To investigate patterns of, and associations between, physical activity at work and in leisure time, television viewing and computer use.
4531 men and 4594 women with complete plausible data, age 44–45 years, participating in the 1958 British birth cohort study.
Physical activity, television viewing and computer use (hours/week) were estimated using a self-complete questionnaire and intensity (MET hours/week) derived for physical activity. Relationships were investigated using linear regression and χ2 tests.
From a target sample of 11 971, 9223 provided information on physical activity, of whom 75 and 47% provided complete and plausible activity data on work and leisure time activity respectively. Men and women spent a median of 40.2 and 34.2 h/week, respectively in work activity, and 8.3 and 5.8 h/week in leisure activity. Half of all participants watched television for 2 h/day, and half used a computer for <1 h/day. Longer work hours were not associated with a shorter duration of leisure activity, but were associated with a shorter duration of computer use (men only). In men, higher work MET hours were associated with higher leisure-time MET hours, and shorter durations of television viewing and computer use. Watching more television was related to fewer hours or MET hours of leisure activity, as was longer computer use in men. Longer computer use was related to more hours (or MET hours) in leisure activities in women.
Physical activity levels at work and in leisure time in mid-adulthood are low. Television viewing (and computer use in men) may compete with leisure activity for time, whereas longer duration of work hours is less influential. To change active and sedentary behaviours, better understanding of barriers and motivators is needed.
The health-damaging effects of a sedentary lifestyle are well established, yet levels of physical inactivity in the United Kingdom and many other countries are high (Armstrong et al., 2000; Hallal et al., 2003; World Health Organisation, 2006; Kruger et al., 2007). Physical activity is a key modifiable factor with a potential role to protect against many chronic diseases, including cardiovascular disease, obesity, osteoarthritis, diabetes and various cancers. Current interest in increasing physical activity is perhaps unprecedented, and from a wide range of stake-holders (Department for Transport, 2000; Department of Health, 2004). Physical activity recommendations encourage people to engage in 30 min of moderate activity on 5 days, but preferably all days of the week (Department of Health, 1995), but information on UK activity patterns is rather fragmented.
People can be physically active across different domains of their lives, for example participating in sport, walking or cycling as a leisure activity or mode of travel, performing domestic or caring activities, or as part of an occupation. The term ‘physical activity’ may have a similarly wide range of interpretations. Activity in leisure time (Paffenbarger et al., 1986; Wannamethee et al., 1998; King et al., 2001; Barengo et al., 2004), work (Stender et al., 1993; King et al., 2001; Barengo et al., 2004) and active travel (Hamer and Chida, 2008) have all been shown to influence disease risk, and yet many studies focus solely on leisure (Kaplan et al., 1996; Davey et al., 2000; Schnohr et al., 2003). Inconsistent relationships between occupational activity and disease risk or mortality have been reported, (Rosengren and Wilhelmsen, 1997; King et al., 2001) but the omission of work activity could be important, given the relatively large proportion of time that adults spend at work.
A recent systematic review of large surveys in the United Kingdom found that ‘lack of time’ was the most common barrier to taking exercise (Foster et al., 2005). Being ‘too tired’ or preferring to ‘rest and relax’ was also frequently reported. Potentially, activity level in one domain of life will influence that in another; for example, people with physically active jobs may engage in less leisure-time activity, whereas inactive leisure-time pursuits such as television viewing and computer use may displace more active ones. Little is known about these relationships, although such knowledge could inform strategies to increase activity.
Using recent data from the 45-year survey of the 1958 British birth cohort, we investigated duration and intensity of physical activity, and duration of television viewing and computer use. We examined leisure-time levels and types of physical activity, the proportion of people meeting current recommendations and how active people were at work. We also investigated whether levels of activity across different domains were related, specifically whether (1) working longer hours were associated with reduced time in leisure activity, television viewing or computer use, (2) greater activity intensity at work was related to lower activity intensity in leisure time, (3) longer duration of television viewing or computer use was related to less active leisure time and (4) longer television viewing was associated with less computer use in leisure time.
Participants and methods
The 1958 cohort includes all children born in England, Scotland and Wales in one week of March 1958 (Power and Elliott, 2006). At 44–45 years, 11 971 participants were invited to a clinical examination undertaken in their home. Contact was not attempted for cohort members who had died, emigrated, permanently refused to participate, or several other reasons (Atherton et al., 2008). In all 9377 participants were visited and 9233 (4578 men and 4645 women) provided information on physical activity. Ethical approval for the 45-year survey was obtained from the South-East Multi-Centre Research Ethics Committee, United Kingdom.
Physical activity was assessed using a modified version of EPAQ2, a previously validated self-complete questionnaire, including work, home and leisure-time activity during the past year (Wareham et al., 2002). Modifications for our study were (i) omission of home activity due to poor validity (Wareham et al., 2002), (ii) addition of ‘free text boxes’ for participants to record work and leisure activities not in the pre-defined lists and (iii) questions on television viewing and computer use were reduced. Questionnaires are available elsewhere (Fuller et al., 2006): work questions were included in a booklet posted to cohort members, (completed and returned to the nurse when she visited); leisure questions were self-completed during the nurse visit.
Two summary measures were derived for leisure and work activity; duration (hours/week) and intensity (MET hours/week). A MET (metabolic equivalent) is the ratio of the energy cost of an activity to resting metabolic rate. For work activity, participants reported hours/week for nine pre-defined activities, and how many times per day (at work) they climbed a flight of stairs or ladder. Time spent climbing stairs and ladders was calculated assuming a 5-day working week, and a time 0.002 h (per flight) and 0.004 h respectively. For leisure activity, frequency and duration for 35 listed activities were reported, after excluding ‘winter sports’ due to reporting difficulties. Eight frequency categories ranged from ‘not done in last year’ to ‘every day’. A seasonal weighting factor was applied to; mowing the lawn (0.5), watering the garden (0.33), football (0.75) and cricket (0.33). MET hours/week were calculated by multiplying activity duration by published MET values (Ainsworth et al., 2000).
Television viewing and leisure-time computer use
Participants reported their average leisure-time use for each in six categories, from ‘none’ to ‘>4 h/day’. In estimating total activity, participants were assigned a duration at the lower limit of their category (those with ‘<1 h/day’ were assigned a duration of 10 min/day), to minimize exclusions due to implausible values for total activity (see data analysis below).
Total activity (hours/week and MET hours/week): was calculated by summing work (where applicable), leisure, television and computer activity.
We scrutinized data for leisure and work activity, to identify implausible values, defined here as hours/week >3 s.d. above the mean for those with complete data. To identify cutoffs for implausible values, data were (square-root) transformed due to positively skewed distributions and values back-transformed to give hours/week for work of 132.63 for men and 117.14 for women and for leisure, 78.63 (men) and 48.74 (women). Work data for 50 men and 30 women, and leisure data for 28 men and 28 women were excluded. For total activity, those with >168 h/week were excluded (six men).
Relationships between work, leisure activity, television viewing and computer use were investigated using linear regression and χ2 tests. Work and leisure hours/week and MET hours/week were analyzed as square root-transformed variables, and work variables were divided into quartiles due to non-linearity in some relationships.
For work activity, about 10% of the sample was missing a frequency or duration; for leisure activity approximately half of all men and women had 1 frequency or duration missing. Non-responders for leisure and work were defined as missing frequency and duration for all activities (except stair and ladder climbing). Those with partial data were those who had recorded a frequency but were missing the corresponding duration, or vice versa, for any activity; the remaining cohort members had complete data.
We present leisure and work activity data based on subsets of participants with complete data for each activity type. Of the 9223 participants who provided some information on physical activity, for work and leisure activity 83 and 99% provided partial data and 75 and 47% provided complete and plausible data, respectively. To establish whether our results were affected by missing data the following sensitivity analyses were performed: (i) analyses repeated using the larger sample including those with partial data, (ii) descriptive and regression analyses repeated weighting by social class at birth and (iii) descriptive analyses repeated using multiple imputation. Mulitple imputation was performed using the ‘mice’ method (multiple imputation by chained equations) using factors related to missingness and predictors of physical activity. Ten copies of the original dataset imputed for missing data were created followed by a combined regression analysis that averages results from each dataset. Results were generally consistent; for example, median values for work, leisure and total activity in women with complete data were 34.20, 5.78 and 52.28 h respectively. Corresponding median values derived from (i) partial data were 32.05, 3.96 and 49.63 h, (ii) weighted analyses were 35.00, 5.69 and 52.39 h and (iii) imputation were 30.04, 6.90 and 50.78 h respectively. Associations were similar except that when participants with partial data were included, the difference in time spent in leisure activity between men working the longest and shortest hours was absent. Analyses were conducted using SPSS for Windows, version 14.0.0 (SPSS Inc., Chicago, IL, USA) except for multiple imputation, performed using STATA version 9.1 (STATACorp LP, TX, USA).
At age 45 years, study participants were predominantly married, with children and described their health as excellent or good (Table 1). In all 47% of women and 4% of men worked part time.
Most men (92%) and women (87%) responding to the work questionnaire had been in paid employment, or done regular organized voluntary work, in the past year. Of these, men worked a median of 40.2 and women 34.1 h/week (Table 2), and men expended a median of 79.5 and women 59.3 MET h/week. Over 90% of working men and women participated in light activities, and spent most of their time in light activity (Table 2). Almost half of men, and over a third of women, undertook moderate activities, and most men (83%) and women (71%) undertook vigorous activities (mostly climbing stairs or ladders), but this accounted for <0.5% of work time (Table 2).
Men spent a median of 8.3 h/week in leisure activity and expended 34.1 MET h/week, and women spent 5.8 h/week and expended 23.7 MET h/week (Table 3). The most popular activities were gardening (>99% participation) and walking, with DIY, swimming and cycling for pleasure popular in men, and swimming, dancing and floor exercises in women (Table 3). Over a third of men (37%) and women (36%) participated in moderate or vigorous leisure activity, at least four times per week, for 30 min per occasion (Table 3). Over 99% of participants reported watching television, with about half watching 2 h/day; approximately two-thirds used a computer during leisure time, most for <1 h/day (Table 3).
Associations between work and leisure activities
Men working the longest hours (top quartile) spent an extra 1.3 h/week in leisure activity compared with men working the shortest hours (bottom quartile) (P=0.05), although this difference was absent if participants with partial data were included. Work and leisure MET h/week were also positively associated in men (Table 4). Longer working hours in women were associated with watching less television; 16% in the top quartile watched >21 h/week of television compared with 25% in the bottom quartile (Table 4). In men, higher work MET hours were associated with watching more television. Men in the top quartile of working hours, and men and women in the top quartile of work MET hours were most likely to not use a computer in leisure time.
Associations between television viewing, computer use and leisure activity
More time watching television was associated with less time in leisure activities: men and women watching no television spent approximately three times as long in leisure activity as those watching >28 h/week (Table 5). Men using a computer in their leisure time for >14 h/week spent the least time in leisure activities, but in contrast, women computer-users spent more time in leisure activities than women not using a computer (Table 5). Television viewing and computer use showed similar relationships with leisure MET hours as for leisure hours. Men and women watching longer hours of television (>21 h/week) were less likely to use a computer, P<0.0001.
Because about half of women worked <35 h/week we examined part-time and full-time separately. Associations were similar for all women except: work hours were unrelated to television viewing in either group, and in part-time workers, leisure activity and computer use were not associated and working longer hours was associated with not using a computer.
We estimated duration and intensity of physical activity at work and in leisure time in mid-adulthood in a nationwide British sample, and investigated how work and leisure activity were related. Firstly, most men and women did not undertake moderate or vigorous activity as part of their occupation. Over 90% spent time in sedentary (light sitting or standing) activities, and for the majority, work was almost entirely sedentary. Men worked longer hours and were more active than women. Secondly, almost two-thirds of 45 year olds did not participate in regular moderate or vigorous activity in leisure time (4 times per week, for 30 min per occasion), and therefore did not meet physical activity recommendations (Department of Health, 1995). Thirdly, we expected longer working hours to displace leisure activity, but found no evidence for this in men or women. We also anticipated that greater activity intensity at work would be associated with lower activity intensity in leisure time, but found that men who accumulated more MET hours/week at work also accumulated more MET hours/week in leisure time. Fourthly, as expected, longer durations of watching television (both sexes) and computer use (men only) were associated with less leisure activity, although women who did not use a computer spent the least time in leisure activity. Lastly, as expected, cohort members who watched longer hours of television (>21 h/week) were less likely to use a computer in leisure time.
Few previous studies have collected such detailed information on physical activity across different domains of activity. Our study includes estimates of work and leisure activity, television viewing and computer use, allowing us to assess how activity in one domain influences that in another, but not home activities such as housework, caring for dependents, due to the poor validity of these measures (Wareham et al., 2002). It is inherently difficult to disentangle the inter-relationships between activity in different domains and identify causality, and the associations we describe may operate in either direction. Despite using a self-complete questionnaire previously validated against accelerometry in a slightly older age group (Wareham et al., 2002), our study participants clearly had difficulty in estimating duration of activity, particularly for less structured leisure activities such as gardening and cycling. For some leisure activities (for example, fishing and DIY) very long durations were recorded. Thus information on frequency was often reported without a corresponding duration; the proportion of missing durations (or frequencies) for any given activity was 20–37%. However, our sensitivity analyses including those with partial data, or using imputation or weighting, did not change the conclusions from the study. We acknowledge that estimating intensity of activity by assigning mean MET values assumes an equal MET value for all individuals performing a given activity. In reality, the relative energy cost may vary with age, gender, body mass, adiposity, movement efficiency and environmental conditions (Ainsworth et al., 2000). Nonetheless importantly, our study suggests that further development work is needed to refine the design and application of questionnaires, if accurate physical activity assessments are to be obtained. The questionnaire could be administered as a personal interview, or in computer-assisted format, to allow checks for extreme responses and to ensure a response for each question. A disadvantage, particularly for large population-based studies, is that these methods are more time-consuming than a self-complete questionnaire, and computer-assisted techniques require advance programming. Alternatively, an additional objective method suitable for assessing activity in domains (for example, accelerometry or heart rate monitoring together with concurrent activity diary) could be used in a subset of participants.
Our study demonstrates that a high proportion of UK 45 year olds are in work, and for most people work is sedentary. Almost all our population (>99%) participated in some form of leisure activity, most commonly gardening, walking and swimming, and it is noteworthy that these are not typically ‘sporty’ activities requiring special facilities or equipment. Yet based on leisure activities, almost two-thirds of the study sample did not meet current physical activity recommendations. Our study shows that for those in work, work contributes a large proportion of total activity hours and leisure a relatively small proportion. However, the intensity of leisure activity is greater, and for many people leisure time will provide the opportunity to accumulate moderate or vigorous activity.
For participants aged 35–54 years in the National Fitness Survey, ‘lack of time’ for increasing physical activity was largely attributed to ‘work’ (particularly in men) and ‘looking after young children’ (more so in women) (Sports Council and Health Education Authority, 1992). Work and childcare were also important reasons for stopping regular exercise (Sports Council and Health Education Authority, 1992). In our study, contrary to expectations, longer working hours and higher work intensity were not associated with reduced leisure activity. Interestingly, men who were less active at work spent less time in leisure activity, an association, which was not due to working hours. However, the few previous studies show inconsistent results (Burton and Turrell, 2000; Kruger et al., 2006; Popham and Mitchell, 2006).
Why people do not engage in more physical activity has been under-researched, but existing studies suggest that changes in life-stage, such as leaving school, having children and children leaving home are crucial points in the maintenance of physical activity (Foster et al., 2005). In our study, working longer hours was not associated with reduced leisure activity, suggesting that there is more to ‘being busy’ than work, or to ‘work reasons’ than solely a time factor preventing participation in leisure activity. Likewise, greater activity at work did not reduce leisure activity, suggesting that work is not a major physical cause of ‘tiredness’ or need to ‘rest and relax’ reported elsewhere (Foster et al., 2005). Leisure time also provides opportunity for sedentary activities and just over half of our study participants watched 2 h of television a day; computer use was lower, at 1 h a day for three quarters of participants. Men who were more active at work watched more television. Our findings suggest that longer working hours were associated with reduced leisure-time computer use (in men), and reduced television viewing (in women). Similarly, longer hours of television viewing (and computer use in men) were associated with reduced leisure activity.
Our study supports the view that active and sedentary behaviours are not simply ‘opposites’ (Biddle, 2007); we found that physical (in)activities in different domains of life do not necessarily co-occur in ways we might expect. Although it might be tempting to assume, for example, that reducing television viewing would lead to increased leisure activity, causality cannot be inferred from a cross-sectional study such as this. Furthermore, active and sedentary behaviours have been shown to exhibit independent effects on health outcomes (Aadahl et al., 2007; Wijndaele et al., 2007). How and why individuals allocate time to various (in)activities is likely to be highly complex; for example, previous work has shown that factors, which predict inactivity itself, are different from those which predict the reasons for inactivity (Owen and Bauman, 1992). To change active and sedentary behaviours to improve health outcomes, a far better understanding of barriers and motivators, real and perceived, is needed.
We are grateful to the Measurement and Epidemiology of Physical Activity group at the Medical Research Council Epidemiology Unit, Cambridge, United Kingdom for sharing information on administering and utilizing data from the physical activity questionnaire (EPAQ2). Data collection at age 45 years was funded by the UK Medical Research Council, Grant G0000934. Statistical analyses was supported by the Medical Research Council and also by the Secretary of State for Health, Department of Health, England (NHS R&D programme). The views and opinions expressed in the article represent those of the authors and do not necessarily reflect those of the Department of Health. Research at the UCL Institute of Child Health and Great Ormond Street Hospital for Children NHS Trust benefits from R&D funding received from the NHS Executive.
About this article
International Journal of Obesity (2017)