Relative and absolute intensity accelerometer metrics decipher the effects of age, sex, and occupation on physical activity

doi:10.21203/rs.3.rs-4251092/v1

Download PDF

Research Article

Relative and absolute intensity accelerometer metrics decipher the effects of age, sex, and occupation on physical activity

https://doi.org/10.21203/rs.3.rs-4251092/v1

This work is licensed under a CC BY 4.0 License

You are reading this latest preprint version

Background

To investigate whether quantifying both the absolute and relative intensity of physical activity (PA) improves understanding of age, sex, and occupation-related differences in PA in healthy adults aged 20–89.

Methods

In the cross-sectional COmPLETE study, participants (N = 460, 48% women, age 55 [IQR 37, 71]) wore accelerometers for up to 14 days and underwent cardiopulmonary exercise testing. Average intensity (AvAcc) and distribution of intensity (IG) of PA across the day were expressed in absolute terms (__ABS) and relative (__REL) to the acceleration at the individual´s maximum intensity, predicted from cardiorespiratory fitness.

Results

After initial increases, AvAcc__ABS and IG__ABS continuously declined beyond age 40–45, whereas AvAcc__REL and IG__REL increased until stabilising at age ~ 70 and declining at age ~ 60, respectively. Cardiorespiratory fitness constantly declined. Women had trivially higher AvAcc__ABS and moderately higher AvAcc__REL, but not IG__ABS and IG_{_REL}, than men. Occupations involving ≥ moderate PA showed higher AvAcc__ABS and AvAcc__REL, but not IG__ABS and IG_{_REL} indicating longer periods of low-intensity PA, compared to sitting/standing occupations.

Conclusions

Distinct age trajectories of absolute and relative metrics as well as cardiorespiratory fitness suggest that the age-related decline in the latter preceded that of PA. Women’s higher AvAcc__ABS and AvAcc__REL relate to more low-intensity PA combined with lower cardiorespiratory fitness rather than more health-enhancing higher-intensity PA. Finally, the intensity profile of occupational PA may provide insight into why occupational PA appears to lack a prophylactic association with health. Quantifying both the absolute and relative intensity of accelerometer-assessed PA provides greater insight than either alone.

Trial registration:

On clinicaltrials.gov (NCT03986892). Retrospectively registered 14 June 2019.

accelerometry

cardiorespiratory fitness

GGIR

activity monitors

Accelerometers have substantially advanced physical activity (PA) research by enabling the accurate and continuous measurement of human movement. This has important implications for PA surveillance and epidemiological research such as providing information on intensity and volume of PA, understanding patterns of inactivity, and allowing for a precise assessment of the relationship of PA and health outcomes.^1–3 While the use of accelerometers has been revolutionary for PA research, further potential value lies in effectively interpreting the raw data to derive meaningful insights on PA patterns.

Various metrics can be derived from raw accelerometer data to describe PA.⁴ Aside from cut-point-based metrics (i.e. light, moderate, and vigorous PA in minutes), alternative cut-point-free metrics are increasingly being used in research.^{2, 3, 5–7} Cut-point-free metrics have the advantage of enhancing comparability across studies, populations, and the most commonly used raw acceleration devices.⁸

The average level of acceleration observed over each 24-hour period assesses the average daily intensity and is commonly used as a proxy for volume of PA (AvAcc__ABS).^{6, 7} The distribution of the acceleration across the day (e.g. the relative proportions of time in higher and lower intensities of PA) provides a complementary measure of the PA profile, the intensity gradient, that gives independent information to total volume (IG__ABS).^{6, 7}This is important as the combination of volume and intensity may explain more variance in health than either alone.^{6, 7}

Looking beyond the absolute intensity of accelerometer-based PA, Orme et al. (9) recently proposed expressing absolute metrics of PA relative to the measured or predicted acceleration corresponding to an individual’s physical capacity/cardiorespiratory fitness (CRF).¹⁰ This approach was recently used by Rowlands et al. (10) to develop the relative intensity gradient (IG__REL), which reflects the intensity distribution of a 24-h day relative to an individual’s maximum (aerobic) acceleration. The authors demonstrated that individuals with low IG__REL, i.e. they had spare capacity, were more likely to increase the absolute intensity of their free-living physical activity (IG__ABS) following a PA intervention.¹⁰ Relative accelerometer metrics may thus have potential to improve personalisation of PA interventions.¹⁰

These developments also pave the way for thinking about the average intensity across the day in relative terms. Expressing both the average daily intensity (AvAcc_{_ABS} & AvAcc_{_REL}) and the distribution of intensity (IG_{_ABS} & IG_{_REL}) across the day in absolute and relative terms may expand our understanding of PA further and be helpful in research and clinical practice.

Both the volume and intensity of accelerometer-assessed PA have been shown to decrease with age^{7, 11} and differ between sexes.¹² However, what is often not considered is the well-documented impact of age and sex on CRF.¹³ Not considering these effects may confound such group comparisons. It is also unclear whether the age trajectories of PA are moderated by the age-related decline in CRF. Thus, leveraging relative alongside absolute accelerometer metrics might offer more nuanced insights into the association of PA with age and sex. Moreover, evidence suggests occupational PA does not confer the same health benefits as leisure-time PA.^{14, 15} Relative accelerometer metrics might elucidate differences in potential underlying mechanisms such as intensity patterning and physiological load between varying occupation types, helping to understand the PA paradox.¹⁴ For example, occupational PA may be at an insufficient relative intensity to lead to meaningful fitness adaptations and/or be sustained over long periods without the recovery periods that would enable meaningful adaptations.¹⁴

Therefore, this study aimed to use a sample of healthy adults aged 20 to 89 years to investigate:

the degree to which AvAcc__REL, AvAcc__ABS, IG__REL, and IG__ABS explain a unique fraction of the variance in PA
whether quantifying the average daily intensity (AvAcc_{_ABS} & AvAcc_{_REL}) and the distribution of intensity across the day (IG_{_ABS} & IG_{_REL}) in absolute and relative terms improves understanding of differences in PA by age, sex, and occupation.

Study design

Data collection took place as part of the population-based, cross-sectional COmPLETE cohort study at the Department of Sport, Exercise and Health at the University of Basel, Switzerland. COmPLETE aimed to determine trajectories of physical fitness components in a healthy population sample between ages 20 and 100. More details are available in the study protocol.¹⁶ The study was approved by the Ethics Committee of North-western and Central Switzerland (EKNZ 2017 − 01451) and all procedures followed the Declaration of Helsinki. COmPLETE was registered on clinicaltrials.gov (NCT03986892).

Study participants

Recruitment took place by sending unaddressed letters to randomly selected postal districts in the Basel area. Healthy adults ≥ 20 years of age, non-smoking, and with a BMI < 30 kg/m² were eligible to partake in this study.¹⁶ The presence of chronic exercise-limiting disease, known pregnancy or breastfeeding, orthopaedic problems hindering the examinations, inability to follow the study procedures, and contraindications for all-out exercise led to exclusion from the study.¹⁶ Before participation, all subjects provided written informed consent. Details regarding the recruitment procedures are given in the study protocol.¹⁶

Study procedures

Anthropometric and occupation data

Body mass and fat percentage were determined by four-segment bioelectrical impedance analysis using the InBody 720 (Inbody Co. Ltd., Seoul, South Korea). The device is valid and reliable for estimating body fat mass.^{17, 18}

Occupational data were collected using the European Health Survey-Physical Activity Questionnaire (EHIS-PAQ).¹⁹ In the original item, occupation type is divided into four categories: no work-related PA, sitting and standing, walking or moderate work-related PA, and vigorous or physically demanding work-related PA.¹⁹ All but two participants in the ‘no work-related PA’ category were retired (n = 151). The level of employment was self-reported on a categorical scale ranging from 0 to 100% of full-time employment in intervals of 10%.¹⁹

Cardiopulmonary exercise testing

Respiratory gases were measured breath-by-breath during one out of five ramp protocols with linear increments of 7, 10, 15, 20, or 30 W/min on a cycle ergometer (Ergoselect 200; ergoline GmbH, Bitz, Germany) using the MetaMax 3B portable metabolic system (Cortex Biophysik GmbH, Leipzig, Germany).¹⁶ The choice of protocol was based on the subject’s estimated CRF to yield an exercise duration between six and 18 min.^{16, 20} The mean of the highest consecutive V̇O₂ values (30-s mean) was defined as V̇O_2peak. Details on the protocols used as well as the exhaustion criteria applied are presented elsewhere.¹³

Accelerometer-based measurement of physical activity

GENEActiv Original (Activinsights Ltd., Kimbolton, UK) triaxial accelerometers were used. Devices were initialised with a sampling frequency of 50 Hz and participants were instructed to continuously wear the device on their non-dominant wrist for up to 14 consecutive days.¹⁶ PA data were processed using the raw data files (.bin format) in the R-package GGIR version 2.9.0.^{4, 21–23} A post-calibration error > 10 milli-gravitational units (mg) led to exclusion of accelerometer files. Euclidean norm minus one g with negative values rounded to zero (ENMO) was the accelerometer metric used to derive the below-described outcomes.²² It reflects the magnitude of dynamic acceleration corrected for gravity averaged over 5-s windows expressed in mg.^{10, 22} In line with other studies using 24-h wear protocols, a valid day was defined as wear time ≥ 14 h.^{7, 24} Moreover, at total of at least four valid weekdays (Monday to Friday) + one Saturday and one Sunday and wear data for each 15 min period of the 24 h cycle were required for inclusion in the statistical analyses.²⁵ These strict criteria were chosen to ensure data are representative of free-living PA which has been shown to differ between weekdays, Saturdays and Sundays in our cohort.²⁵

Accelerometer outcomes

The following metrics were averaged across all valid days and refer to a 24-h cycle from midnight to midnight. Their corresponding names in GGIR outputs are available in Supplemental Digital Content (SDC) Table 1. The concept of all analytical metrics is illustrated in Fig. 1.

Table 1

Cohort characteristics, overall and stratified by sex.
Characteristic	Overall, N = 460	Male, N = 238	Female, N = 222
Age, years	55 (37, 71) [20–89]	56 (37, 71) [20–89]	54 (36, 71) [20–89]
Body mass index, kg·m^-2	23.5 (21.8, 25.4)	24.4 (22.7, 26.4)	22.5 (20.9, 24.4)
Body fat, %	23.0 (17.7, 28.9)	20.0 (15.1, 25.2)	26.9 (21.7, 33.0)
Smoking status
Non-Smoker	364 (79%)	187 (79%)	177 (80%)
Ex-smoker > 10 years	95 (21%)	50 (21%)	45 (20%)
Unknown	1 (0%)	1 (0%)	0 (0%)
Occupation type
At least moderate PA	132 (29%)	50 (21%)	82 (37%)
No work-related PA	82 (18%)	49 (21%)	33 (15%)
Sitting/standing	246 (53%)	139 (58%)	107 (48%)
AvAcc__ABS, mg	30.1 (25.3, 36.7)	30.2 (24.9, 37.0)	29.9 (25.5, 36.0)
AvAcc__REL, %	2.94 (2.42, 3.59)	2.71 (2.21, 3.15)	3.24 (2.67, 3.91)
IG__ABS	-2.484 (-2.614, -2.331)	-2.441 (-2.581, -2.304)	-2.536 (-2.644, -2.368)
IG__REL	-3.076 (-3.267, -2.921)	-3.085 (-3.284, -2.921)	-3.066 (-3.249, -2.921)
Peak oxygen uptake, mL·kg^-1·min^-1	34.4 (26.9, 42.4) [13.5–65.1]	37.9 (29.7, 46.0) [16.4–65.1]	30.6 (24.2, 37.8) [13.5–53.5]
Pred. maximum ENMO, mg	1’058.8 (824.2, 1’305.8)	1’165.4 (911.5, 1’418.2)	940.6 (741.2, 1’162.3)
Data are presented as median (25th, 75th percentile), [Range], or n (%). Abbreviations: __ABS, absolute; AvAcc, average intensity; ENMO, Euclidean norm minus one; IG, intensity gradient; PA, physical activity; __REL, relative.

Absolute average intensity (AvAcc__ABS) is the arithmetic average of ENMO across a 24-h day in mg.
Relative average intensity (AvAcc__REL) is AvAcc__ABS expressed in relation to the predicted maximum (aerobic) acceleration (mg) of an individual. We predicted maximum acceleration based on the linear relationship of ENMO and oxygen uptake (V̇O₂).²⁶ Based on the assumption of the relationship between V̇O₂ and ENMO being linear for aerobic activities and the regression model by Hildebrand et al.²⁶ (SDC Fig. 1), we calculated maximal ENMO corresponding to the individuals’ CRF (V̇O_2peak measured during cardiopulmonary exercise testing) by extrapolation. This is an estimate of the maximum aerobic acceleration an individual can produce. Further details are in the SDC.
Absolute intensity gradient (IG__ABS) describes the intensity distribution of PA across a 24-h day. Time accumulated in incremental acceleration bins (bin size = 25 mg) is expressed in relation to the mid-point of each intensity bin.^{5, 10} The slope of a linear regression including both log-transformed variables (to linearise the curvilinear relationship) is the IG of an individual.^{6, 10}
Relative intensity gradient (IG__REL) describes the relative intensity distribution of PA across a 24-h day expressed in relation to an individual’s maximum (aerobic) acceleration (mg) (as described earlier).²⁶ It is derived from the epoch-level comma-separated values (.csv) files generated by GGIR using the custom-built R script available at www.github.com/Maylor8/Relative-Intensity-Gradient.¹⁰ Calculations are similar to that of IG_{_ABS} only that log-transformed accumulated time is regressed on log-transformed relative intensity instead of absolute intensity.¹⁰ The mid-percentage of relative intensity bins of 5% (5 to > 300% of maximum acceleration) is used in this case.¹⁰
MX metrics reflect the acceleration (mg) above which a person’s most active X minutes are accumulated.²⁷ The relative MX metrics were calculated using the individual’s predicted maximum aerobic acceleration: MX / maximum acceleration * 100.^{9, 10} This was done to display the absolute and relative intensity of the PA profile of the cohort and selected subgroups (i.e. stratified by age groups, sex, and occupation type). MX plots were generated for periods of 1-720 min (12 h) both on the original scale and Z-transformed (to better visualise differences in longer periods) using customised open-source code available at: www.github.com/Maylor8/RadarPlotGenerator.

*** insert Fig. 1 here ***

Statistical analyses

Statistical analyses and figures were done in R version 4.2.0.²⁸ Accelerometer metrics in text are presented as Z-scores unless stated otherwise.

Correlation analyses: Correlations between absolute and relative cut-point-free metrics were examined using Spearman’s Rank Correlation. A principal component analysis of AvAcc__REL, AvAcc__ABS, IG__REL, and IG__ABS, was conducted to investigate the degree of information overlap and unique information among all relative and absolute accelerometer metrics.²⁹

Regression analyses: The association between age, sex, occupation type and each of the absolute and relative cut-point-free metrics was examined using a series of multiple linear regression models. Dependent variables (AvAcc__REL, AvAcc__ABS, IG__REL, IG__ABS, and MX metrics) were Z-transformed for better comparability. A directed acyclic graph identified the minimal necessary parameters included in the models to estimate unbiased effects (Suppl. Figure 2).³⁰ Age (continuous variable), sex (dichotomous variable: female, male), and occupation type (categorical variable: at least moderate PA, no work-related PA, sitting/standing) were included as independent variables. Moreover, models were adjusted for body fat percentage (continuous variable), seasonality (categorical variable: spring, summer, autumn, or winter), and level of employment (categorical variable: not working, 10–50%, 60–90%, and 100% of full-time employment). We used third-degree polynomials to model all continuous independent variables.³¹ Fulfilment of model assumptions was verified using residual diagnostics. Dependent variables were log-transformed before Z-transformation in case assumptions for normality distribution and homogeneity of residuals were not fulfilled. If our data indicated an association based on the estimate and the lower or upper limits of the 95% CI included zero (± 5% of the width of the interval), this was flagged accordingly. Effect sizes were categorised as small (≥ 0.1 to < 0.3), medium (≥ 0.3 to < 0.5) and large (≥ 0.5) on the Z-scale.³²

Sensitivity analyses: Due to the high number of retired individuals reporting ‘no work-related PA’, occupational analyses were re-run without retired individuals. Moreover, occupational analyses were re-run including only weekdays to examine potential difference in the strength of association compared to including all days.

Table 1 shows cohort characteristics stratified by sex. The flow of participants is presented in Suppl. Figure 3. Of a total of 629 participants in the COmPLETE study, 460 were included in the final statistical analyses.

*** insert Table 1 here ***

Relationship between relative and absolute cut-point free accelerometer metrics

All relative and absolute cut-point-free accelerometer metrics were moderately and positively correlated except for AvAcc__REL and IG__ABS (Suppl. Figure 4A). The results of the principal component analysis indicate that AvAcc__REL and IG__ABS contribute via a different dimension to explaining the variance in the data, whereas AvAcc__ABS and IG__REL contribute rather similar information to the overall variance (Suppl. Figure 4B).

Association of relative and absolute metrics with age

The association of relative and absolute accelerometer metrics with age are compared to each other in Fig. 2 and Suppl. Figure 5. AvAcc__ABS, IG__ABS, increased from age 20, plateauing briefly around age ~ 40. Beyond age 40–45, both AvAcc__ABS and IG__ABS continuously declined. In contrast, AvAcc__REL and IG__REL continued to increase before stabilising at age ~ 70 or declining at age ~ 60, respectively. Detailed results of the linear regression models are in Suppl. Table 2 (Z-transformed dependent variable), Suppl. Table 3 (original scales), and Suppl. Table 4 (in percentages).

*** insert Fig. 2 here ***

The lower absolute intensity of the PA profile in the older participants was most evident during the most intense 60 minutes of the day, with effect sizes ranging from small to large across the absolute MX metrics M1 to M60. However, in longer, less intense periods (M120 to M720) the associations ranged from small to trivial. For the latter, our data were also compatible with negligible to moderate effects (Fig. 3a, abs. MX_acc). In contrast, relative intensity was lowest in the youngest age-group across all time periods (1 min to 12 h), with small to large between-group differences observed for all relative MX metrics depending on the age group (Fig. 3b, rel. MX_acc).

*** insert Fig. 3 here ***

Association of relative and absolute metrics with sex

While AvAcc__ABS was only trivially associated with sex, being 5% higher in women than men (0.18, 95% -0.01 to 0.37), AvAccel__REL was 13% higher in women (0.45, 95% CI 0.27 to 0.64). Sex differences for the IG were negligible (IG__ABS: -0.05, 95% -0.24 to 0.13; IG__REL: 0.12, 95% -0.11 to 0.34). Detailed results of the regression models are in Suppl. Tables 2–4.

The trend towards a higher AvAcc__ABS over the PA profile in women was due to higher intensities during longer periods, M360 (6 h) to M480 (8 h) (Fig. 3a, abs. MX). In contrast, the higher AvAcc__REL was due to small to moderately higher relative intensity in women than in men across short as well as long time periods, M2 to M720 (12 h) (Fig. 3b, rel. MX).

Association of relative and absolute metrics with occupation status

Occupation type was moderately associated with AvAcc__ABS and AvAcc__REL (Fig. 4, Suppl. Tables 2–4). Explicitly, predominantly sitting or standing at work was independently associated with lower AvAcc__ABS (-9.81%, 95% CI -15.42 to -4.2) and AvAcc__REL (-11.02%, 95% CI -16.6 to -5.44) compared to at least moderate work-related PA (Fig. 4). There was a negligible association of occupation type with IG__REL or IG__ABS (Fig. 4, Suppl. Table 2). Results were consistent when considering only weekdays (Suppl. Table 5). Considering only weekend days, associations of occupation type with AvAcc_{_ABS} and AvAcc_{_REL} were smaller and negligible (Suppl. Table 6).

*** insert Fig. 4 here ***

The higher AvAcc__ABS over the PA profile in those with moderate work-related PA relative to those in sit/stand occupations was due predominantly to higher intensities during the longer periods, with small to moderate associations evident for M120 to M720 (12 h) (Fig. 3a).

Similarly, the higher AvAcc__REL in those with moderate work-related PA relative to those in sit/stand occupations was predominantly due to higher relative intensity during long periods of the day, with small to medium associations observed for M720 to M360 (Fig. 3b).

Our study demonstrated that 1) the relative intensity of PA increased with age while the absolute intensity plateaued. This together with a continuously declining CRF indicates the age-related decline in cardiorespiratory fitness might preceded that of PA. 2) Women had trivially higher AvAcc_{_ABS}, yet moderately higher AvAcc_{_REL}. The higher average intensity across the day (volume) was due to more time spent at lower intensity accelerations. However, in combination with the women’s lower CRF, this resulted in the greater sex difference in relative average intensity across the day, rather than more higher-intensity PA, as reflected in no difference in absolute IG. 3) Individuals with at least moderate occupational PA have higher AvAcc__ABS and AvAcc__REL, but not IG. This occurs primarily through more intense longer periods, unlike leisure-time PA that tends to focus on shorter periods of higher PA (thus increases IG__ABS and IG__REL) and allows time for recovery.

Association of relative and absolute metrics with age

Up to 40 years of age, trajectories of AvAcc__ABS and IG__ABS indicate favourable changes with a higher absolute volume of PA in a day together with less time spent inactive and/or more time spent at higher absolute intensities of PA. Further, both relative parameters increased similarly, indicating that the increase in absolute intensity is reflected in an increase in physical effort.

Beyond 40 years of age, we documented a decrease in PA volume and a shift towards PA at lower absolute intensities and/or more inactive time, as shown previously.⁷ However, the corresponding increase in, AvAcc__REL shows that despite a lower volume of PA, it is accumulated at a rising percentage of an individual’s capacity (i.e., AvAcc__REL). Simultaneously, IG__REL stabilised. The dissimilarity in slopes between the two relative metrics indicates that they may reflect different information beyond 50 years of age. Assessing the relative IG alone, might lead to falsely assuming that daily PA evokes a similar physiological burden upon the individual despite advancing age. There is a narrowing of the absolute intensity spectrum attainable with advancing age due to the ageing-related decline in physical function.^{13, 33} Activities of daily life that may have been on the lower end of the relative intensity spectrum with younger age (e.g., grocery shopping, household chores) may now require much more relative effort. Consequently, these types of activities take longer and make up for a larger fraction of the daily volume of PA, increasing the AvAcc__REL despite the lower absolute intensity (AvAcc__ABS and IG__ABS) of the activities. This may suggest an increasing physiological burden across the day. However, the stability of the IG__REL post age ~ 40–45, before declining post age ~ 60, suggests that distribution of the more intense activity is adapted in accordance with declining capacity. This suggests that all four parameters are important to fully understand the ageing-related changes in PA patterns and age is an important variable to consider for personalising PA interventions and evaluation. A detailed illustration of how the four metrics interrelate with ageing is in Fig. 5.

*** insert Fig. 5 here ***

MX plots further elucidate these age differences in absolute and relative metrics. While all age groups are similarly active at an absolute intensity for the longer duration MX metrics, the intensity of PA drops notably for the shorter periods of higher intensity in 61-90-year-olds compared to the younger individuals. Given the consistency in absolute MX metrics for 20–40 and 41-60-year-olds, the higher relative intensity of the MX metrics in 41-60-year-olds again suggests that the age-related decline in CRF precedes the decrease in PA intensity (Fig. 2). This also indicates that older adults are under a higher physiological load during long portions of the day compared to younger adults. Indeed, relative intensity seems to converge between middle-aged and older adults regarding shorter, more intense periods, despite the absolute intensity being much lower in the 61-90-year olds. We may hypothesise that while middle-aged individuals still engage in more intense activities, older individuals do not, but nevertheless they still experience the same higher physiological load due to lower CRF. This may be tackled by replacing inactive time with health-enhancing PA at an appropriate relative intensity to improve CRF and reduce daily physiological load.

Association of relative and absolute metrics with sex

In our study, AvAcc__REL and AvAcc__ABS, but not IG__ABS and IG__REL, tended to be higher in women. Higher AvAcc__ABS, but not IG__ABS, in women is consistent with UK Biobank,¹² but not with Mielke et al. (34) who noted no sex differences in AvAcc__ABS among Australian adults. However, the presence of sex differences may vary depending on geographical/societal context.³⁵ The greater difference in AvAcc__REL than AvAcc__ABS in women is explained by lower levels of CRF compared to men.¹³ The greater volume in women was due to more time spent at lower intensity accelerations, rather than accumulating higher intensity PA, as reflected in no difference in IG__ABS.

Association of relative and absolute metrics with occupation status

Individuals engaging in more physically demanding work exhibit higher average absolute and relative intensity of PA than individuals in more sedentary roles or those not engaging in work-related PA. This aligns with observations from the cross-sectional National Health and Nutrition Examination Survey (NHANES), which reported more weekday PA for individuals with active jobs than those with inactive jobs.³⁶ Our findings extend these by providing a relative PA perspective that is to date unique. Both the average absolute and relative intensity of PA is higher indicating that the higher relative intensity in individuals with more active jobs is due to them engaging in more intense PA rather than having a lower CRF. Confirming this, a sensitivity analysis showed no difference in V̇O_2peak between occupation types (data not presented). Our data suggests that higher relative intensity seen in those with greater absolute volume of PA is predominantly due to higher absolute intensities during longer periods (i.e., MX_acc > 120 min), explaining the lack of difference in IG by occupation in absolute or relative terms. Importantly, this also shows that despite their higher occupational PA, these workers do not have higher CRF.

These findings may have relevant implications for the PA paradox,¹⁴ as they support the hypothesis that the patterns of PA associated with occupational PA may not be optimal for improving CRF or cardiovascular health, e.g., be of too low intensity (i.e. <60% of maximum aerobic capacity) and/or too long duration, without appropriate recovery periods.¹⁴ On the other hand, the average relative intensities of occupational PA over an 8-hour working day may exceed the recommended levels for such longer periods (> 30–35% of maximal aerobic capacity).^{14, 37} Moreover, the higher relative intensity during prolonged periods means more time spent at elevated heart rate, which is a known risk factor for cardiovascular disease and mortality.^{14, 38}

Limitations

Predicted maximum acceleration, used to determine relative intensity of PA, is derived from the extrapolation of acceleration up to the value corresponding to V̇O_2peak. This may induce inaccuracy compared to directly measuring maximum acceleration using ambulatory tests such as the incremental shuttle walking test.⁹ However, we used the gold standard, cardiopulmonary exercise testing, to measure V̇O_2peak and our simulation showed good accuracy of the regression models by Hildebrand et al. (26). Even if there may be a prediction error, using the same model to calculate maximum acceleration ensures comparability between participants within this study. PA patterns and the associations of accelerometer metrics with age, sex, and occupation type might differ between cohort characteristics, geographic and societal contexts. The findings of our study are hence primarily generalisable to a Central European healthy adult population. The categorisation by occupation type relied on self-reports, which were limited to a few categories and it is possible that some participants may not have been working during their monitoring period, potentially skewing the results. Yet, a sensitivity analysis showed consistent results when including only weekdays (Suppl. Table 5). Furthermore, the occupational aspect is primarily relevant for those still in the workforce. A considerable portion of our participants over the age of 65 years were retired (n = 153) and all except for two fell into the ‘no work-related PA’ category. Further, the main analyses were adjusted for age and level of employment in % of full-time. Moreover, subgroup analyses with retired participants excluded were similar (Suppl. Figures 6&13). Lastly, age-, sex-, and occupation-specific variations in PA patterns caused by fragmentation, frequency, duration, and temporal distribution were not investigated in this study but might explain some of the associations.

Our findings help decipher age, sex, and occupation type-related PA patterns by providing a perspective considering the individual´s physical capacity. Firstly, age-related changes in absolute and relative intensity indicated the age-related decline in CRF preceded a drop in absolute intensity and volume of PA. Despite increasing physiological load, absolute volume and intensity distribution of PA increased up to age 40 and declined thereafter. Secondly, the slightly higher PA volume in women is mainly explained by more time spent at lower intensity accelerations rather than higher (potentially) more health-enhancing intensity PA. This is accompanied by overall higher relative intensity due to lower CRF in women compared to men. Thirdly, we provided insights into hypotheses relating to the underlying mechanisms of the PA paradox. Higher occupational PA seems to increase absolute average intensity of PA and thus physiological load, but is not accommodated by higher CRF, as indicated by higher relative intensity. The increased PA occurs primarily through higher intensities over longer periods of time, unlike leisure-time PA that focuses on shorter periods of health-enhancing PA at higher intensities and allows time for recovery. This knowledge may allow for better personalisation of PA interventions and interpretation of accelerometer-based PA data.

AvAcc__ABS Absolute average intensity

AvAcc__REL Relative average intensity

CRF Cardiorespiratory fitness

ENMO Euclidean norm minus one

IG__ABS Absolute intensity gradient

IG__REL Relative intensity gradient

PA Physical activity

SD Standard deviation

SDC Supplemental Digital Content

Ethics approval and consent to participate

This study was approved by the Ethics Committee of North-western and Central Switzerland (EKNZ 2017–01451). All participants in this study provided informed written consent.

Consent for publication

Not applicable.

Data availability statement

The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions.

Competing interests

The authors declare that they have no competing interests.

Funding

The COmPLETE study was funded by the Swiss National Science Foundation (Grant no. 182815). RK is funded by the Swiss National Science Foundation (Grant No. PZ00P3_208999/1). AVR is supported by the Lifestyle Theme of the Leicester NHR Leicester Biomedical Research Centre and NIHR Applied Research Collaborations East Midlands (ARC-EM).

Authors contributions

FS conceptualised the manuscript and wrote the original draft. FS analysed, visualised, and interpreted the data. DI and EL contributed to the statistical analyses. JW, AST, TH, and RK conceptualised the study and defined the methods. JW and RK collected the data. FS, JW, and RK were responsible for data curation. RK, JW, EL, DI, TH, AVR, and AST revised the manuscript. AST obtained funding for the project. All authors approved the final version of the manuscript.

Acknowledgements

We value the contributions of all participants in the COmPLETE study. We acknowledge the use of thermometer vectors by Vecteezy (Figures 1&5; https://www.vecteezy.com/free-vector/thermometer).

Fairclough SJ, Rowlands AV, del Pozo Cruz B, Crotti M, Foweather L, Graves LEF, et al. Reference values for wrist-worn accelerometer physical activity metrics in England children and adolescents. Int J Behav Nutr Phys Act. 2023;20(1):35. 10.1186/s12966-023-01435-z.
Dempsey PC, Rowlands AV, Strain T, Zaccardi F, Dawkins NP, Razieh C, et al. Physical activity volume, intensity and incident cardiovascular disease. Eur Heart J. 2022;43(46):4789–800.
Dawkins NP, Yates T, Edwardson CL, Maylor B, Henson J, Hall AP, et al. Importance of Overall Activity and Intensity of Activity for Cardiometabolic Risk in Those with and Without a Chronic Disease. Med Sci Sports Exerc. 2022;54(9):1582–90. 10.1249/mss.0000000000002939.
Migueles JH, Rowlands AV, Huber F, Sabia S, van Hees VT. GGIR: A Research Community–Driven Open Source R Package for Generating Physical Activity and Sleep Outcomes From Multi-Day Raw Accelerometer Data. J Meas Phys Behav. 2019;2(3):188–96. 10.1123/jmpb.2018-0063.
Rowlands AV, Fairclough SJ, Yates TOM, Edwardson CL, Davies M, Munir F, et al. Activity Intensity, Volume, and Norms: Utility and Interpretation of Accelerometer Metrics. Med Sci Sports Exerc. 2019;51(11):2410–22.
Rowlands AV, Edwardson CL, Davies MJ, Khunti K, Harrington DM, Yates T. Beyond Cut Points: Accelerometer Metrics that Capture the Physical Activity Profile. Med Sci Sports Exerc. 2018;50(6):1323–32. 10.1249/mss.0000000000001561.
Schwendinger F, Wagner J, Knaier R, Infanger D, Rowlands AV, Hinrichs T, et al. Accelerometer metrics: healthy adult reference values, associations with cardiorespiratory fitness, and clinical implications. Med Sci Sports Exerc. 2023;56(2):170–80. 10.1249/MSS.0000000000003299.
Migueles JH, Molina-Garcia P, Torres-Lopez LV, Cadenas-Sanchez C, Rowlands AV, Ebner-Priemer UW, et al. Equivalency of four research-grade movement sensors to assess movement behaviors and its implications for population surveillance. Sci Rep. 2022;12(1):5525. 10.1038/s41598-022-09469-2.
Orme MW, Lloyd-Evans PHI, Jayamaha AR, Katagira W, Kirenga B, Pina I, et al. A Case for Unifying Accelerometry-Derived Movement Behaviors and Tests of Exercise Capacity for the Assessment of Relative Physical Activity Intensity. J Phys Act Health. 2023;20(4):303–10. 10.1123/jpah.2022 – 0590.
Rowlands AV, Orme MW, Maylor BD, Kingsnorth A, Herring LY, Khunti K, et al. Can quantifying the relative intensity of a person’s free-living physical activity predict how they respond to a physical activity intervention? Findings from the PACES RCT. Br J Sports Med. 2023. 10.1136/bjsports-2023-106953.
Wanner M, Hartmann C, Pestoni G, Martin BW, Siegrist M, Martin-Diener E. Validation of the Global Physical Activity Questionnaire for self-administration in a European context. BMJ Open Sport Exerc Med. 2017;3(1):e000206. 10.1136/bmjsem-2016-000206.
Rowlands AV, Dempsey PC, Gillies C, Kloecker DE, Razieh C, Chudasama Y, et al. Association Between Accelerometer-Assessed Physical Activity and Severity of COVID-19 in UK Biobank. Mayo Clin Proc Innov Qual Outcomes. 2021;5(6):997–1007. https://doi.org/10.1016/j.mayocpiqo.2021.08.011.
Wagner J, Knaier R, Infanger D, Königstein K, Klenk C, Carrard J, et al. Novel CPET Reference Values in Healthy Adults: Associations with Physical Activity. Med Sci Sports Exerc. 2020;35(1):26–37. 10.1249/mss.0000000000002454.
Holtermann A, Krause N, van der Beek AJ, Straker L. The physical activity paradox: six reasons why occupational physical activity (OPA) does not confer the cardiovascular health benefits that leisure time physical activity does. Br J Sports Med. 2018;52(3):149. 10.1136/bjsports-2017-097965.
Cillekens B, Huysmans MA, Holtermann A, van Mechelen W, Straker L, Krause N, et al. Physical activity at work may not be health enhancing. A systematic review with meta-analysis on the association between occupational physical activity and cardiovascular disease mortality covering 23 studies with 655 892 participants. Scand J Work Environ Health. 2022;48(2):86–98. 10.5271/sjweh.3993.
Wagner J, Knaier R, Infanger D, Arbeev K, Briel M, Dieterle T, et al. Functional aging in health and heart failure: the COmPLETE Study. BMC Cardiovasc Disord. 2019;19(1):180. 10.1186/s12872-019-1164-6.
Anderson LJ, Erceg DN, Schroeder ET. Utility of multifrequency bioelectrical impedance compared with dual-energy x-ray absorptiometry for assessment of total and regional body composition varies between men and women. Nutr Res. 2012;32(7):479–85. https://doi.org/10.1016/j.nutres.2012.05.009.
McLester CN, Nickerson BS, Kliszczewicz BM, McLester JR. Reliability and Agreement of Various InBody Body Composition Analyzers as Compared to Dual-Energy X-Ray Absorptiometry in Healthy Men and Women. J Clin Densitom. 2020;23(3):443–50. 10.1016/j.jocd.2018.10.008.
Baumeister SE, Ricci C, Kohler S, Fischer B, Töpfer C, Finger JD, et al. Physical activity surveillance in the European Union: reliability and validity of the European Health Interview Survey-Physical Activity Questionnaire (EHIS-PAQ). Int J Behav Nutr Phys Act. 2016;13(1):61. 10.1186/s12966-016-0386-6.
Midgley AW, Bentley DJ, Luttikholt H, McNaughton LR, Millet GP. Challenging a dogma of exercise physiology: does an incremental exercise test for valid VO2max determination really need to last between 8 and 12 minutes? Sports Med. 2008;38(6):441–7. 10.2165/00007256-200838060-00001.
Hees VTv, Fang Z, Langford J, Assah F, Mohammad A, Silva ICMd, 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(7):738–44. 10.1152/japplphysiol.00421.2014.
van Hees VT, Gorzelniak L, Dean León EC, Eder M, Pias M, Taherian S, et al. Separating Movement and Gravity Components in an Acceleration Signal and Implications for the Assessment of Human Daily Physical Activity. PLoS ONE. 2013;8(4):e61691. 10.1371/journal.pone.0061691.
Sabia S, van Hees VT, Shipley MJ, Trenell MI, Hagger-Johnson G, Elbaz A, et al. Association Between Questionnaire- and Accelerometer-Assessed Physical Activity: The Role of Sociodemographic Factors. Am J Epidemiol. 2014;179(6):781–90. 10.1093/aje/kwt330.
Herrmann SD, Barreira TV, Kang M, Ainsworth BE. How Many Hours Are Enough? Accelerometer Wear Time May Provide Bias in Daily Activity Estimates. J Phys Act Health. 2013;10(5):742–9. 10.1123/jpah.10.5.742.
Schwendinger F, Wagner J, Infanger D, Schmidt-Trucksäss A, Knaier R. Methodological aspects for accelerometer-based assessment of physical activity in heart failure and health. BMC Med Res Methodol. 2021;21(1):251. 10.1186/s12874-021-01350-6.
Hildebrand M, van Hees VT, Hansen BH, Ekelund U. Age Group Comparability of Raw Accelerometer Output from Wrist- and Hip-Worn Monitors. Med Sci Sports Exerc. 2014;46(9):1816–24. 10.1249/mss.0000000000000289.
Rowlands AV, Dawkins NP, Maylor B, Edwardson CL, Fairclough SJ, Davies MJ, et al. Enhancing the value of accelerometer-assessed physical activity: meaningful visual comparisons of data-driven translational accelerometer metrics. Sports Med Open. 2019;5(1):47. 10.1186/s40798-019-0225-9.
R Core Team. R: A language and environment for statistical computing Vienna, Austria: R Foundation for Statistical Computing. 2021 [ https://www.R-project.org/].
Mevik B-H, Wehrens R. The pls Package: Principal Component and Partial Least Squares Regression in R. J Stat Softw. 2007;18(2):1–23. 10.18637/jss.v018.i02.
Shrier I, Platt RW. Reducing bias through directed acyclic graphs. BMC Med Res Methodol. 2008;8(1):70. 10.1186/1471-2288-8-70.
Perperoglou A, Sauerbrei W, Abrahamowicz M, Schmid M. A review of spline function procedures in R. BMC Med Res Methodol. 2019;19(1):46. 10.1186/s12874-019-0666-3.
Cohen J. Statistical power analysis for the behavioral sciences. Academic; 2013.
Lichtenstein E, Wagner J, Knaier R, Infanger D, Roth R, Hinrichs T, et al. Norm Values of Muscular Strength Across the Life Span in a Healthy Swiss Population: The COmPLETE Study. Sports Health. 2022. 10.1177/19417381221116345.
Mielke GI, Burton NW, Brown WJ. Accelerometer-measured physical activity in mid-age Australian adults. BMC Public Health. 2022; 22(1): 1952. 10.1186/s12889-022-14333-z.
Amagasa S, Inoue S, Shibata A, Ishii K, Kurosawa S, Owen N, et al. Differences in Accelerometer-Measured Physical Activity and Sedentary Behavior Between Middle-Aged Men and Women in Japan: A Compositional Data Analysis. J Phys Act Health. 2022;19(7):500–8. 10.1123/jpah.2022-0098.
Van Domelen DR, Koster A, Caserotti P, Brychta RJ, Chen KY, McClain JJ, et al. Employment and Physical Activity in the U.S. Am J Prev Med. 2011;41(2):136–45. https://doi.org/10.1016/j.amepre.2011.03.019.
Jørgensen K. Permissible loads based on energy expenditure measurements. Ergonomics. 1985;28(1):365–9. 10.1080/00140138508963145.
Korshøj M, Lidegaard M, Kittel F, Van Herck K, De Backer G, De Bacquer D, et al. The relation of ambulatory heart rate with all-cause mortality among middle-aged men: a prospective cohort study. PLoS ONE. 2015;10(3):e0121729.
Supplemental Digital Content.
Supplemental_Digital_Content.docx.

No competing interests reported.

SupplementalDigitalContent.docx

Download PDF

Editorial decision: Revision requested
16 Aug, 2024
Reviews received at journal
13 Aug, 2024
Reviews received at journal
13 Aug, 2024
Reviewers agreed at journal
12 Aug, 2024
Reviews received at journal
07 Aug, 2024
Reviewers agreed at journal
02 Aug, 2024
Reviewers agreed at journal
02 Aug, 2024
Reviewers agreed at journal
30 Jul, 2024
Reviewers invited by journal
26 Jul, 2024
Editor invited by journal
22 Apr, 2024
Submission checks completed at journal
11 Apr, 2024
Editor assigned by journal
11 Apr, 2024
First submitted to journal
11 Apr, 2024

You are reading this latest preprint version

Relative and absolute intensity accelerometer metrics decipher the effects of age, sex, and occupation on physical activity

Status:

Version 1

Abstract

Background

Methods

Results

Conclusions

Trial registration:

Figures

INTRODUCTION

MATERIALS AND METHODS

Study design

Study participants

Study procedures

Anthropometric and occupation data

Cardiopulmonary exercise testing

Accelerometer-based measurement of physical activity

Accelerometer outcomes

Statistical analyses

RESULTS

Relationship between relative and absolute cut-point free accelerometer metrics

Association of relative and absolute metrics with age

Association of relative and absolute metrics with sex

Association of relative and absolute metrics with occupation status

DISCUSSION

Association of relative and absolute metrics with age

Association of relative and absolute metrics with sex

Association of relative and absolute metrics with occupation status

Limitations

PERSPECTIVE

Abbreviations

Declarations

References

Additional Declarations

Supplementary Files

Status:

Version 1