In a nationally representative sample of British adults aged 60-64 y we found evidence that PAEE, MVPA, light-intensity PA and sedentary time vary by sex, past PA, health status and behaviours, and socio-demographic parameters, and that patterns of variation are similar when either objective or self-reported instruments are used to assess PA. Overall, median (IQR) objectively measured MVPA was 26.0 (12.3-48.1) and 41.0 (18.8, 73.0) min/day in women and men, respectively.
Both PAEE and time spent in MVPA were greater among men than women. Self-reported sedentary time was greater among men, but objective estimates indicated that there was no difference between genders. Obese individuals and those with fair/poor self-rated health had lower PAEE and MVPA, and longer sedentary time compared with their counterparts. Employed participants had greater PAEE and more time in MVPA but less sedentary time (objectively measured) than retired people. Those who self-reported being active in the past had greater PAEE and time in MVPA and were less sedentary than those reporting less activity in the past. Objective estimates indicated that participants spent 75% of their time at the intensity below 1.5 MET (sedentary). Domestic PA had the greatest relative contribution to the total self-report PAEE (MET-h/day), whereas among the employed, total self-reported PAEE was mainly driven by occupational PA.
Comparison with other studies
Research examining PA in older populations has mainly relied on self-report measures [17, 51–53] and only a few epidemiological studies have used objective methods specifically with older adults [20, 34, 54]. However, none of the previous studies using objective methods examined detailed PA-patterning according to various socio-demographic, behavioural or clinical parameters. The observed differences in PAEE and MVPA by sex are plausible and in keeping with the findings of previous studies which used questionnaires [49, 55, 56] or objective methods [54, 34] to assess PA.
The substantially lower self-reported sedentary time in women than men is consistent with previous studies . However, the discrepancy in sex-differences by method (non-significant difference in objectively measured sedentary time between the sexes) might be explained by the gender-specific sedentary pursuits; it is possible that not all sedentary pursuits in which women would normally engage were captured by the EPAQ2, and women did therefore not have the opportunity to report such activities. Another possible explanation would be a differential reporting bias of sedentary time, i.e. greater underestimation among women .
Parsons et al. used the similar modified version of the questionnaire in the 1958 British birth cohort who were aged 45 at the time of assessment and still in the workforce. In this other birth cohort higher PAEE was reported than in our population which supports the notion that PA declines with age [20, 51]. Further, our observation that retired persons had greater PAEE during leisure time (as ascertained by self-report) than their full-time employed counterparts is in line with a recent systematic review indicating that LTPA and exercise increase after the transition to retirement . In contrast, total objectively assessed PAEE in our study was lower in retired participants compared with full-time employed individuals, which suggests a redistribution of PA from the occupational domain to leisure time, but also an overall decrease in PA after retirement. Higher LTPA in part-time employed relative to full-time employed participants is consistent with some studies , but contrary to Parsons et al. who reported a positive association between working hours and LTPA in men. This may reflect the fact that our sample of part-time employed participants included those who had retired from their main occupation but continued to work in a different occupation. Besson et al. reported EPAQ2 results from the EPIC-Norfolk population which was of similar age when the questionnaire was administered (1998–2000) as the population in the present study (2006–2010) and found comparable levels of total and domain-specific PAEE.
Manini et al. assessed PAEE in a slightly older population [mean age 74.8 (±2.9) y, N = 302, 50% women] using the doubly labelled water technique and reported mean PAEE of 672 kcal/day. Mean PAEE in our population was 36.2 kJ/kg/day which equals 680 kcal/day (mean body weight was 78.9 kg). This suggests similar PAEE in these populations; however differences in assessment methods and population sampling procedures need to be considered. Objectively assessed levels of PAEE and MVPA in our study are lower than those observed in the InterAct study where the same method was used in 54-y old adults from 10 European countries (median PAEE of 40.5 and 44.0 kJ/kg/day in women and men, respectively; and median MVPA-time of 72.5 and 80.8 min/day in women and men, respectively) . Troiano et al. reported that accumulated mean time in MVPA assessed by accelerometer (>2020 counts per min, ~3 MET) in the NHANES- population was 12.4 min/day in women and 16.7 min/day in men aged 60–69. However, when only MVPA occurring in 10-min bouts or longer was included, these estimates were as low as 5.8 and 6.5 min/day in women and men, respectively . Both total accumulated and bout-based estimates are considerably lower than those observed in our study. Davis et al.  examined PA by accelerometry in 163 European women and men (Better Ageing project), aged over 70 and reported a mean PAEE estimate of 16.8 and 20.1 kJ/kg/day in women and men, respectively and mean MVPA of 16.7 and 23.8 min/day in women and men, respectively. The observed differences may reflect different methods used to assess PA or unequal wear time and period of day during which the monitor was worn. In the NHANES and the Better Ageing project, a uniaxial accelerometer was worn around the waist during awake hours over 7 days , whereas in our study, a combined HR and acceleration sensor was worn for 5 days. In addition, hip-worn accelerometers do not capture activities such as cycling or upper body movement well, whereas the combined sensing performs better at distinguishing intensity across most activities.
With respect to intensity distribution, our finding that over 75% of the time was spent below 1.5 MET, is congruent with the results of Evenson et al. suggesting that adults aged 60–69 spend most of their awake time sedentary. Mean sedentary time during awake hours in the NHANES was 617 and 569 min/day in women and men, respectively . Assuming 8 h of sleep, these findings are comparable to ours. Lower PAEE and less time in MVPA in current smokers than non-smokers is in keeping with the results of the EPIC-Norfolk  cohort showing a greater proportion of inactive adults among current smokers, which highlights the fact that unhealthy behaviours are likely to cluster and suggests the need to promote PA as part of general lifestyle modification advice in this population subgroup.
Several other studies have examined the patterning of objectively measured PA by socio-demographic factors. Cleland et al. indicated that Australian adults (mean age 31 y) with manual occupation and lower education had 5.8 times higher pedometer-assessed PA than their counterparts. However, it is difficult to directly attribute the observed differences to age alone, given the different methods used to assess PA. The inverse relationship of BMI with PAEE and MVPA is in line with our previous findings from the EPIC-Norfolk cohort which showed a significant cross-sectional relationship between body weight and inactivity, and demonstrated that weight gain over time is associated with future physical inactivity . Although BMI is an indicator of overall body composition, it does not distinguish between fat and fat-free mass, but PA in this age-group has been found to play a major role in the preservation of fat-free mass [62–64].
Strengths and limitations
The major strength of our study lies in the combined use of objective and self-reported measures to obtain complementary information on PA. Objective monitoring gives estimates of total free-living PAEE and time spent at different intensity levels, whereas questionnaire data provide valuable information on the type and context of activity and allow an estimation of domain-specific PAEE and contribution of different domains to total PAEE. In addition, our sample is unique as it is nationally representative of British adults aged 60–64 at the time of assessment , which is a population of particular public health interest given the changes in lifestyle and health occurring with transition to retirement. Moreover, we examined the patterns of time spent in light-intensity PA which adds to the existing knowledge of PA in this population and provides important information for the design of PA-interventions tailored for people in later life. Lastly, the birth cohort design allowed us to explore the variation of PA-measures by past PA, independent of the confounding effect of age.
However, important caveats should be considered in the interpretation of our findings. Firstly, objectively assessed sedentary behaviour includes sleep, which did not allow us to make conclusions about the amount of time spent sedentary during awake hours. Secondly, the possibility of social desirability bias in self-reported PA and sedentary time, as well as bias arising from assigning energy cost to these behaviours , cannot be excluded. In addition, the possibility of reverse causality and/or reporting bias cannot be ruled out in cross-sectional associations with health-related factors including BMI, long-term limiting illness or disability, and self-rated health.
The interpretation of how our findings relate to current UK-guidelines for PA  is not straight-forward. Since we did not have a full week of free-living PA-monitoring, we cannot accurately assess total weekly duration of MVPA on which the guidelines are based (≥150 min/week). If we extrapolate our objectively measured MVPA-estimates per day into weekly values assuming that MVPA is pursued all days of the week, the findings broadly suggest that 55% of women and 69% of men in our sample meet UK-guidelines for MVPA (≥150 min/week). If however, we use the common interpretation of the guidelines that MVPA should exceed 30 min/day, then our results indicate that 43% of women and 60% of men meet the recommended level. Furthermore, current guidelines also state that adults should perform muscle-strengthening activities at least twice a week. If we count the individuals who report doing ≥1 h/week of weight-training on top of accumulating ≥30 min/day of objectively measured MVPA, the proportion of participants meeting this combined target is only 2.2% (1.8% among women and 2.4% among men) indicating a low prevalence of sufficient activity. This proportion would be even lower if any further restrictions on MVPA time accumulated in bouts (e.g. lasting 10 min or longer) were imposed.
This study demonstrates the utility of the combined use of self-report and objective monitoring to assess the levels and context of PA in early old age. Our findings indicate a detrimental cross-sectional relationship of poor health with PA which may in turn lead to reduced independent living and loss of function. Promoting PA earlier in adulthood and maintaining it in later life should be considered as the means to achieve health benefits and reduce the growing healthcare spending associated with deteriorating health in older individuals. Since PAEE and time in MVPA were materially lower among retired participants compared with the employed, encouragement to substantially increase LTPA needs to be given to retired adults. Given the robust evidence that physical inactivity, overweight and obesity are risk factors for exit from paid employment via disability pension [66, 67], particularly in older workers , PA-promotion in the working population should be considered as a primary preventive effort to reduce premature exit from the labour market and improve sustainable work ability, as well as establishing healthy habits before retirement. A recent meta-analysis has suggested that PA-promotion in this age group has some effect on increasing PA . Our findings highlight specific population strata which could be targeted for such PA-promotion efforts and further demonstrate that individuals in this age-group spend a large proportion of their time in light-intensity PA (1.5-3 MET). The variation within the light PA category warrants further research to clarify potential health benefits.