Is occupational physical activity associated with all-cause mortality in UK Biobank?

Objectives To investigate associations between occupational physical activity (OPA) and all-cause mortality. Methods From baseline (2006-2010), 452,884 UK Biobank participants (aged 40-69 years) were followed for a median 11.1 (IQR: 10.4-11.8) years. OPA was categorised by cross-tabulating degree of manual work and walking/standing work amongst those in paid employment (n=264,424), whereas categories of occupational status were used for those not in paid employment (n=188,460). Cox proportional hazards models were used to estimate hazard ratios (HR) and 95% confidence intervals (CI) for all-cause mortality by occupational category, and for working hours/week and non-occupational physical activity stratified by occupational category. Results During 4,965,616 person-years of follow-up, 23,310 deaths occurred. Compared to those in sedentary jobs, retirement was associated with lower mortality in women (HR=0.74, CI:0.68-0.81) and men (HR=0.85, CI:0.79-0.92), whereas unemployment was associated with higher mortality in men (HR=1.26, CI:1.10-1.45). There was no evidence of mortality differences by OPA category within the working population. Working <35 hours/week versus 35-40 hours/week was associated with lower mortality in both women (HR=0.86, CI:0.79-0.93) and men (HR=0.81, CI:0.75-0.88), with no interaction by OPA. Non-occupational physical activity was associated with lower mortality in both women (HR=0.90 per 5 kJ/day/kg, CI:0.84-0.96) and men (HR=0.88 per 5 kJ/day/kg, CI:0.84-0.92), with no interaction by OPA. Conclusion Work classified as having higher levels of OPA may not be as active as reported, or the types of physical activity performed in those jobs are not health-enhancing. Irrespective of OPA category or employment status, non-occupational physical activity appears to provide health benefits.

is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint The copyright holder for this this version posted December 19, 2020. ;https://doi.org/10.1101https://doi.org/10. /2020 • There were no differences in mortality between categories with different levels of self-48 reported OPA 49 • Physical activity outside of work was associated with lower hazard of all-cause 50 mortality and there was no interaction with occupational physical activity, indicating 51 similar benefits across different jobs types. 52

How might it impact on clinical practice in the future? 53
• Health professionals should be aware that occupations assumed to involve more 54 physical activity may not be as active as reported, or the types of physical activity 55 performed in those jobs may not be health-enhancing. 56 • Physical activity during leisure-time should continue to be recommended to adults of all 57 paid and unpaid occupational categories. 58 . CC-BY 4.0 International license It is made available under a perpetuity.
is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint The copyright holder for this this version posted December 19, 2020. ;https://doi.org/10.1101https://doi.org/10. /2020 BACKGROUND 59 The benefits of physical activity for health are well established, [1,2] with guidelines from the 60 UK Chief Medical Officers [3] and the World Health Organization [4] recommending the 61 equivalent of 150 minutes of moderate-intensity aerobic physical activity each for maintenance 62 of good physical and mental health. No distinctions are made between physical activity in 63 leisure-time, transport, home, occupational domains; the total volume of activity is assumed to 64 be beneficial regardless of the domain in which it was accrued. Contradictory to this advice is 65 the suggestion that occupational physical activity (OPA) does not confer the same benefits, 66 and may even be harmful to health.
[5] One meta-analysis reported that male (but not female) 67 workers with high level OPA were at 18% higher risk of mortality compared to those at low 68 levels.
[6] Reasons proposed for these findings include OPA being performed at lower 69 intensities, for protracted periods, and in a static posture,[7] but the existing evidence also has 70 limitations including the use of crude self-reported OPA measures, limited adjustment for non-71 occupational physical activity, and residual confounding for socio-economic status and lifestyle 72 factors (e.g. smoking). [8,9] Prior studies within occupational strata have reported that active 73 jobs were associated with lower morality.[10,11] 74 UK Biobank is a large prospective cohort study total and domain-specific physical activity 75 data, as well as occupational variables. These data can be combined in such a way that both 76 work category and hours can be used to characterise exposure to different types and volumes 77 of OPA. A range of lifestyle, socio-economic and health-related variables are collected using a 78 standardised protocol, and it is also possible to calibrate self-reported non-occupational 79 physical activity to objective measures of physical activity using the accelerometer sub-80 cohort[12] to better control for physical activity outside of work. UK Biobank has sufficient 81 sample size and accrued deaths to allow stratification by sex and occupational categories. 82 This can better address issues of confounding patterned by occupational group with strata 83 larger than many occupational cohorts. This presents opportunities for improving our 84 understanding of OPA, particularly in the UK, where there are few contemporary analyses. In 85 . CC-BY 4.0 International license It is made available under a perpetuity.
is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint The copyright holder for this this version posted December 19, 2020. ;https://doi.org/10.1101https://doi.org/10. /2020 this study, we investigated whether occupational category and hours of work in different job 86 categories were associated with all-cause mortality. 87

METHODS 88
Participants and study design 89 UK Biobank is an ongoing prospective cohort study of men and women aged 40-69 years 90 residing within 25 miles of one of 22 assessment centres in England, Scotland, and Wales. degree of standing/walking work, and working duration in hours per week were self-reported 100 using a touch-screen questionnaire.
[15] Participants with missing data for these variables 101 were excluded from analyses (n=9,362), as were those reporting paid employment status but 102 zero working hours (n=186). For those in paid employment, degree of manual work and 103 degree of standing/walking work were both reported as one of four categories: "never/rarely", 104 "sometimes", "usually", "always". Responses of "usually" and "always" were collapsed for both 105 manual work and standing/walking, with the two variables cross-tabulated (Supplementary 106 Table 1) to create six mutually exclusive OPA categories: "no manual, no standing/walking", 107 "no manual, some standing/walking", "no manual, usually standing/walking", "some manual, 108 some standing/walking", "some manual, usually standing/walking", "usually manual, usually 109 standing/walking". Total physical activity estimated by median wrist acceleration showed a 110 . CC-BY 4.0 International license It is made available under a perpetuity.
is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint The copyright holder for this this version posted December 19, 2020. ;https://doi.org/10.1101https://doi.org/10. /2020 6 predictable association across these categories within the accelerometery sub-cohort, [16] 111 indicating face validity (Supplementary Figure 2). 112 The first set of analyses used as the exposure (hereafter defined as "occupational categories") 113 OPA categories for those in paid employment, and occupational status of those not in paid 114 employment (retired, unable to work due to illness/disability, caring for home/family, student, 115 unemployed, unpaid work). The second set of analyses included only those in paid 116 employment with tertiles of working hours as the exposure (<35, 35-40, >40 hours per week) 117 and stratification by OPA category. In supplementary analyses we repeated both of the above 118 but OPA category was replaced by SOC group for those in paid employment while including 119 the same occupational status categories for those not in paid employment. Further details on 120 the SOC group classifications are provided in Supplementary Table 2. 121 To investigate whether OPA moderates the association of physical activity outside of work 122 with mortality, we examined the association between non-occupational physical activity energy 123 expenditure (PAEE) and all-cause mortality in two sets of stratified analyses (by OPA 124 categories and by tertiles of working hours). We previously showed how self-reported 125 variables representing multiple behaviours could be combined to predict total PAEE in UK 126 is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint March 2020 in all three nations. 137

Statistics 157
For each exposure, a Cox proportional hazards model with age as the underlying timescale 158 was used to estimate hazard ratios (HR) and 95% confidence intervals (CI) for all-cause 159 mortality. The proportional hazards assumption for categorical variables was examined using 160 log-log plots; the baseline hazard function was stratified by levels of those variables that did 161 . CC-BY 4.0 International license It is made available under a perpetuity.
is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint The copyright holder for this this version posted December 19, 2020. ; https://doi.org/10.1101/2020.12.18.20248428 doi: medRxiv preprint 8 not satisfy this assumption (fruit and vegetable intake, income, education). Fractional 162 polynomials showed that working hours per week did not meet the log-linear assumption so 163 this variable was categorised using tertile boundaries, while meat consumption (log[x+1]) and 164 Townsend index ([x] 2 ) were transformed. Variance inflation factors and Pearson correlations 165 indicated no strong collinearity between variables. All analyses were stratified by sex a priori, 166 based on findings from previous studies.
[6] Wald tests were used to examine the potential 167 interactions of working hours and non-occupational PAEE with OPA category. Models 168 included all covariates listed above, as well as separately omitting BMI and resting heart rate 169 which may be on the causal pathway between physical activity and mortality. Individuals with 170 missing covariate data (n=37,965) were excluded, as were those who died in the first two 171 years of follow-up (n=2,096) to mitigate potential reverse causation. For the same reason, we 172 conducted sensitivity analyses excluding those with prevalent coronary heart disease, stroke, 173 or cancer at baseline (n=63,755). All analyses were performed using STATA/SE 16.1 174 (StataCorp, TX, USA). 175

Patient and public involvement 176
Patients and members of the public were not formally involved in the design, analysis or 177 interpretation of this study. 178

179
In a sample of 452,884 participants, 23,310 deaths occurred during a median 11.1 (IQR: 10.4-180 11.8) years of follow up (4,965,616 person-years). Baseline characteristics of the participants 181 by sex and occupational status are shown in Table 1 is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint The copyright holder for this this version posted December 19, 2020. ; https://doi.org/10.1101/2020.12.18.20248428 doi: medRxiv preprint is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint The copyright holder for this this version posted December 19, 2020. ; https://doi.org/10.1101/2020.12.18.20248428 doi: medRxiv preprint or standing/walking were most common (women 38%; men 34%), whereas jobs involving the 188 highest levels of manual work and standing/walking were less common (women 9%; men 189 16%). Supplementary Table 8 shows the distribution of participants across OPA categories 190 within SOC strata. Participants in managerial, professional, and administrative SOCs tended 191 to report less manual work and standing/walking, whereas participants in elementary, skilled 192 trade, personal service, and operative SOCs tended to report more manual work. 193 Figure 1 shows hazard ratios and 95% confidence intervals of all-cause mortality for 194 occupational categories compared with the referent of a paid job involving no manual work or 195 standing/walking (e.g. sedentary office work). Among women and men in paid work, there 196 were no differences in hazard of all-cause mortality. The hazard of mortality was lower in 197 retired women, but twice as high in women unable to work due to illness. The hazard was 198 higher in men unable to work due to illness, unemployed men, and in men caring for home or 199 others. Additional adjustment for resting heart rate and BMI did not alter these findings 200 (Supplementary Figure 3). When OPA category was replaced with SOC group 201 (Supplementary Figure 4), men with "elementary" or "process, plant or machine operative" 202 SOCs had higher hazards of all-cause mortality than those in "senior managerial positions" 203 (the category we assumed to be most similar to sedentary desk work with large numbers in 204 both sexes), however no such associations were observed in women. Similar associations 205 were observed in the model adjusting for resting heart rate and BMI (Supplementary Figure 4). 206 Figure 2 shows hazards of all-cause mortality for tertiles of working hours per week within 207 different OPA strata. Women working 35-40 hours per week had higher hazard than those 208 working 1-34 hours per week, but women working the longest hours had lower hazard than 209 those in the middle tertile. Among men, working 35-40 or >40 hours per week was associated 210 with similarly high hazards of mortality, compared with those working 1-34 hours per week. 211 There was no evidence of interaction between working hours and OPA category (p=0.49 and 212 p=0.90 for women and men, respectively). Additional adjustment for resting heart rate and 213 BMI did not materially alter these findings (Supplementary Figure 5). There was no evidence 214 . CC-BY 4.0 International license It is made available under a perpetuity.
is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint The copyright holder for this this version posted December 19, 2020. ;https://doi.org/10.1101https://doi.org/10. /2020 of interaction between working hours and SOC category (p=0.32 and p=0.68 for women and 215 men, respectively; Supplementary Figure 6). 216 Figure 3 shows associations between non-occupational PAEE and all-cause mortality across 217 occupational strata. For those in paid employment, non-occupational PAEE was associated 218 with lower hazard of mortality in both sexes with no evidence of interaction by occupational 219 group (p=0.19 and p=0.11 for women and men, respectively). For those not in paid 220 employment, non-occupational PAEE was also associated with lower hazard of mortality in 221 both sexes with evidence of interaction by occupational group in men (p=0.02) but not women 222 (p=0.40). Following additional adjustment for resting heart rate and BMI, hazard ratios were 223 attenuated across all strata (Supplementary Figure 7). 224 The inverse association between non-occupational PAEE and mortality was reasonably 225 consistent across tertiles of working hours with no evidence of interaction observed in either 226 women (p=0.69) or men (p=0.61) (Figure 4). Following additional adjustment for resting heart 227 rate and BMI, hazard ratios were attenuated across all strata (Supplementary Figure 8). We 228 observed similar results when repeating all of the above analyses with the exclusion of 229 participants with baseline prevalent coronary heart disease, stroke or cancer (data not shown). 230

DISCUSSION 231
In this study of 452,884 women and men including 264,424 paid workers in occupations with 232 varying degrees of manual work and standing/walking, we found little evidence that all-cause 233 mortality varied by category of OPA. Working full-time rather than part-time hours was 234 associated with higher hazard of mortality but there was no pattern indicating that hours in 235 some OPA categories were more harmful than others. Retirement was associated with lower 236 mortality in both men and women but not working due to illness at baseline was predictably 237 not beneficial for survival. Non-occupational physical activity was beneficial across 238 occupational categories, supporting universal physical activity guidelines. is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint The copyright holder for this this version posted December 19, 2020. ;https://doi.org/10.1101https://doi.org/10. /2020 We found no evidence of an association between OPA and all-cause mortality after controlling 240 for non-occupational physical activity, working hours, and a range of demographic, clinical, 241 and lifestyle variables. This is somewhat in contrast to a meta-analysis of 193,696 people 242 reporting that men with high level of OPA were at higher risk of mortality than those at the low 243 level (HR=1.18, 95% CI: 1.05-1.34, I 2 =76%), and the corresponding result for women which 244 showed some evidence of an inverse association (HR=0.90, 95% CI: 0.80-1.01, I 2 =0%).
[6] 245 Our main findings from a single UK cohort are however in agreement with studies from 246 Europe [18][19][20][21][22] and the USA[23] indicating no association. However, in our SOC analysis, we 247 did observe higher all-cause mortality for "elementary occupations" and "process, plant and 248 machine operatives" in men. Discrepancies between our findings and previous work may 249 partly be explained by variation in working patterns and conditions between populations and 250 eras. For example, the strongest effect size (HR 3.40, 95% CI: 1.94-5.96) in the above meta-251 analysis is from a Taiwanese study with baseline in 1990, [24] likely not generalisable to the 252 UK between 2006 and 2010. Alternatively, our findings could suggest that in this UK 253 population, the combination of two self-reported OPA variables is insufficient to characterise 254 the intensity level of work throughout the day or week, making groups more difficult to 255 distinguish and biasing effect estimates towards the null. The SOCs for which we observed 256 higher harmful associations with all-cause mortality (including assembly line and construction 257 workers, cleaners, and drivers) are perhaps more consistent in terms of activity intensity and 258 thus better characterised, but the potential risks of the actual physical activity performed as 259 part of these occupations should be investigated further using objective measures of physical 260 activity labelled by domain. Accelerometers have been combined with work diaries to show 261 that for mostly (71%) "blue-collar" workers in Denmark, reallocating time to MVPA at work was 262 positively associated with long term sickness absence, whereas an inverse association was 263 observed for reallocating time to MVPA in leisure-time. [25] 264 In contrast to objective monitoring, self-reported categorical data do not detail the pattern of 265 work bouts intensity across each day. Although a strength of this work was calibration of our 266 . CC-BY 4.0 International license It is made available under a perpetuity.
is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint The copyright holder for this this version posted December 19, 2020. ;https://doi.org/10.1101https://doi.org/10. /2020 Characteristics like these may be patterned by occupational group, and these strata could be 294 used to investigate specific working cohorts, such as in previous studies. [10,11] 295 We used baseline data to assign occupational status and were unable to account for any 296 changes during the follow-up period. In a sample aged 40-69 at baseline, retirement or 297 changes in work due to illness during follow-up are of particular concern. We were also unable 298 to account for potentially complex work histories leading up to baseline, account for changes 299 to nature of work over time, [34] or generalise our findings to younger workers. There is also 300 evidence of a healthy volunteer selection bias in UK Biobank such that it is not representative 301 of the general population, [35] particularly in relation to smoking and education [36] which are 302 notable confounders for this study. 303 In summary, analysis of this population of UK adults aged 40-69 years old showed limited 304 evidence of an association between OPA and all-cause mortality, although potential 305 measurement error and residual confounding mean that we are unable to rule out the 306 possibility of either health benefits or risks. Until stronger evidence is available from a 307 combination of domain labels and objective assessment of the temporal pattern of activity, 308 individuals should continue to maximise their physical activity volume during leisure-time 309 irrespective of their occupation. 310

Contributors 311
MP and SB conceptualised the study. MP undertook the analyses with input from TS, SJS, 312 KW, AM, and SB. MP drafted the manuscript with critical revisions from all authors. All authors 313 approved the final version. 314 The lead author (MP) affirms that this manuscript is an honest, accurate, and transparent 315 account of the study being reported; that no important aspects of the study have been omitted; 316 and that any discrepancies from the study as planned have been explained. 317 Acknowledgements 318 . CC-BY 4.0 International license It is made available under a perpetuity.
is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint The copyright holder for this this version posted December 19, 2020. ;https://doi.org/10.1101https://doi.org/10. /2020