The primary hypothesis being tested is whether habitual levels of PA are associated with obesity in countries spanning the epidemiologic transition, using both objective and self report PA. This combination allows for the assessment of patterns of PA, both by intensity (moderate and vigorous) as well as day of the week (week versus weekend) and finally domains of PA (work, travel and recreation). As expected, the prevalence of obesity was significantly different among our 5 sites, with adults from Ghana presenting with the lowest rates (38% overweight/obese) and US adults with the highest (78% overweight and obese). This is very much in keeping with the current published prevalence rates  and confirms that our samples are characteristic of their populations.
Overall, men accumulated almost 20 minutes more a day of objectively measured moderate and vigorous PA (MVPA, in 1-minute bouts) than women, with Ghanaian and South African men accumulating the greatest amount. Notably, the site averages for MVPA among the men were all in excess of the current WHO recommendations  for PA, while none of the site averages for the women exceeded 30 min/d for MVPA. Women from Ghana were the closest to the recommended goals, accumulating 26 min/d, while US women accumulated only half of the recommended amount of daily PA (15 min/d). This sex difference in daily PA has previously been reported. For example, Cook et al. (2009) found that rural black South African men accumulated significantly more objectively measured pedometer steps than women. In this study, however, the majority of the participants (66%) were classified as physically active (>10,000 steps/d), with only 8% of these participants were classified sedentary (<5000 steps/d). Our data is however, in line with data published by Hallal et al.  reporting that US men accumulate 33 min/d of MVPA and US women around 19 min/d of MVPA.
The lower daily PA among the women from the US, is in keeping with the premise that rural dwellers are more physically active than urban dwellers, or/and that populations in developing countries are more active than populations in developed countries [27–29]. Dumith et al.  performed a meta-analysis using self reported PA data collected using the IPAQ instrument, in 76 countries, spanning the Human Development Index (HDI). Physical inactivity was higher amongst women compared to men, with differences being greatest among men and women from low HDI countries. Results from the meta-regression indicated that countries with a greater prevalence of physical inactivity, increased as a function of HDI. Similarly, the impact of urbanization on PA has repeatedly been reported. Both Assah  and Cook  compared rural and urban African populations and in both instances found that the total PA was higher in the rural populations.
In an effort to understand the HDI differences in PA, we explored week versus weekend MPVA. In our study; >90% of men from South Africa were classified as participating in manual labor, compared to only 54% of men in the Seychelles and it could be expected that MPVA would be significantly different according to the day of the week, i.e. week versus weekend, for South Africans at the very least. However, we found that there was no difference in the amount of MVPA accumulated by the most active men in our study (i.e. Ghana and South Africa, Figure 4. It should be noted here, that this may be an artifact of the instrument itself, since it is known that accelerometry may not be sensitive to any increases in energy expenditure during activities, such as weight-bearing manual labor, walking up stairs or activities in postures requiring significant isometric muscle activities [30–32]. On the other hand, this observation might be real, and in the case of South African and Ghanaian men, at the very least, there may be some continuation of casual manual labor on weekends, and outside of the regular work week. Instead it was the men from the US and Seychelles, who completed between 7–10 minutes less on the weekend days, suggesting that they are accumulate their PA from work and travel.
This pattern persisted for the women, where women from South Africa and Ghana accumulated similar amounts of MPVA, irrespective of the time of the week, while women from the US, Jamaica and Seychelles, accumulated significantly less MVPA on the weekend. This difference was most striking amongst the Seychellois where their weekend MVPA time was reduced by more than 50% (27.9 vs. 13.4 min/d) and be in line with a higher HDI country status, where the work week is traditionally Monday through Friday. Due to the limitations of the GPAQ instrument, we were unable to separate out week versus weekend PA domains, which may have provided insight into differences between upper and low HDI weekly PA patterns. Our results, however, are different from the South African study by Cook , where participants accumulated significantly more steps on the weekend days versus week days and these study differences may reflect the different settings. All of the participants in that study were from rural villages, although in our study the participants from Ghana were also from one rural village and here we did not find differences among the different days of the week.
With regards to PA patterns and adiposity, we found that lean participants accumulated significantly more MVPA than overweight and obese participants. This finding has previously been reported in a number of studies [33–36]. Among the men, the difference between lean and obese men was as much as 17 minutes on week days (47 vs. 30 min). The difference was not as marked for the women, where lean women accumulated only 6 minutes more than obese women on week days (25 vs. 19 min, p < 0.001); and is probably a result of the overall lower objectively measured PA among all the women. In the South African study, it was found that obese participants walked on average almost 2000 steps per day less . Previous studies have estimated 2000 steps to be approximately 1 mile per day , or approximately 15 minutes of walking according 2009 National Household Travel Survey . This is of similar magnitude to the difference found among the men in the current study (17 min/d). Interestingly however, regardless of BMI status, the magnitude of the difference between groups was similar for week versus weekend days and it is apparent from our data that our participants were not as active on the weekends, at least in Jamaica, the Seychelles and USA, all considered middle and upper HDI countries. This is in contrast to the South African study, which found that participants were significantly more active on Saturdays as a result of increased travel physical activity (walking).
This may have important implications for public health messaging, where current WHO and US Surgeon General PA [26, 39–41] guidelines suggest that adults accumulate 30 minutes of MVPA on most days (5 or more) and in bouts of 10 minutes or more [8, 10]. It is apparent from our objective PA data that, and at least among the men in our study, that most already accumulate 30 minutes of MVPA, at least in 1-minute bouts, verified by objective monitoring and in spite of their BMI status or whether they were engaged in manual labor or not. Even amongst the obese men (>30 kg/m2), the mean weekly accumulated MVPA was 30 min, in 1-minute bouts. When the data is analyzed in 10-minute bouts, only South African men meet PA guidelines. While among the women, none of the 3 BMI groups achieved an average of 30 minutes of objectively measured MVPA per day in 1-minute bouts; however, those in the lean and overweight groups did get close to 25 minutes. When using the 10-minute PA guideline, women in Ghana, South Africa and the Seychelles all accumulate ten minutes of daily MVPA. Although the aim of our study want not directed towards a formal validation of the GPAQ, we noted substantial differences in the amount of MVPA between the instruments. This raise questions about the accuracy of the prevalence estimates of the proportion of sufficiently active individuals obtained by the GPAQ instrument from these countries and locations. For example, using self report data, the men in this study all accumulate well over 60 minutes of daily PA, with those in Ghana and South Africa, reporting over 4 hours. The agreement between the GPAQ instrument and objective MVPA in our study, among the men, were all in excess of 0.2 (p < 0.001), indicating a moderate association, whereas the agreement amongst the women were all very low and only significant among women from Ghana, Jamaica and US. Despite this, women from all 5 sites reported accumulating in excess of 30 minutes of PA per day. One of the mainstays of the global public health response to deal with the world-wide obesity epidemic is to increase overall daily PA, to the point of 30 minutes on most days for health [26, 39, 40] and 60 minutes a day to prevent obesity . To be clear, both the WHO and US surgeon general PA guidelines indicate that the accumulation of PA is not restricted to only leisure or recreational time, but may include “leisure time PA, transportation (e.g. walking or cycling), occupational (i.e. work), household chores, play, games, sports or planned exercise, in the context of daily, family, and community activities” [26, 39]. It is apparent, at least from our data comparing 5 different HDI settings, and among the men, that most already accumulate close to or even more than the recommended amount of daily PA and this provides support for the need for either revisiting the dose of PA recommended for health benefits and to focus more public health messaging on energy intake, in addition to PA.
The large discrepancy between objectively measured PA and self report PA in our study is consistent with observations by Cook et al. , who in fact found fairly low levels of sedentarism (8%) among South African adults from a rural setting, compared to previously published reports of sedentarism, using self-report data, of between 34-39% among men and women, respectively. This highlights several issues related to trying to capture PA among participants from settings outside of the developed world, particularly for multi-country studies such as ours, which rely on only self report tools (e.g. GPAQ) to assess PA. It may be that respondents are simply better able to report leisure time PA .
Finally, this discrepancy highlights the importance of using objectively measured PA; Troiano found that for the US population, approximately 50% of the adults reported meeting the Surgeon general’s guidelines on PA (accumulate moderate of vigorous aerobic activity for at least 30 minutes/day), while the objective data indicated that far less than 5% of the adult population met this guideline . It is clear from our data, that objectively measured and self-reported PA may provide different results. Cook et al. has suggested several reasons for this disconnect; firstly it is not uncommon for the misclassification of sedentary and light activity as moderate activity . Secondly, the time spent being PA may be overestimated, for e.g. the GPAQ probes PA measured in bouts of 10-minutes and it is not unusual for adults to round up their estimated time. In addition, there are activities which are poorly captured by an accelerometer (i.e. weight bearing activities, such as during construction work) which add to the discrepancy between assessment methods but also highlights that not all discrepancy is due to the limitations of self-report; these methods can be seen as complimentary. Nonetheless, if the goal of WHO PA guidelines is to encourage people to get more physically active and intervention efforts were targeted only at inactive individuals as per self-reported activity, we would likely be missing substantial proportions of the population when the majority of adults are already reporting fairly high levels of PA. With this in mind, population-wide efforts to increase physical activity may be a safer strategy.