A systematic review of correlates of sedentary behaviour in adults aged 18–65 years: a socio-ecological approach

Of the 74 studies included, 62 examined the relationship between sedentary behaviours and individual factors. Four categories of factors were identified: behavioural (lifestyle, physical activity and sedentary habits (n = 30)), physical, biological and genetic (age, gender, body composition, health status and medication (n = 26)), psychological (stress and depressive symptoms, attitudes and perceptions (n = 25)) and socioeconomic status (educational levels, employment/occupational status and income (n = 23)). All individual factors were assessed using self-report questionnaires apart from some of the physical, biological and genetic factors (e.g., body mass index and heritability) that were measured objectively. Table 2 provides a detailed overview.

Thirty studies examined lifestyle factors: alcohol consumption (n = 7), food intake (n = 5), smoking (n = 7) and physical activity (n = 17). Alcohol consumption was found to be unrelated to sedentary behaviours in the three of the five studies [23‐25] that examined its correlation as an individual factor. The remaining two found it to be positively associated with time spent sedentary in transport (driving) [26] and to overall weekend sedentary time [27]. Similarly when combined with diet, it was shown to have a positive association [28, 29] to TVSE and overall leisure sitting time. Four studies investigating food cravings, snacking and high calorie snacking found that sedentariness was highly associated with all four [27, 30‐32]. In six of the seven studies investigating smoking, it was shown to be positively associated with sedentary time as measured by TVSE, time spent driving and total sitting time [24‐29, 33]. In terms of physical activity, 18 studies examined its relationship with sedentary behaviour. The majority looked at overall physical activity levels (n = 11) and found there to be an inverse association [24, 25, 31, 34‐38]. This was also the case for the three studies that explored levels of physical activity outside of work-time [34, 39, 40]. One study examined the association between retirement and physical activity levels in a number of different sedentary domains (TVSE, leisure reading, occupational sitting and domestic sitting) and reported no correlation between it and any of the domains [41]. Similarly, a lack of physical activity [42], vigorous physical activity [23] and total time exercising [43] were not significantly associated with sedentariness. One study conducted a four-armed randomised trial investigating whether (i) supervised exercise, (ii) supervised exercise with advice to decrease sedentary time or (iii) advice to decrease sedentary time and increase non-exercising physical activity levels [44] would change total sedentary time, as measured by an inclinometer. Results revealed that structured exercise was ineffective; only those in the group that were given advice to try and change their sedentary behaviours and increased their daily physical activity levels showed a significant change in total sedentary time. Finally, sedentary habits such as TV viewing and cell phone use (gaming and suing wifi) were found to be positively associated with total time spent sitting [24, 34, 45] and TVSE [36, 46].

Twenty-six studies investigated physical, biological or genetic factors with all of them evaluating age as a correlate of sedentary behaviour. Fourteen of twenty studies supported a positive relationship between age and sedentary behaviours (the older the person, the more sedentary). Eight studies looked at total sitting time, five of which positively correlated with age. Overall the results were mixed between positive correlations and no significant correlations. No studies reported a negative correlation and furthermore findings could not be differentiated by their sedentary measurements.

Gender was investigated in 19 studies. Ten reported the female gender to be inversely related to sedentariness [23, 37, 43, 47‐53] with two reporting males to be more sedentary when associated with total time spent in front of a computer screen and overall leisure sitting time [53, 54]. The remainder found little or no association with either gender [12, 33, 45, 55‐57]. The majority of studies reported gender differences defined sedentariness as total sitting time [43, 47, 49‐51] or TVSE [23, 48] with one using accelerometry [37], one using heart rate [56] and one reporting barriers to changing sedentary time in Oman via semi-structured interviews [52]. Four studies reported the male gender to be positively associated with sitting time in transport [53, 58‐60].

The relationship between sedentary behaviour and body mass index (BMI) was evaluated in 25 papers, the majority of which investigated its association with leisure screen time. Seventeen of these studies [23, 25, 29, 31, 32, 34‐36, 49, 55, 56, 60‐65] reported a positive relationship with the remainder showing no correlation. Nine studies examined this association using total sitting time and two used accelerometers as an objective measure [30, 37] and one heart rate [56]. The two studies that used accelerometry reported no significant relationship between BMI and total sedentary time while the majority of the remaining studies showed that the higher the BMI the higher the level of sedentariness. Two studies looked at the association between occupational sitting and BMI and reported a positive association [54, 66]. Overall results suggest that there is a strong relationship between increased BMI and higher level of sedentary behaviours.

Chronic diseases (e.g., diabetes, cardiovascular disease) were shown to have a positive relationship with sedentary time in three of the four studies [25, 32, 67] that included them. In none of the studies were they the primary factors being investigated. In contrast, illness, previous surgery, disability and injury were shown to have no significant correlation to total sitting time [24, 30, 47, 68]. One study investigated the role of heritability [69], one pregnancy [70], and another the role of hormone treatment and medication [25]. All three reported significant correlations with sedentariness.

Fifteen studies included psychological factors. However few investigated more than one factor. Five studies investigated depressive symptoms and four found that symptoms of depression, anxiety and tension were positively related to total screen time [29, 31, 42, 57]. Similarly perceived stress levels [26, 27, 42] and perceived tiredness [47] were also positively associated, whereas perceived health [25, 31] and perceived benefits of reducing sedentary behaviours [53‐55] were found to be inversely associated with sedentariness as measured by occupational sitting, TVSE and total sitting time. One study investigating perception of personal appearance and content with body image found no relationship to sedentariness [71]. Rhodes et al. [72] investigated whether planned behaviour is related to sedentary behaviours. They found mixed results; attitude and intention were negatively correlated with sedentariness as measured by total sitting time while perceived control and norm were not. Conroy et al. [45] also reported that intention and habit in terms of regulating sedentary behaviour were negatively associated. Overall the limited available evidence is supportive for a positive relationship between perceived feelings of depression, stress and anxiety and TVSE and a negative relationship between sedentary behaviour and planned behaviour to overcome sedentariness.

Twenty-two studies investigated educational levels and their relationship to sedentary behaviours. Nine examined TVSE, eleven used total sitting time or occupational sitting, two total leisure sitting time and one accelerometry for total sedentary time [22]. Of the nine that examined the correlation between educational levels and TVSE, eight reported significant inverse correlations [9, 22, 23, 31, 40, 42, 48, 55] and one reported no significant relationship [63]. In terms of total sitting time, five studies focused on occupational sitting as a domain of total sitting found there was a positive relationship with educational attainment [10, 22, 40, 53, 54] whereas the studies that investigated total sitting time without classifying domains found it to be negatively correlated. The exception was total sitting time as measured by actigraphy; it was found to have a positive association with educational attainment [22].

Occupation and employment were explored as a potential factor in seventeen studies. In relation to TVSE, it was positively related to unemployment while the opposite was true for employment (negatively correlated) [9, 12, 23, 31, 48, 55, 63]. Storgaard et al. [12] investigated employment in relation to all leisure sitting time as opposed to just TVSE and reported it to be similar to TVSE; positively related to unemployment and negatively related to employment.

Type of employment was reported in some of the studies. Manual employment, investigated in four studies [22, 37, 38, 73] was positively correlated to sedentariness outside of work where sedentariness was measured as total sitting time both subjectively and objectively. In contrast, working in an office was more likely to result in less sedentary time outside of work. Chau et al. [38] and Jans et al. [74] reported increased total sitting time associated with working in an office. Thorp et al. [75] found call centre employees to be more sedentary during the working day than customer service workers. Five further studies exploring type of occupation showed that those in professional roles were more likely to have a higher level of occupation sitting than those in non-professional positions [22, 27, 40, 53, 54].

Five studies focused on whether sedentary behavior differed based on work and non-work time days. Overall, work days corresponded to more sedentary time [39, 74‐77] and a greater amount of time spent in prolonged sitting, when compared to non-work days [75, 76]. Saidj et al. [53] reported the more sedentary the occupation type the higher the association with increased sedentary time in other domains (work, transport, leisure, screen time) during weekdays but not during the weekend.

Total sitting time, sitting time during work and traveling to and from work was significantly higher for full-time workers than for part-time workers [27, 66, 74]. Also, work commitments as barriers for physical activity were inversely associated with reported time spent sedentary [47, 52, 67]. Munir reported that vigor at work was associated with less occupational sitting in men and women, but the association was unclear across gender for absorption at work, dedication or job performance [66]. In women, change at work, return to study or new work was associated with an increase in total sitting [68].

Other factors relating to employment that were investigated were retirement and studying. From the retirement perspective, four studies found that retirement resulted in an increase in sedentary behaviour [41, 68, 73, 78]. Retirement was associated with an increase in total [68, 78], leisure SB (screen, reading, total) and domestic sitting [41, 73]. Van Uffelen et al. [24] reported a significant increase in overall sitting time in women who were studying.

Sixteen studies examined the relationship between income and sedentary behaviours using various measures of sedentariness. One study reported a positive association with sitting time spent in transport [77] and three studies that focused on occupational sitting time [22, 40, 54] also found a positive relationship. Of the remaining studies, TVSE as a sedentary measure was used in seven studies, five of which reported a negative relationship [22, 30, 40, 48, 73] while the remaining two showed no correlation [9, 63]. Six studies examined income and its relationship to total sitting time [10, 24, 30, 37, 59, 79, 80]. Finally one study focused on leisure sitting time [28] and found it to be inversely related. In terms of total sitting time and its relationship to income, all studies but one [10] reported a relationship, mainly in a positive direction. Finally in terms of objectively measured total sedentary time and its relationship with household income, both Kozo et al. [59] and Stamatakis et al. [22] found a clear positively association.

The relationship between interpersonal factors and sedentary behaviours was examined in 22 studies, 14 were cross-sectional; 3 longitudinal, 3 prospective and two were qualitative. Two domains of interpersonal factors were identified, family-related (marital status, living arrangements, family functioning, number of children, family commitment (n = 17)) and social factors (social norms, social interaction, cohesion, support and participation, sense of community; (n = 7)). All interpersonal factors were assessed using self-administered or interview-administered validated questionnaires. Table 3 provides detailed findings relating to interpersonal factors from the 22 identified papers.

Eight studies investigated the relationship between “marital status" and sedentary behaviour. In Japanese adults, Ishii et al. [63] reported that unmarried subjects were likely (odds ratio [OR], 2.02; 95 % CI, 1.32–3.10) to spend more time in TVSE than married subjects (>14 h/week). Van Uffelen et al. [24] found sitting time to be significantly higher in single women [24] while van Dyck et al. [60] showed adults living with a partner [60] sat less. Another study conducted in Hong Kong reported contradictory findings. Xie et al. found TV viewing time to be higher in married persons [23].

No relationship was found between TV viewing and marital status in a group of low-income women [42], neither between occupational sitting and marital status in men or women [66]. Clark et al. [68] using a prospective study design investigated the relationship between life events and sedentary behaviour and found change in marital status was not associated with changes in total sitting [68].

In terms of “living arrangement”, whether someone lived alone or with others was not associated with screen time [63] or TV viewing [9]. One study did however report that men living alone were more likely to watch TV for 4 h/day or more [48]. Only one study investigated the association between TV viewing time and lower “family functioning” (likert score) in a group of low-income women and found a positive correlation (r = 0.28, p < 0.01) that remained significant in a multivariable model including stress and depressive symptoms [42]. The impact of the “number of children” on sedentary behaviour as assessed by TV viewing time was not significant in Li’s study in low-income women [42], nor was it significant in a longitudinal study by Ding [9]. In contrast, several authors found that that overall sitting time was lower with more children [24] or with the birth of child [24, 53, 68, 71]. Kozo et al. investigated several sub-types of sedentary behaviour and found that having no children was related to more TV/video viewing, computer/Internet use, sitting and talking with friends or listening to music, total sitting time or accelerometer-measured sedentary time [59]. Two studies found a positive association between number of children and transport sitting time [59, 77]. Family commitment defined as providing care for other members of the family was investigated in the Australian Longitudinal Study on Women’s Health [24]. It revealed that women who cared for others spent less time sitting, particularly younger women. Similarly, having more dependents was associated with decreased occupational sitting time in men and women [66]. In contrast, Salmon et al. reported family commitments as a factor that decreased physical activity resulting in an increase in total sedentary time [47]. This finding was further supported by George et al. [67] and Mabry et al. [52] in their qualitative studies. They both reported family commitments as a barrier to decreasing sedentary time. Taken together, these results show that family-related factors show inconsistencies for their relationship with sedentariness.

Six studies investigated other social factors such as social norms [52, 55], social cohesion, interaction, support and participation and sense of community [9, 16, 52, 62]. Social norms were found to correlate with leisure screen time in one study. Although other factors were not significantly associated with sedentary behaviour they were significant mediators of the impact of education on this unhealthy behaviour [62]. No social factors were found to correlate with weekend sitting time with or without interaction with retirement status [16] Finally a study by Uijtdewilligen et al. [27] investigating daily hassles found no significant association with screen time [71].

Of the 74 studies included, 33 considered environmental exposure. Environmental exposures/resources is categorised under five domains, four of which are previously proposed in other publications [81]; physical environment, services available in the environment, socio-demographic environment, neighbourhood safety and the additional domain of the home/work indoor environment. Twenty four studies account for factors related to the physical environment, fourteen thirteen considered variables of the socio-demographic environment (neighbourhood socio-economic status, deprivation), nine examined factors relating to neighbourhood safety and eight investigated service environment variables (recreation facilities, access to services, proximity of destinations). Finally, four studies considered the indoor environment at home or at work (furniture, number of TVs/PCs).

In term of measurement, self-reported/perceived and objective assessments of environmental characteristics were quite equally distributed across the studies. The most commonly used objective measure of environmental factors was GIS techniques and composite environmental measures (e.g., neighbourhood walkability index, neighbourhood deprivation index). Table 4 provides a detailed account of the included studies, the investigated environmental variables and measurement tools.

Mixed results were observed regarding the effect of living in a rural or urban area, dependent on the type of sedentary behaviour considered. Van Uffelen et al. [24] and Uijtdewilligen et al. [27] found that living in an urban area resulted in higher total sitting time among women compared to those living in a rural town. Likewise, Clark et al. [48] reported that living in a regional city outside of the state capitals was associated with an increased likelihood of watching two or more hours of television per day. Pomerleau et al. [28] found similar associations between urban area and sedentary behaviour during leisure time but this was dependent on nationality; Estonian men and Lithuanian women living in towns and cities were more sedentary than their rural counterparts however the opposite is true for Latvian men and women. Two studies showed that living in a rural area was positively associated with transport sitting time [15, 77].

In terms of aesthetics, only one of six studies reported a significant negative association between neighbourhood aesthetics and overall sitting time [60] while five studies reported no associations with sedentary behaviours. Considering green spaces, an increase in the density [12] and a greater proximity [14] were associated with a decrease in sedentary behaviour time. Five studies considered neighbourhood walkability showing mixed results. Three of these reported a negative association with sedentary behaviours. Sugiyama et al. [34] found that women in high-walkability neighbourhoods spend less time watching TV than their counterpart in moderate or low walkability neighbourhoods. Similarly, Kozo et al. [59] showed neighbourhoods with high walkability decrease total sitting time among both men and women. One of the five studies reported an unexpected positive association [82]: the higher the neighbourhood walkability index, the higher the time spent sitting and finally, two studies reported no associations with walkability and total screen time [8] or total sedentary time assessed objectively [59]. Only one of four studies considering walking and cycling facilities found a negative association between bike facilities and sedentary behaviour [11] and no significant associations were found with footpaths. Only one study explored the interaction between work status and neighbourhood walkability reporting a significant correlation [9]. Air or noise pollution and weather were found to be significantly associated with an increase in sedentary time. Findings relating to traffic safety were contradictory [10, 48, 60]. Lastly, no clear patterns of association were observed regarding street connectivity or land mix and sedentary behaviours.

Regarding the services/destinations resources, only Van Dyck et al. [60] found interactions with gender resulting in significant negative linear relations among men between access, proximity and number of destination close to home and sitting time; whereas six studies found no association with sedentary behaviours. Fields et al. [11] reported a negative association between public transport facilities (i.e., bus stop) and leisure time spent sedentary whereas Wallmann-Sperlich et al. [10] found no significant association with sitting time and public transport facilities.

Neighbourhood socioeconomic status was one of the most commonly investigated environmental factor (n = 7 studies) and results are contradictory. Of the seven studies, two studies reported a negative association between neighbourhood socioeconomic status and TV viewing among women [8, 34]. Proper et al. [49] reported greater sitting time for disadvantaged neighbourhoods during weekend days and less sitting time during weekdays while Kozo et al. [59] and Stamatakis et al. [79] found positive associations irrespective of the day of the week. Three studies also reported a positive association between neighbourhood deprivation and leisure screen time and total sitting time [18, 22, 79, 80]. One study reported a significant positive association between residential density and total sitting time [60] whereas results were mixed for time spent sedentary in transport [60, 83] and null for leisure screen time [11].

Two studies found associations between indoor equipment and sedentary behaviours. Van Dyck et al. [55, 82] found that after adjustment for socio-demographic factors, the size of the largest TV set and the number of computers in the home was positively associated with TV viewing time and leisure-time internet usage. Regarding the indoor work-environment, Crespo et al. [86] found a worksite promotion index including shower facilities at work, lockers for clothes at work and safe bicycle storage to be associated with greater sedentary time, but at the same time also associated with increased levels of recreational physical activity. Finally Saidj et al. found a negative association between habitat surface area and leisure time sitting cross-sectional, while the association was no more significant in a longitudinal perspective [17].