To our knowledge, our study is the largest in which posture-discriminating accelerometry was used to objectively measure total amount and patterns of sedentary behaviour in a sample of adults comprising participants with type 2 diabetes, IGM or NGM. The results showed that participants with type 2 diabetes had the most sedentary time, up to 26 min more per day compared with participants with IGM or NGM. Each extra hour of sedentary time was associated with increased odds for type 2 diabetes of 22%. No statistically significant differences were observed between participants with NGM and those with IGM. More time spent sedentary was also associated with a 1.13 times higher odds for one to two metabolic syndrome criteria and a 1.39 times higher odds for the metabolic syndrome (three to five criteria), independent of higher intensity physical activity. The number of sedentary breaks per day was highly comparable and not significantly different between the participants with type 2 diabetes, those with IGM and those with NGM. A statistically significant difference in the number of breaks was seen between participants with the metabolic syndrome and those without the metabolic syndrome, but the difference was only two breaks per day. Also, the odds ratio for the metabolic syndrome was statistically significant but small, and therefore not clinically relevant. Similar results were found for average sedentary bout duration: a statistically significant difference in bout duration was seen between participants with the metabolic syndrome and those without the metabolic syndrome, but the difference was less than 1 min. Also, the OR for the metabolic syndrome was statistically significant: a 1.09 times higher odds for three to five criteria when the average sedentary bout was longer. Last, the number of prolonged sedentary bouts did not differ between the groups according to diabetes status or the groups according to the metabolic syndrome, and the odds ratios were not statistically significant.
A major strength of our study was the use of posture-based measurement with the activPAL accelerometer, worn on the thigh, which has been shown to be a highly accurate method for assessing sedentary behaviour [20, 21]. Thus, our estimates of the amounts of sedentary time and the determination of sedentary breaks are more accurate than data based solely on acceleration, which cannot discriminate between postures. Further, we used accelerometry in a large sample of middle-aged and older adults with type 2 diabetes and IGM, which enabled us to examine and quantify associations of several objectively measured sedentary behaviour variables with type 2 diabetes. Also, waterproofed attachment of the activPAL on the thigh enabled us to collect 24 h accelerometry data, which not only resulted in complete data assessments, but could also have improved wear-time compliance, as demonstrated in a recently published study in children [27]. Wear time in our study population was, on average, 6.3 days with 15.7 h of waking time, and 85.3% of our study population provided at least 6 valid days of data. Another strength was adjustment for important confounders including BMI and high-intensity physical activity, which excludes the possibility that these factors account for the associations of sedentary behaviour with type 2 diabetes and the metabolic syndrome. The adjustment for BMI predominantly caused the differences between models 1 and 2, yet BMI could be part of the pathway between sedentary behaviour and type 2 diabetes. Consequently, the analyses could have been subject to overadjustment.
A few limitations should also be mentioned, of which study design is the most important. As our analyses were cross-sectional in nature, causal relationships could not be examined. It may therefore be possible that participants with type 2 diabetes had more sedentary time because of their poorer health. However, when participants with type 2 diabetes on insulin medication (who may be considered to have more severe type 2 diabetes and could for that reason have more sedentary time) were excluded from the analysis, the results did not change. This may suggest that sedentary behaviour at least partly preceded type 2 diabetes, as the associations were similar among participants who did not necessarily have to spend more time sedentary because of their health (data not shown). Furthermore, previous prospective studies have demonstrated that sedentary behaviour predicts markers of insulin resistance [5, 13]. Taken together, these findings support the hypothesis that the direction of the association is predominantly from sedentary behaviour to health outcomes, although large-scale prospective studies are warranted to provide better insights into the directions of the associations. Further, although we adjusted for a broad range of confounding factors, it is possible that some unmeasured factors, for example dietary intake, partly explain the associations. Finally, sedentary behaviour was measured during 1 week only, and this may not truly reflect habitual behaviour.
Several previous studies have used accelerometry to objectively measure sedentary behaviour and examined its associations with the metabolic syndrome. In line with our results, larger amounts of sedentary time have been associated with metabolic risk, although the reported effect sizes were smaller [8, 11, 15–17]. To date, no studies have reported associations between objectively measured sedentary time as exposure variable and type 2 diabetes as outcome measure. However, a meta-analysis of studies with self-reported measures of sedentary behaviour showed a risk of 112% for type 2 diabetes in the group with the highest compared with the lowest amounts of sedentary time [28]. Furthermore, large amounts of objectively measured sedentary time have been associated with (markers of) insulin resistance [5, 6, 10] and, as mentioned earlier, the metabolic syndrome. As these factors are precursors to type 2 diabetes, the results of these studies in combination with the results of the meta-analysis may support our findings of an increased risk for type 2 diabetes with increasing amount of sedentary time. Physiological mechanisms that could explain our findings have not yet been studied extensively, but results from animal studies suggest that responses to contractile (in)activity of muscle cells can play a role in glucose metabolism as reductions in lipoprotein lipase (LPL), an enzyme that contributes to the metabolism and transport of lipids, were seen after periods of inactivity [29, 30].
Few observational studies have reported associations between patterns of sedentary behaviour and metabolic health, but the findings that have been reported are inconsistent. Some studies have shown associations between the number of sedentary breaks and metabolic risk factors [6, 15, 18], while others have not [5, 16]. These inconsistencies could be caused by different methods for measuring breaks, which can be based on change in acceleration or on change in posture. Also, differences exist in the definition of a break, which can be any interruption of sedentary time or interruptions of at least 1 min, although we did not find different results for sedentary breaks of any duration and breaks of at least 1 min. Further, comparability of studies is hampered by differences in adjustment strategies and, furthermore, associations could be different among younger and older adults and among adults with and without (a higher risk for) type 2 diabetes, because their metabolic profiles differ. Studies on prolonged sedentary bouts and sedentary bout duration are scarce and no study on associations with type 2 diabetes has been reported. However, the studies of Bankoski [15] and Healy [7] show no statistically significant associations of bout length or number of bouts ≥30 min with metabolic variables, except waist circumference. Further, in our study we used three measures for expressing the sedentary behaviour pattern, but other measures could also be used to study the pattern of sedentary time [31]. In order to compare studies examining associations of numbers of both sedentary breaks and bouts, future studies should use similar measures, and ideally adjust for similar confounders.
To conclude, this was the largest study that objectively measured total amount and patterns of sedentary behaviour in a sample of adults with type 2 diabetes, IGM or NGM. The results showed that an extra hour of sedentary time was associated with increased odds of 22% for type 2 diabetes and of 39% for the metabolic syndrome, independent of high-intensity physical activity. The pattern in which sedentary time was accumulated, as expressed by number of sedentary breaks, number of prolonged sedentary bouts and average sedentary bout duration, was only weakly associated with an increased risk for the metabolic syndrome. Future studies in participants with type 2 diabetes should be conducted to confirm our results, and to explore dose–response relationships and causality. Nevertheless, our findings could have important implications for public health as they suggest that sedentary behaviour may play a significant role in the development and prevention of type 2 diabetes, independent of high-intensity physical activity. Consideration should be given to including strategies to reduce the amount of sedentary time in diabetes prevention programmes.