The results showed a generally high level of overall sitting time of 5 hours/day in the German population, with men sitting significantly longer than women. In both genders age and PA were negatively associated and the educational level was positively associated with sitting time. Interestingly, the level of income did not significantly contribute as an independent correlate of sitting time. Only one environmental correlate ‘Walking is unsafe because of the traffic in my neighbourhood’ was independently associated with sitting time in women. In men, no associations with environmental correlates were found. The overall variance of the multivariate model ranged from 16.5% for men to 8.9% for women.
Prevalence
The median sitting time in the German population was 5 hours per day, which represents approximately 31% of an adult´s assumed 16 waking hours a day. Regarding the 20-country comparison [
17], the results were congruent with the overall median of the investigated countries and similar to those in such investigated European countries as Belgium, Sweden or Spain. Compared to collected IPAQ data from the Netherlands, the UK and the USA, which showed sitting times ranging from 5.5 hours to >7 hours [
21], the sitting time for the German population falls within the lower range. A possible explanation could be the use of a convenience sample in the study by Rosenberg et al. [
21]. Their study consisted mainly of university staff and students with a generally high educational and socioeconomic status who may have overall higher sitting times, as also seen in the present study.
Regarding prolonged sitting times of six hours and more, the present study revealed a reduction in prevalence points of about 13.3 compared to the study sample in 2002 (30.1% vs. 43.4%) [
19]. The extent of this finding was not expected and is of crucial importance to explain it. A possible explanation could be that the study samples are not entirely comparable due to a higher mean age and slightly higher income levels in the present study. Furthermore, the low response rate in the current study has to be considered as it implies a possible selection bias of health-interested respondents who answered the survey and reported less sitting time. In addition, it should be kept in mind that the cut-off level of > 6 hours is an artificial threshold and small shifts of minutes per day for people close to the cut-off might result in large differences.
Studies using objective measurements such as accelerometers to assess sitting time detected even higher durations for sedentary behaviour, for example in the United States with 7.7 hours/day [
2], in China with 8,5 hours/day [
37], or as in Australia where participants spend 57% of their waking hours sedentary [
6]. It is well known that objective measurements have been associated with higher sedentary behaviour than self-reported behaviour. This reflects a higher sensitivity of objective measurements of overall sitting time and overcomes issues of recall bias of self-reported measures [
38]. It is not possible to compare these numbers with German populations since there is lack of representative objectively collected data. However, the sitting item of the IPAQ, which in contrast to the GPAQ distinguishes between sitting time during weekdays and weekend day, but otherwise offers the same question phrasing, mirrored reasonable agreement compared to accelerometer counts/min <100 [
21]. Nevertheless, studies using objective measurements to determine sitting time are warranted.
The significantly higher amount of sitting time among men in our study corresponds with that of past studies [
2,
39]. Bauman et al. [
17] reported higher sitting times among men in seven out of 20 countries. Contradictory findings were reported from the US [
38], indicating a lower prevalence of women in screen time, but a higher prevalence of women for ‘sitting most of the day’ than for men, resulting in a longer duration of overall sitting time for women with reference to accelerometer counts. Results from Australia pointed out that there were gender-specific dissimilarities on looking at the different domains of sitting for watching TV, general leisure and home computer use during the usual weekday and weekends [
40]. To understand these gender-specific patterns of sitting time, it is necessary to examine in more detail, i.e. screen time, non-screen time or the different domains of sitting, such as at work, in transport and during leisure to develop well-directed interventions.
Correlates of sitting time
The second aim of the present study was to explore sitting behaviour in respect to different socio-demographic and environmental correlates. Multivariate models examining the association between overall sitting time and the above-mentioned correlates explained more of the variance in men (R
2 = 16.5%) than in women (R
2 = 8.9%). From a public health perspective, the low variance might still be of significance for developing interventions of the population level to reduce sitting time. However, the results also showed that a large part of the model variance remains unexplained by the included correlates. Models including different correlates, such as social norms, psychosocial or home environment correlates (e.g. home entertainment, labour-saving devices) might be most promising for explaining sitting behaviour [
23]. Consequently, on-going research has been assigned to investigate possible correlates of sitting, considering the different types and domains of sedentary behaviour [
40] and recognizing the relevant contextual factors [
23].
The present results confirmed decreasing sitting time with increasing age for both genders and replicated recent findings [
17]. The greater use of technology, sitting occupations and passive modes of transport among younger adults could account for this behaviour. However, opposite age relationship patterns were found in other studies using self-reports [
38], as well as objective measuring tools [
2,
37]. Reasons for this discrepancy could be seen in the more challenging task of answering the self-report sitting time question for older people. This might affect the accuracy of the response [
41]. Furthermore, Healy et al. [
38] indicated a sitting domain specific age-related influence, showing increasing sitting times with age for TV viewing and screen time, but decreasing values for computer use. Therefore, ongoing research that investigates the effect of age on sitting with objective as well as domain-specific self-report data management is warranted to better identify sitting patterns related to age.
The present finding that there is no association between overweight and sitting time can partly be explained by the results of a recent systematic review [
3], which revealed only limited evidence for a longitudinal relationship between sedentary behaviour, weight gain, and the risk of obesity. Moreover, studies suggested a relationship between overweight and more specific aspects of sitting, such as TV watching [
42], but not overall sitting time as collected in the present study. Also, self-reported BMI as in the present study may lead to misclassifications, which could explain the missing association.
Studies have shown that the level of education was positively associated with sitting time [
17], especially during weekdays [
43]. This was confirmed by our results for men and women and indicates that reasonable interventions to reduce sitting time have to be developed, especially for people with higher levels of education. However, studies investigating more specific sitting behaviours indicate that people with lower education have longer TV viewing time during leisure [
40]. Interestingly, the income level was not independently associated with sitting time and fades in the model, which might be due to the fact that the correlate of income level ‘hides’ behind the educational level. Burton et al. [
40] also did not reveal an overall association of sitting time with income level, but demonstrated longer home computer-use times in the mid-income group.
Based on our current findings of the socio-demographic correlates, we can conclude that the main target groups for reducing overall sitting time are especially men and younger and more educated adults. This might be a surprising conclusion as it is different from what we know from the field of PA promotion. Future studies should focus on contextual factors considering the domain and the type of sedentary behaviour to develop effective action for high-risk groups such as men or perhaps managers, university students, office workers etc. in order to reduce sitting time. However, considering measurement issues (e.g. response bias increasing with age) and the versatile nature of sedentary behaviour as a distinct class of behaviours, future studies must identify target groups depending on their dominant sedentary behaviour instead of overall sitting time.
The present findings suggest a strong negative association between PA and sitting time for both genders. Decreasing levels of PA have been associated with increasing overall sitting time before [
5,
17,
19]. However, it has to be emphasized that the evidence is not consistent in this matter and several studies detected no association [
21], inconsistent association [
44] or even positive associations, indicating that PA and sitting behaviour are independent constructs [
20]. Keeping in mind that in the present study all domains of PA (work including household chores, transport and leisure) as well as overall sitting time were assessed it also seems reasonable that people with high PA do not report on high sitting time, because of the limited time. Especially studies looking at distinct sitting behaviours during leisure time and specific leisure PA did not find negative associations between PA and sitting [
20]. Consequently, domain-specific studies, looking at PA as well as sitting behaviour, are required.
Overall, the association between the environmental correlates and overall sitting time was weak in the present study, which may be due to the fact that the environmental questions, which were based on the ALPHA questionnaire, were developed for a PA context and not for sitting. However, we found a significant association for women between a higher perceived neighborhood safety and an increasing overall sitting duration. This finding was unexpected and may originate from a selection bias in that people with higher educational and income levels choose safer neighborhoods which was associated with longer sitting times. A further explanation could be the missing distinction of sedentary behaviour domains (household, leisure time, transport and occupation) as suggested by the ecological model [
23]. This may also be one rationale for the missing association between overall sitting time and the other environmental correlates in the present study. Here, investigations of the association between more specific sitting times, i.e. time during motorized transport and environmental correlates, could be promising [
27]. All in all, it has to be emphasized that research considering a possible association between sitting time and environmental correlates is just evolving and that future studies need to investigate the specificity of the environment (home, neighborhood, recreation and workplace environments) and the diverse domains of sitting, for example, investigating the neighbourhood environmental correlates of time sitting in cars or home environmental correlates of leisure-time sitting and screen-based entertainment sitting time [
23].
Limitations and strengths
Although the sample was representative of the German population concerning age, gender, federal state, residential density and household size, the low response rate in the study is a limitation. Nevertheless, referring to the overall decline of response rates during recent decades [
28] and considering survey research showing that no difference in empirical findings was a given characteristic of study protocols which accepted a low response rate as compared to studies with a higher response rate due to more aggressive attempts to make contact [
29], the present response rate seems acceptable and appropriate for investigating the given research question. However, the potential for a survey non-response bias or a selection bias of the health-interested population should be acknowledged. A further limitation in this study is the outcome of ‘overall sitting time’, with no differentiation between weekdays and weekend days and no domain-specific information concerning sitting behaviour. Furthermore, our information on sitting time was obtained by self-report. Consequently, our results might be biased due to misclassifications or social desirability. Future research should use both objective and subjective assessments of sitting time to capture important domain- and behaviour-specifıc sitting time information on weekdays and weekend days and to objectively measure total sitting time, as well as patterns of sitting [
38]. Another limitation in this study is the adaption of the response scale from the ALPHA questionnaire, which may have an impact on the validity of the questions and may aggravate comparability with other research including environmental correlates. Strengths of this study include the reasonably large sample size and the inclusion of correlates of multiple domains in terms of understanding health behaviours.