senior Editor,

PLOS Medicine

plosmedicine.org

-----------------------------------------------------------

Requests from the editors:

*We'd recommend that the data availability statement provided in the article submission system is updated to include the github links for code availability (provided in the authors' response and main text).

-----------------------------------------------------------

Comments from the reviewers:

Reviewer #1:

The revisions have improved this paper substantially, particularly the update of follow-up and the re-specification of bouts around the currently discussed bout duration boundary of 10 minutes. There are, however, a few outstanding concerns remaining as outlined below.

Title: You have not assessed change in physical activity but between-individual differences - change title accordingly (eg, delete "changing").

Abstract:

It is not a trial so your conclusion should be that activity is associated with lower mortality, not "reduced" mortality. Same issue with "increased" mortality risk - it should read "higher". Make global check for causal language.

Introduction:

There are other papers from the cohort reporting on the general association between accelerometry-measured exposure and mortality. For example, a recent paper by Strain et al (Nature Medicine, 2020) reported on the associations between accelerometry-measured activity and mortality using the same dataset but did not look at the potential importance of bouts - suggest reword intro slightly to leap from there.

Note that the recently launched WHO guidelines for physical activity have also lost the 10-min bout requirement, and there is increased recognition of light activity too - may allow you a slightly elevated pitch to refer to global guidelines rather than just US/UK national?

Methods:

Invitation to join the accelerometry subsample was not random; everyone with a valid email in the cohort were invited, except those currently taking part in substudies.

Thanks for confirming that you have calibrated your accelerometer data to gravity; they would otherwise not have been valid and we could stop right there! Please include appropriate citation for the method - the Biobank tool uses the van Hees method (with temperature correction) and other tools use either that or the Lukowic method, both of which are sufficient but you do have to use one.

To clarify the point about accelerometer axis orientation: All accelerometers used were Axivity AX3 but there are two versions of that. The first batch of accelerometers used in UKB had the X and the Y axes swopped around in the casing, compared to subsequent batches used. There are no other differences as the chip is the same; it simply means a rotation of 90 degrees on the wrist mount. The meta data of the .cwa files includes this information through its serial numbers but you can also see it in the data as the axis angles (pitch and roll) with gravity (low-frequency component in the data) differ between the two batches. The activity classification work by Willett et al (Sci Rep 2018 - the correct citation for the method used in Doherty et al, 2018) used the orientation of the X and Y axes of the more recent batch, so is directly applicable to that part of the UKB data but not directly to the data collected with the first batch of monitors. Although only some of the several signal features used in the activity classification method are orientation-sensitive, the optimal application of the method requires remapping X acceleration to Y and vice versa as an initial step for those who were measured with the earliest version of the accelerometer.

The splitting of the MVPA exposure category around the 10-min boundary is excellent and will make the results of this analysis far more relevant to the current debate on the issue. I have to say, the other categories do not seem entirely logical to me, eg why split MVPA at 40 minutes and not 30 minutes? And why not split the sub-10 category further, eg at 5-min, as that is where all the behavioural variance is (and one notes that exact example of 5-min bouts in your summary statement)? And why insist on alignment of sedentary and MVPA bout durations when the notion of short and long are materially different for such behaviours?

I don`t think you need to keep saying "at the request of the Reviewers" in the Methods as you have already stated it when describing deviations from your analysis plan but I refer to the Editor here for specific journal policy.

With respect to the approach of defining exposure as nested vs mutually exclusive, I appreciate the opportunity to do the reflections for you since you ask; it is certainly not wrong to conduct a mutually exclusive categorization analysis as you have done but given that the recent debate was not about which bout duration activity should have (short or long) to confer health benefits but rather if it should have a minimum duration of 10min or not, the analysis that more directly addresses that issue is one that contrast associations of exposures with and one without a minimum bout duration definition (and not a maximum one).

Results:

Main table 1 doubles up as a drop-out analysis - fine although I would have preferred a clean main table of the data in the second column and then this one in supplement.

I see no more main tables? Are supplementary table S3a and S4 not the most essential set of results? Or do figure 2 and 3 serve this function as main results? We are not helped by the formatting issue here - there is a dead-end bookmark in my review copy.

Discussion:

Suggest delete "uniquely" from opening statement - it reads very odd and probably isn`t even true unless you caveat it somehow but that would make it worse.

Suggest make reference to global public health recommendations by the WHO instead - seems more convincing. It is not true that the removal of 10-min MVPA bout duration requirements from US and UK guidelines were purely based on trial evidence - observational epidemiological evidence such as the type you report here was also considered, including some of the papers you cite here yourself - please rewrite.

Your points about the limitations of self-report are well made - in fact so well made that you should include them in your paper. The fact remains though that the overall conclusions about time in different intensities do align well with previous reports from the full 500k participants in the cohort with longer follow-up.

With respect to the proposed text edits on the accelerometer orientation issue, it is wrong to blame this on the participants as those in the early batch most likely did what they were told to do with that version of the monitor (so they did not wear it incorrectly) - please rephrase, including making the point that your hybrid approach of using vector-magnitude-based MVPA mitigates against this being a major issue for most of your main analysis (as vector magnitude is not orientation-specific).

"… any true effect of activity on mortality being short-lived" - I can`t quite decide whether this is an unfortunate phrasing of words or rather smart or funny but if this was unintentional, you should probably rephrase it!

An issue which requires further discussion is the fact that 1-min epoch data are 1-min bouts - maybe just say exactly that where you mention that it might change if you had used another epoch setting which is a bit vague.

Also, whilst I agree that some previous analyses have not accounted for overall time in sedentary or MVPA, you have not accounted for intrinsic differences in the intensity of MVPA between bouts; it is perfectly possible that those who go for longer MVPA bouts are also those who run faster when doing MVPA! In the recent Strain paper from the cohort, total volume was a main driver of the association with mortality, with additional benefit suggested from increased fraction of that volume coming from MVPA. Your analysis does not account for volume differences and intensity differences within exposure categories (which also makes your cutpoint decision far more influential) but the previous analysis does - a paragraph to discuss this issue should suffice.

General formatting errors throughout for references to within-document items, eg figs/tables.

-----------------------------------------------------------

Reviewer #2: I appreciate the authors effort to address my concerns. Well done!

-----------------------------------------------------------

Reviewer #3: The authors have addressed my concerns and I now recommend publication

Peter Flom

-----------------------------------------------------------

Any attachments provided with reviews can be seen via the following link:

[LINK]

PLoS Med. 2021 Sep 15;18(9):e1003757. doi: 10.1371/journal.pmed.1003757.r005

Author response to Decision Letter 2

2 Jan 2021

Attachment

Submitted filename: reviewer-responses-20201218.docx

Click here for additional data file.^{(55.9KB, docx)}

PLoS Med. doi: 10.1371/journal.pmed.1003757.r006

Decision Letter 3

Richard Turner

15 Jul 2021

Dear Dr. Millard,

Thank you very much for re-submitting your manuscript "Association of physical activity intensity and bout length with mortality: a prospective study of 79,503 participants in UK Biobank" (PMEDICINE-D-20-02260R3) for consideration at PLOS Medicine. We do apologize for the long delay in our response.

I have discussed the paper with editorial colleagues and our academic editor, and I am pleased to tell you that, once the remaining editorial and production issues are fully dealt with, we expect to be able to accept the paper for publication in the journal.

The remaining issues that need to be addressed are listed at the end of this email. Any accompanying reviewer attachments can be seen via the link below. Please take these into account before resubmitting your manuscript:

[LINK]

In revising the manuscript for further consideration here, please ensure you address the specific points made by each reviewer and the editors. In your rebuttal letter you should indicate your response to the reviewers' and editors' comments and the changes you have made in the manuscript. Please submit a clean version of the paper as the main article file. A version with changes marked must also be uploaded as a marked up manuscript file.

Please also check the guidelines for revised papers at http://journals.plos.org/plosmedicine/s/revising-your-manuscript for any that apply to your paper. If you haven't already, we ask that you provide a short, non-technical Author Summary of your research to make findings accessible to a wide audience that includes both scientists and non-scientists. The Author Summary should immediately follow the Abstract in your revised manuscript. This text is subject to editorial change and should be distinct from the scientific abstract.

We hope to receive your revised manuscript within 1 week. Please email us (plosmedicine@plos.org) if you have any questions or concerns.

We ask every co-author listed on the manuscript to fill in a contributing author statement. If any of the co-authors have not filled in the statement, we will remind them to do so when the paper is revised. If all statements are not completed in a timely fashion this could hold up the re-review process. Should there be a problem getting one of your co-authors to fill in a statement we will be in contact. YOU MUST NOT ADD OR REMOVE AUTHORS UNLESS YOU HAVE ALERTED THE EDITOR HANDLING THE MANUSCRIPT TO THE CHANGE AND THEY SPECIFICALLY HAVE AGREED TO IT.

To enhance the reproducibility of your results, we recommend that you deposit your laboratory protocols in protocols.io, where a protocol can be assigned its own identifier (DOI) such that it can be cited independently in the future. Additionally, PLOS ONE offers an option to publish peer-reviewed clinical study protocols. Read more information on sharing protocols at https://plos.org/protocols?utm_medium=editorial-email&utm_source=authorletters&utm_campaign=protocols

Please review your reference list to ensure that it is complete and correct. If you have cited papers that have been retracted, please include the rationale for doing so in the manuscript text, or remove these references and replace them with relevant current references. Any changes to the reference list should be mentioned in the rebuttal letter that accompanies your revised manuscript.

Please note, when your manuscript is accepted, an uncorrected proof of your manuscript will be published online ahead of the final version, unless you've already opted out via the online submission form. If, for any reason, you do not want an earlier version of your manuscript published online or are unsure if you have already indicated as such, please let the journal staff know immediately at plosmedicine@plos.org.

Please let me know if you have any questions, and we look forward to receiving the revised manuscript.

Sincerely,

Richard Turner PhD

Senior Editor, PLOS Medicine

rturner@plos.org

------------------------------------------------------------

Requests from Editors:

Please adapt the data statement to: "... available from UK Biobank for researchers who meet the criteria for access to confidential data ..." or similar.

Please remove the word "prospective" from the title (we think that your study is a retrospective analysis of prospectively gathered data). So as to comply with journal style we suggest: "Physical activity intensity and bout length and mortality in 79,503 UK Biobank participants: a population-based cohort study".

At line 22, should that be "... less light activity ..."?

At line 24 and elsewhere, we suggest making that "sedentary time".

In the abstract, we suggest quoting the 95% CI and p values immediately after the point estimates, e.g., "... (HR 1.02 [95% CI 1.01 to 1.02, p<0.001] per 10 minutes more sedentary time ...)".

At line 28, we suggest removing "UK Biobank had a 5% recruitment response", as this can be mentioned in the main text.

At line 32 please make that "was associated" as you are reporting findings, with the surrounding wording adapted to be consistent with this.

At line 82, for example, please adapt the reference call-out to the style: "... amounts alone [4,5]." (noting the absence of spaces within the square bracket).

Please restructure the paragraph beginning at line 109, where the elements of discussion should be removed or transferred to the Discussion section. We anticipate 1-2 sentences stating the aim and, to an extent, the methodology of your study. The sentence beginning at line 118 should be removed.

You mention an analysis plan at line 124. Please include this as an attachment if available, referred to in the text.

At line 164 and elsewhere as appropriate, please use the style "follow-up".

We ask you to adopt more circumspect language in places. For example, at line 385 to avoid over-generalizing we suggest amending the wording to: "In this study, we found that time spent in MVPA was associated with lower mortality ..." or similar (with the tenses adjusted to match in the rest of the paragraph).

Please use the journal name abbreviation "PLoS ONE" in the reference list.

Please add "[preprint]" to any preprints cited, reference 29 perhaps being one.

Please remove the information on funding and the UK Biobank application from the end of the text. The former will appear in the article metadata in the event of publication, via entries in the submission form, and the latter can be moved to the Methods section.

***

PLoS Med. 2021 Sep 15;18(9):e1003757. doi: 10.1371/journal.pmed.1003757.r007

Author response to Decision Letter 3

21 Jul 2021

Attachment

Submitted filename: reviewer-responses-20210721.docx

Click here for additional data file.^{(27KB, docx)}

PLoS Med. doi: 10.1371/journal.pmed.1003757.r008

Decision Letter 4

Richard Turner

27 Jul 2021

Dear Dr. Millard,

Thank you very much for re-submitting your manuscript "Association of physical activity intensity and bout length with mortality: an observational study of 79,503 UK Biobank participants" (PMEDICINE-D-20-02260R4) for consideration at PLOS Medicine.

Following further discussions among the editors, we will need to ask you to respond to some minor comments from our academic editor before we are in a position to proceed further. The remaining issues that need to be addressed are listed at the end of this email: please take these into account before resubmitting your manuscript.

We hope to receive your revised manuscript within 1 week. Please email us (plosmedicine@plos.org) if you have any questions or concerns.

To enhance the reproducibility of your results, we recommend that you deposit your laboratory protocols in protocols.io, where a protocol can be assigned its own identifier (DOI) such that it can be cited independently in the future. Additionally, PLOS ONE offers an option to publish peer-reviewed clinical study protocols. Read more information on sharing protocols at https://plos.org/protocols?utm_medium=editorial-email&utm_source=authorletters&utm_campaign=protocols

Please let me know if you have any questions, and we look forward to receiving the revised manuscript.

Sincerely,

Richard Turner, PhD

Senior Editor, PLOS Medicine

rturner@plos.org

------------------------------------------------------------

Requests from Editors:

Does "income (pounds)" need to be specified in table 1?

Comments from Academic editor:

I note an issue with supplementary figures S10-12 which show the compositional analysis results (judged by some to be the correct way to examine the main question posed in this paper, so fairly essential). Specifically, the authors need to explain what the two curves in each colour or panel represents, and secondly comment on what the results actually show; the all-too-brief statement in the manuscript says that that these analyses show the same as the main results but in some cases they appear to show the exact opposite or indeed very complex U-shaped associations! This needs a proper description and a proper discussion.

A minor point is to sharpen up and strengthen the statement in the summary of “what do these findings mean”: The authors have analysed the accelerometer data in 1-min time resolution, and so the shortest bout of any activity category is 1 minute. Although it is not directly reported anywhere (but perhaps should be), we can deduce that the most common “short” MVPA bout is very close to 1 minute, given that the median time spent in total MVPA is 92 min per day and the median time spent in short MVPA bouts is 72 min per day. Hence, the example of 5-min bout duration for MVPA is a bit odd and probably wrong or at least sending the wrong message; what we have learned from this analysis is that MVPA bouts as short as 1 minute in duration have benefits for human health – a more accurate and more powerful message, providing very strong support for the most recent health recommendations.

***

PLoS Med. 2021 Sep 15;18(9):e1003757. doi: 10.1371/journal.pmed.1003757.r009

Author response to Decision Letter 4

4 Aug 2021

Attachment

Submitted filename: reviewer-responses-20210803.docx

Click here for additional data file.^{(25.9KB, docx)}

PLoS Med. doi: 10.1371/journal.pmed.1003757.r010

Decision Letter 5

Richard Turner

5 Aug 2021

Dear Dr Millard,

On behalf of my colleagues and the Academic Editor, Dr Brage, I am pleased to inform you that we have agreed to publish your manuscript "Association of physical activity intensity and bout length with mortality: an observational study of 79,503 UK Biobank participants" (PMEDICINE-D-20-02260R5) in PLOS Medicine.

Before your manuscript can be formally accepted you will need to complete some formatting changes, which you will receive in a follow up email. Please be aware that it may take several days for you to receive this email; during this time no action is required by you. Once you have received these formatting requests, please note that your manuscript will not be scheduled for publication until you have made the required changes.

In the meantime, please log into Editorial Manager at http://www.editorialmanager.com/pmedicine/, click the "Update My Information" link at the top of the page, and update your user information to ensure an efficient production process.

PRESS

We frequently collaborate with press offices. If your institution or institutions have a press office, please notify them about your upcoming paper at this point, to enable them to help maximise its impact. If the press office is planning to promote your findings, we would be grateful if they could coordinate with medicinepress@plos.org. If you have not yet opted out of the early version process, we ask that you notify us immediately of any press plans so that we may do so on your behalf.

We also ask that you take this opportunity to read our Embargo Policy regarding the discussion, promotion and media coverage of work that is yet to be published by PLOS. As your manuscript is not yet published, it is bound by the conditions of our Embargo Policy. Please be aware that this policy is in place both to ensure that any press coverage of your article is fully substantiated and to provide a direct link between such coverage and the published work. For full details of our Embargo Policy, please visit http://www.plos.org/about/media-inquiries/embargo-policy/.

To enhance the reproducibility of your results, we recommend that you deposit your laboratory protocols in protocols.io, where a protocol can be assigned its own identifier (DOI) such that it can be cited independently in the future. Additionally, PLOS ONE offers an option to publish peer-reviewed clinical study protocols. Read more information on sharing protocols at https://plos.org/protocols?utm_medium=editorial-email&utm_source=authorletters&utm_campaign=protocols

Thank you again for submitting to PLOS Medicine. We look forward to publishing your paper.

Sincerely,

Richard Turner, PhD

Senior Editor, PLOS Medicine

rturner@plos.org

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

S1 STROBE Checklist. STROBE, STrengthening the Reporting of OBservational studies in Epidemiology.

(DOC)

Click here for additional data file.^{(89KB, doc)}

S1 Fig. Directed acyclic graph illustrating path between bout length and survival through a common cause of bout length and total time spent sedentary.

(PDF)

Click here for additional data file.^{(59.8KB, pdf)}

S2 Fig. Directed acyclic graph illustrating hypothesised confounding factors and potential mediators.

(PDF)

Click here for additional data file.^{(78.9KB, pdf)}

S3 Fig. Distributions of the number of valid days of accelerometer data for participants included in our samples.

(PDF)

Click here for additional data file.^{(49.5KB, pdf)}

(PDF)

Click here for additional data file.^{(88.5KB, pdf)}

(PDF)

Click here for additional data file.^{(81.2KB, pdf)}

S6 Fig. Results of sensitivity analysis starting follow-up 1 and 2 years after accelerometer wear.

(PDF)

Click here for additional data file.^{(72.1KB, pdf)}

S7 Fig. Results of sensitivity analysis using ML-only approach: Association of time spent in MVPA and sedentary bouts of a given length, with all-cause mortality.

MVPA, moderate-vigorous physical activity.

(PDF)

Click here for additional data file.^{(114KB, pdf)}

S8 Fig. Results of sensitivity analysis starting follow-up 1 and 2 years after accelerometer wear: Association of time spent in MVPA and sedentary bouts of a given length, with all-cause mortality.

MVPA, moderate-vigorous physical activity.

(PDF)

Click here for additional data file.^{(102KB, pdf)}

S9 Fig. Association of time spent in MVPA and sedentary bouts of a given length, with all-cause mortality using hybrid approach and “other day” imputed data.

MVPA, moderate-vigorous physical activity.

(PDF)

Click here for additional data file.^{(87.1KB, pdf)}

S10 Fig. Results of sensitivity analysis using isometric log ratio transformed activity variables: Association of time spent in activity categories, with all-cause mortality.

(PDF)

Click here for additional data file.^{(74.3KB, pdf)}

S11 Fig. Results of sensitivity analysis using isometric log ratio transformed activity variables: Association of time spent in MVPA bouts of a given length, with all-cause mortality.

MVPA, moderate-vigorous physical activity.

(PDF)

Click here for additional data file.^{(108.4KB, pdf)}

S12 Fig. Results of sensitivity analysis using isometric log ratio transformed activity variables: Association of time spent in sedentary bouts of a given length, with all-cause mortality.

(PDF)

Click here for additional data file.^{(76.9KB, pdf)}

S13 Fig. Directed acyclic graph illustrating the potential for collider bias due to selection into the study sample.

(PDF)

Click here for additional data file.^{(59.3KB, pdf)}

S1 Table. Summary statistics of the UK Biobank participants who are in our sample, compared with those who were invited to wear an accelerometer but are not in our sample, for complete days version.

(DOCX)

Click here for additional data file.^{(43KB, docx)}

S2 Table. Summary of time spent in activity classifications on average per day.

(DOCX)

Click here for additional data file.^{(41.3KB, docx)}

S3 Table. Correlations between activity summary variables of main analysis (hybrid approach) and ML-only sensitivity approach.

(DOCX)

Click here for additional data file.^{(52.9KB, docx)}

S4 Table. Associations of transferring time between overall activity categories, with all-cause mortality.

(DOCX)

Click here for additional data file.^{(56.1KB, docx)}

S5 Table. Associations of transferring time from MVPA bout length category to other activity category, with all-cause mortality.

MVPA, moderate-vigorous physical activity.

(DOCX)

Click here for additional data file.^{(60.2KB, docx)}

S6 Table. Associations of transferring time from sedentary bout length categories to other activity category, with all-cause mortality.

(DOCX)

Click here for additional data file.^{(49.5KB, docx)}

S1 Text

(DOCX)

Click here for additional data file.^{(58KB, docx)}

Attachment

Submitted filename: reviewer-responses20201028.docx

Click here for additional data file.^{(64.9KB, docx)}

Attachment

Submitted filename: reviewer-responses-20201218.docx

Click here for additional data file.^{(55.9KB, docx)}

Attachment

Submitted filename: reviewer-responses-20210721.docx

Click here for additional data file.^{(27KB, docx)}

Attachment

Submitted filename: reviewer-responses-20210803.docx

Click here for additional data file.^{(25.9KB, docx)}

Data Availability Statement

[pmed.1003757.ref001] 1.Lee I-M, Shiroma EJ, Lobelo F, Puska P, Blair SN, Katzmarzyk PT, et al. Impact of Physical Inactivity on the World’s Major Non-Communicable Diseases. Lancet. 2012;380:219–29. doi: 10.1016/S0140-6736(12)61031-9 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pmed.1003757.ref002] 2.Ekelund U, Tarp J, Steene-Johannessen J, Hansen BH, Jefferis B, Fagerland MW, 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]

[pmed.1003757.ref003] 3.Physical activity guidelines: UK Chief Medical Officers’ report. 2019. Available from: https://www.gov.uk/government/publications/physical-activity-guidelines-uk-chief-medical-officers-report.

[pmed.1003757.ref004] 4.Piercy KL, Troiano RP, Ballard RM, Carlson SA, Fulton JE, Galuska DA, et al. The Physical Activity Guidelines for Americans. JAMA. 2018;320:2020–8. doi: 10.1001/jama.2018.14854 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pmed.1003757.ref005] 5.Bull FC, Al-Ansari SS, Biddle S, Borodulin K, Buman MP, Cardon G, et al. World Health Organization 2020 guidelines on physical activity and sedentary behaviour. Br J Sports Med. 2020;54:1451–62. doi: 10.1136/bjsports-2020-102955 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pmed.1003757.ref006] 6.Jakicic JM, Kraus WE, Powell KE, Campbell WW, Janz KF, Troiano RP, et al. Association between Bout Duration of Physical Activity and Health: Systematic Review. Med Sci Sports Exerc. 2019;51:1213–9. doi: 10.1249/MSS.0000000000001933 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pmed.1003757.ref007] 7.Tarp J, Child A, Whitet T, Westgate K, Bugge A, Grøntved A, et al. Physical activity intensity, bout-duration, and cardiometabolic risk markers in children and adolescents. Int J Obes (Lond). 2018;42:1639–50. doi: 10.1038/s41366-018-0152-8 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pmed.1003757.ref008] 8.Glazer NL, Lyass A, Esliger DW, Blease SJ, Freedson PS, Massaro JM, et al. Sustained and shorter bouts of physical activity are related to cardiovascular health. Med Sci Sports Exerc. 2013;45:109–15. doi: 10.1249/MSS.0b013e31826beae5 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pmed.1003757.ref009] 9.Saint-Maurice PF, Troiano RP, Matthews CE, Kraus WE. Moderate-to-Vigorous Physical Activity and All-Cause Mortality: Do Bouts Matter? J Am Heart Assoc. 2018;7:e007678. doi: 10.1161/JAHA.117.007678 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pmed.1003757.ref010] 10.Strath SJ, Holleman RG, Ronis DL, Swartz AM, Richardson CR. Objective physical activity accumulation in bouts and nonbouts and relation to markers of obesity in US adults. Prev Chronic Dis. 2008;5:A131. [PMC free article] [PubMed] [Google Scholar]

[pmed.1003757.ref011] 11.Brooke HL, Atkin AJ, Corder K, Brage S, van Sluijs EM. Frequency and duration of physical activity bouts in school-aged children: A comparison within and between days. Prev Med Rep. 2016;4:585–90. doi: 10.1016/j.pmedr.2016.10.007 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pmed.1003757.ref012] 12.Masse L, Fuemmeler B, Anderson C, Matthews CE, Trost SG, Catellier DJ, et al. Accelerometer data reduction: a comparison of four reduction algorithms on select outcome variables. Med Sci. 2005;37:S544–54. doi: 10.1249/01.mss.0000185674.09066.8a [DOI] [PubMed] [Google Scholar]

[pmed.1003757.ref013] 13.Rowlands A, Pilgrim E, Eston R. Patterns of habitual activity across weekdays and weekend days in 9–11-year-old children. Prev Med. 2008;46:317–24. doi: 10.1016/j.ypmed.2007.11.004 [DOI] [PubMed] [Google Scholar]

[pmed.1003757.ref014] 14.Trost S, Pate R, Sallis J, Freedson P, Taylor WC, Dowda M, et al. Age and gender differences in objectively measured physical activity in youth. Med Sci. 2002;34:350–5. doi: 10.1097/00005768-200202000-00025 [DOI] [PubMed] [Google Scholar]

[pmed.1003757.ref015] 15.Riddoch CJ, Mattocks C, Deere K, Saunders J, Kirkby J, Tilling K, et al. Objective measurement of levels and patterns of physical activity. Arch Dis Child. 2007;92:963–9. doi: 10.1136/adc.2006.112136 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pmed.1003757.ref016] 16.Diaz KM, Howard VJ, Hutto B, Colabianchi N, Vena JE, Safford MM, et al. Patterns of Sedentary Behavior and Mortality in U.S. Middle-Aged and Older Adults: A National Cohort Study. Ann Intern Med. 2017;167:465–75. doi: 10.7326/M17-0212 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pmed.1003757.ref017] 17.Fry A, Littlejohns TJ, Sudlow C, Doherty N, Adamska L, Sprosen T, et al. Comparison of Sociodemographic and Health-Related Characteristics of UK Biobank Participants With Those of the General Population. Am J Epidemiol. 2017;186:1026–34. doi: 10.1093/aje/kwx246 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pmed.1003757.ref018] 18.Sudlow C, Gallacher J, Allen N, Beral V, Burton P, Danesh J, et al. UK Biobank: An Open Access Resource for Identifying the Causes of a Wide Range of Complex Diseases of Middle and Old Age. PLoS Med. 2015;12:e1001779. doi: 10.1371/journal.pmed.1001779 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pmed.1003757.ref019] 19.Doherty A, Jackson D, Hammerla N, Plötz T, Olivier P, Granat MH, et al. Large Scale Population Assessment of Physical Activity Using Wrist Worn Accelerometers: The UK Biobank Study. PLoS ONE. 2017;12:e0169649. doi: 10.1371/journal.pone.0169649 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pmed.1003757.ref020] 20.Stolwijk AM, Straatman H, Zielhuis GA. Studying seasonality by using sine and cosine functions in regression analysis. J Epidemiol Community Health. 1999;53:235–8. doi: 10.1136/jech.53.4.235 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pmed.1003757.ref021] 21.Doherty A, Smith-Byrne K, Ferreira T, Holmes MV, Holmes C, Pulit SL, et al. GWAS identifies 14 loci for device-measured physical activity and sleep duration. Nat Commun. 2018;9:5257. doi: 10.1038/s41467-018-07743-4 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pmed.1003757.ref022] 22.Willetts M, Hollowell S, Aslett L, Holmes C, Doherty A. Statistical machine learning of sleep and physical activity phenotypes from sensor data in 96,220 UK Biobank participants. Sci Rep. 2018;8:7961. doi: 10.1038/s41598-018-26174-1 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pmed.1003757.ref023] 23.van Hees VT, Fang Z, Langford J, Assah F, Mohammad A, da Silva ICM, et al. Autocalibration of accelerometer data for free-living physical activity assessment using local gravity and temperature: an evaluation on four continents. J Appl Physiol. 2014;117:738–44. doi: 10.1152/japplphysiol.00421.2014 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pmed.1003757.ref024] 24.Ainsworth BE, Haskell WL, Herrmann SD, Meckes N, Bassett DRJR, Tudor-Locke C, et al. 2011 Compendium of Physical Activities: A Second Update of Codes and MET Values. Med Sci Sport Exerc. 2011;43:1575–81. doi: 10.1249/MSS.0b013e31821ece12 [DOI] [PubMed] [Google Scholar]

[pmed.1003757.ref025] 25.Bell JA, Hamer M, van Hees VT, Singh-Manoux A, Kivimaki M, Sabia S. Healthy obesity and objective physical activity. Am J Clin Nutr. 2015;102:268–75. doi: 10.3945/ajcn.115.110924 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pmed.1003757.ref026] 26.Groenwold RHH, Palmer TM, Tilling K. To adjust or not to adjust? When a “confounder” is only measured after exposure. Epidemiology. 2021;32:194–201. doi: 10.1097/EDE.0000000000001312 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pmed.1003757.ref027] 27.Millard LAC, Tilling K, Lawlor DA, Flach PA, Gaunt TR. Physical activity phenotyping with activity bigrams, and their association with BMI. Int J Epidemiol. 2017;46. doi: 10.1093/ije/dyx093 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pmed.1003757.ref028] 28.McGregor DE, Palarea-Albaladejo J, Dall PM, Hron K, Chastin S. Cox regression survival analysis with compositional covariates: Application to modelling mortality risk from 24-h physical activity patterns. Stat Methods Med Res. 2020;29:1447–65. doi: 10.1177/0962280219864125 [DOI] [PubMed] [Google Scholar]

[pmed.1003757.ref029] 29.Orme M, Wijndaele K, Sharp SJ, Westgate K, Ekelund U, Brage S. Combined influence of epoch length, cut-point and bout duration on accelerometry-derived physical activity. Int J Behav Nutr Phys Act. 2014;11:34. doi: 10.1186/1479-5868-11-34 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pmed.1003757.ref030] 30.Groenwold RHH, Sterne JAC, Lawlor DA, Moons KGM, Hoes AW, Tilling K. Sensitivity analysis for the effects of multiple unmeasured confounders. Ann Epidemiol. 2016;26:605–11. doi: 10.1016/j.annepidem.2016.07.009 [DOI] [PubMed] [Google Scholar]

[pmed.1003757.ref031] 31.Strain T, Wijndaele K, Dempsey PC, Sharp SJ, Pearce M, Jeon J, et al. Wearable-device-measured physical activity and future health risk. Nat Med. 2020;26:1385–91. doi: 10.1038/s41591-020-1012-3 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pmed.1003757.ref032] 32.Swanson JM. The UK Biobank and selection bias. Lancet. 2012;380:110. doi: 10.1016/S0140-6736(12)61179-9 [DOI] [PubMed] [Google Scholar]

PERMALINK

Association of physical activity intensity and bout length with mortality: An observational study of 79,503 UK Biobank participants

Louise A C Millard

Kate Tilling

Tom R Gaunt

David Carslake

Deborah A Lawlor

Roles

Abstract

Background

Methods and findings

Conclusions

Author summary

Why was this study done?

What did the researchers do and find?

What do these findings mean?

Introduction

Methods

Participants

Fig 1. Participant flow diagram.

Data collection

Physical activity measurement

All-cause mortality

Potential confounders

Accelerometer data preprocessing

Statistical analyses

Dealing with missing accelerometer data

Deriving physical activity bouts

Deriving summary variables reflecting time spent in activity categories, overall, and in bout length strata

Estimating the association of overall time spent in activity categories with all-cause mortality taking account of total time spent in that activity and coupling of spending more time in one activity category with less time in another category

Relating time spent in short, medium, and long bouts of MVPA or sedentary behaviour with all-cause mortality

Results

Table 1. Summary statistics of the UK Biobank participants in our sample.

Associations of overall time spent in activity categories, with all-cause mortality

Fig 2. Associations of less time spent in baseline activity category coupled with more time in comparison category, with all-cause mortality.

Associations of MVPA and sedentary bout length with all-cause mortality

Fig 3. Associations of time spent in MVPA and sedentary bouts of given duration, with all-cause mortality.

Discussion

Study strengths and limitations

Supporting information

Acknowledgments

Abbreviations

Data Availability

Funding Statement

References

Decision Letter 0

Adya Misra

Roles

Decision Letter 1

Adya Misra

Roles

Author response to Decision Letter 1

Decision Letter 2

Emma Veitch

Roles

Author response to Decision Letter 2

Decision Letter 3

Richard Turner

Roles

Author response to Decision Letter 3

Decision Letter 4

Richard Turner

Roles

Author response to Decision Letter 4

Decision Letter 5

Richard Turner

Roles

Associated Data

Supplementary Materials

Data Availability Statement

ACTIONS

PERMALINK

RESOURCES

Similar articles

Cited by other articles

Links to NCBI Databases