Are temporal patterns of sitting associated with obesity among blue-collar workers? A cross sectional study using accelerometers
This study showed that sitting in BB throughout the day and at work was negatively associated with BMI, fat percentage and waist circumference, while sitting in LB was positively associated with waist circumference and BMI, but not with fat percentage. Temporal patterns of sitting during leisure time were not significantly associated with any obesity indicator, while total time sitting during leisure showed a tendency of being positively associated with the obesity indicators.
As hypothesized, we found a negative association between BB of sitting and obesity indicators during whole day and work. The time spent in BB at work in this population was, on average, 42 min (SD 24 min). According to Fig. 2, spending 10 min in BB (0–5 min) at work was associated with waist circumference of ~97 cm while spending 40 min in BB was associated with waist circumference of ~92 cm. The associations with BB of sitting persisted after adjusting for several potential confounders, including MVPA and total sitting time. This finding suggests that those who spent more time in BB of sitting were less likely to be obese than those who spent less time in BB, independent of their total sitting time and level of MVPA. This result agrees with some previous studies on sedentary behavior such as breaks in prolonged sitting (e.g., transitions from a sedentary to an active state lasting ?1min) and obesity indicators, after adjusting for total sitting time and MVPA [18, 19]. Sitting in BB in our study could be considered as a ‘proxy’ for ‘breaking up’ sustained sitting periods sitting by various physical activities associated with blue-collar work.
Our hypothesis was also confirmed with respect to a positive association between sitting in LB and obesity indicators (i.e., BMI and waist circumference). In other words, workers who spent less time in LB of sitting were less obese and vice versa. For example, Fig. 2 shows that a worker spending 1 h at work in LB had waist circumference of ~94 cm while a worker spending 3 h in LB had waist circumference of ~99 cm. The effect of LB persisted after adjusting for potential confounders including MVPA and total sitting time. This suggests, together with the findings on BB, that the temporal distribution of sitting time is important to obesity in its own right, and that this relationship is independent of the extent of MVPA and total sitting time.
While based on animal studies, a possible and previously suggested explanation for the inverse associations of LB and BB with obesity indicators could be that prolonged sitting may lead to a loss of contractile stimulation of weight bearing muscles [39, 40]. This could suppress lipoprotein lipase activity which, in turn, could impair several aspects of lipid metabolism (such as triglycerides uptake and HDL production) [39, 40] and contribute to the development of obesity on the long term. On the other hand, frequent interruptions of sitting could facilitate lipid metabolism and glucose removal from the blood due to intermittent muscle contractions [41, 42], which may, in the long term, decrease the probability of becoming obese. Frequent interruptions of sitting by short breaks could also lead to larger total energy expenditure than fewer but longer breaks, and thus a more pronounced effect on obesity indicators, as suggested by others [18]. However, these hypotheses require further investigation, since we used a cross-sectional study design and did not measure any metabolic variables.
Most of our results concerning associations between temporal patterns of sitting and obesity indicators persisted after adjusting for total sitting time. Total sitting time per se was not significantly associated with any obesity indicators in the analyses of whole days and work, and tended to be in analyses of leisure time. These findings agree with previous studies reporting no significant associations between objectively measured total sitting time and obesity indicators such as BMI [43, 44], weight status [43, 44], percent body fat, waist hip ratio [45], and waist circumference [44]. The total sitting time is distributed in sitting periods of different durations, which may, according to our results, have different direction of association with obesity. Addressing only the total sitting time may therefore mask important associations of the structure of sitting time with obesity. Our results, suggesting that the temporal pattern of sitting is important to obesity, independent of the total sitting time, encourage interventions on the temporal pattern of sitting for preventing obesity.
Another interesting finding in our study was the lack of clear associations between the temporal pattern of sitting and obesity indicators during leisure. We found a slight tendency of a negative association between BB of sitting and waist circumference. However, it did not reach significance after adjustment for total sitting time and BB of sitting during work. On the other hand, we found a tendency of a positive association between total sitting time at leisure and obesity outcomes. It could be that total sitting time is so long during leisure (on average 5.8 h, or 65Â % of total measured time) that the temporal pattern of sitting gets less important. We also found that LB of sitting during leisure was not significantly associated with obesity indicators, as opposed to LB during work. Until now, to the best of our knowledge, no previous study has investigated this association specifically during leisure. One explanation that LB of sitting during leisure showed a weaker association could be that sitting behavior during leisure is more heterogeneous and, to a larger extent, associated with confounding factors such as eating snacks during TV viewing. This increased uncertainty of the contents of LB in leisure-time sitting which would, for statistical reasons, lead to a weaker association with any outcome.
We found that waist circumference was more strongly associated with temporal pattern of sitting, followed by BMI and fat percentage. Waist circumference is a measure of central adiposity in the relatively small visceral adipose tissue compartment, which has been shown to be closely related to physiological disturbances caused by weight gain than the total mass of adipose tissues in the body [46]. Thus it is of note that we observed a stronger association of temporal sitting patterns with waist circumference than with body fat and BMI, suggesting that temporal patterns of sitting are, indeed, relevant to obesity related health outcomes.
Methodological considerations, strength and limitations
A major strength of our study is the study population of blue-collar workers varying little in socioeconomic status but offering a great variation in sitting time, yet with a considerable average prevalence of sitting, i.e. slightly more than 50Â % of the time. Also, sitting time was measured using two accelerometers which allowed us to separate standing and lying from sitting. We also used a validated software, Acti4, discriminating activities with an excellent sensitivity and specificity [29]. Additionally, we utilized exposure variation analysis (EVA) to determine the temporal pattern of sitting. EVA is a versatile generic approach for quantifying the level and frequency of activities, as demonstrated by the use of EVA for analyses of, e.g. working postures [47] and physical activity intensities [36].
In our analyses, we adjusted for potential confounders such as MVPA and total sitting time [17, 21, 25] to identify any independent association of temporal patterns of sitting with obesity indicators. However, adjusting for MVPA and total sitting time did not change the results to any major extent. Our results also persisted after adjustment for wear time, indicating no bias due to between-worker differences in measurement time. Moreover, we also mutually adjusted for sitting variables during work and leisure when investigating their independent association with obesity indicators.
The main limitation of the study is the cross-sectional study design, which does not allow inferences about causal relationships between sitting patterns and obesity. Thus, further prospective studies assessing the direction of the association between accurately measured temporal patterns of sitting at work and leisure and obesity are needed, as a basis for discussing causation. Since our study included a convenience sample of companies with a high fraction of blue-collar workers, our results may not be generalizable to the general population of blue-collar workers in Denmark, let alone in industrialized countries in general.