Workplace neighbourhood built environment and workers’ physically-active and sedentary behaviour: a systematic review of observational studies

Background Many desk-based workers can spend more than half of their working hours sitting, with low levels of physical activity. Workplace neighbourhood built environment may influence workers’ physical activities and sedentary behaviours on workdays. We reviewed and synthesised evidence from observational studies on associations of workplace neighbourhood attributes with domain-specific physical activity and sedentary behaviour and suggested research priorities for improving the quality of future relevant studies. Methods Published studies were obtained from nine databases (PubMed, Web of Science, PsycINFO, Scopus, Transport Research International Documentation, MEDLINE, Cochrane, Embase, and CINAHL) and crosschecked by Google Scholar. Observational studies with quantitative analyses estimating associations between workplace neighbourhood built environment attributes and workers’ physical activity or sedentary behaviour were included. Studies were restricted to those published in English language peer-reviewed journals from 2000 to 2019. Results A total of 55 studies and 455 instances of estimated associations were included. Most instances of potential associations of workplace neighbourhood built environment attributes with total or domain-specific (occupational, transport, and recreational) physical activity were non-significant. However, destination-related attributes (i.e., longer distances from workplace to home and access to car parking) were positively associated with transport-related sedentary behaviour (i.e., car driving). Conclusions The findings reinforce the case for urban design policies on designing mixed-use neighbourhoods where there are opportunities to live closer to workplaces and have access to a higher density of shops, services, and recreational facilities. Studies strengthening correspondence between the neighbourhood built environment attributes and behaviours are needed to identify and clarify potential relationships. Protocol registration The protocol of this systematic review was registered on the International Prospective Register of Systematic Reviews (PROSPERO) on 2 December 2019 (registration number: CRD42019137341). Supplementary Information The online version contains supplementary material available at 10.1186/s12966-020-01055-x.


Background
Many desk-based workers spend the majority of their working hours being sedentary; this is markedly greater than the proportion of time spent sedentary during nonworking hours [1,2]. An ecological model of health behaviour suggests that workers' physically-active and sedentary behaviours are influenced by multiple factors [3]. Among the multiple influences, built environment factors can be particularly influential on desk-based workers' sedentary behaviour [4]. The workplace built environment comprises the built environment attributes inside (e.g., workstations and spatial layout of buildings) and on the land parcel of the workplace building (e.g., workplaceexclusive car parking) and the neighbourhood surrounding the workplace (e.g., neighbourhood walkability and destination access). Previous reviews relating to the workplace built environment have mainly focused on examining inside environments as potential influences on physically-active and sedentary behaviours during work time [4][5][6][7]. However, workplace neighbourhood environment may also be an important influence on the physical activity and sedentary behaviours of workers. Given there is greater capacity for workers to undertake moderate-to vigorous-intensity physical activity (e.g., walking and cycling) in the neighbourhood around and beyond the immediate workplace setting [8], the influence of workplace neighbourhood environment merits examination [9].
Synthesising the broader body of empirical evidence relating to workplace neighbourhood built environment is essential for informing urban design policies to support physical activity and reduce sedentary behaviours among workers. A previous systematic review examining associations of workplace built environments, both inside and neighbourhood attributes, with physical activity and sedentary behaviour reported inconclusive findings for neighbourhood attributes [10]. However, the search strategy used did not employ specific terms relating to characteristics of the neighbourhood built environment such as neighbourhood walkability, destinations, and safety. Furthermore, the previous findings did not distinguish the purpose of active and sedentary behaviours (e.g., for occupation, transport, or recreation). Therefore, the associations may be confounded as the environmental correlates varied by domains [11,12].
Therefore, the aim of our review was to examine findings from observational studies and synthesise current evidence on associations of workplace neighbourhood built environment attributes (including those not located on the workplace precinct), with domain-specific physical activity and sedentary behaviour among desk-based workers. We further provided suggestions based on the results for improving the evidence on urban design policies to influence workers' physical activity and sedentary behaviours.

Methods
The protocol of this systematic review was published on the International Prospective Register of Systematic Reviews (PROSPERO) on 2 December 2019 (registration number: CRD42019137341).

Database search strategy
This systematic review was conducted in October 2019 following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines [13]. Systematic searches were conducted in nine databases: PubMed, Web of Science, PsycINFO, Scopus, Transport Research International Documentation, MEDLINE, Cochrane, Embase, and CINAHL. We also used Google Scholar to confirm missing studies. The last search was carried out in the beginning of January 2020. Three sets of search terms were used: environment variables (e.g., workplace, worksite, and neighbourhood), physical activity (physical activity and walking), and sedentary behaviour (sedentary behaviour and prolonged sitting). Supplementary Material 1 shows all the search terms and syntax used for the search.

Screening
The database search produced 2077 articles after removing duplicates. They were screened by two independent reviewers (CYL and YL). The inclusion criteria were: i) published after 2000 in peer-reviewed journals; ii) fulltext was written in English; iii) observational studies with quantitative analyses, and iv) estimated associations of self-reported or objective built environmental measures around the workplace with physical activity or sedentary behaviour among workers. We conducted the review on articles published after the year 2000 because studies on this topic began to emerge around that time [10]. Those studies which examined only other environmental measures, for example, the social (e.g., organisational support), informational (e.g., posters or programs), and interior (e.g., workstations) environments, which were not incorporated with neighbourhood workplace built environment were excluded. We targeted workers who mostly conduct sedentary desk-based work in a primary work location (e.g., office workers), so studies that focused on factory workers, drivers, and clinical nurses were excluded. The screening process based on title and abstract removed 1945 articles. Two independent reviewers (CYL and YL) read the full text of the remaining 132 articles to check their eligibility. This process identified 55 articles to be included in the review by the two reviewers . The consistency of the screening process between the two reviewers was over 95%. Any uncertainty of inclusion of articles was discussed with a third reviewer (MJK) until consensus was reached. Figure 1 shows the flow chart illustrating the process of database search and screening.

Data extraction
All relevant information from the full-texts of the eligible articles was extracted by one reviewer (CYL) and crosschecked by the second reviewer (YL). We extracted the following information: study location; sample; study design; built environment attributes and measurement methods (i.e., perceived or objectively-measured); outcome variables and measurement methods (i.e., reported or objectivelymeasured); covariates; and results. Associations between built environment attributes and outcome variables were reported in various ways, including Spearman or Pearson correlation coefficients, regression beta coefficients, and odds ratios.
The outcome variables were categorised into total and different domains of physical activity and sedentary behaviour, including occupational; transport; and recreational physical activity or sedentary behaviour, based on the definitions employed in each study. Adapting the categories of neighbourhood built environment attributes employed in the previous reviews [11,12,69], we divided the workplace neighbourhood built environment into five categories (Fig. 2): i) Composite environmental indices: a) a composite index including multiple neighbourhood built environments across different types (e.g., walkability calculating the density, land use mix, and connectivity); and, b) a composite index mixing neighbourhood built environment with other attributes such as interior built environment and/or workplace policies together (e.g., a scale measuring facilities around workplace, social climate, and organisational supports). ii) Route-related attributes: these included routes for pedestrians or cyclists and street connectivity or intersection density. iii) Destination-related attributes: these included the presence, density, and diversity of destinations which were assumed to increase physical activity such as shops, transport stops, recreational facilities; and the distance between workplace and home or city centre. Of note, the presence of car parking, which was specifically examined as a matter of convenience to drive cars (i.e., a type of sedentary transport behaviour), was also examined in some studies. iv) Safety: these comprised a low volume of traffic for pedestrians and bicyclists, low crime rates, and lighting along the commuting routes. v) Aesthetics: these included general aesthetics, greenness, and being free of litter.

Quality assessment
The scientific rigour of the selected articles was reviewed and assessed by two independent reviewers (CYL and YL), using the study quality assessment tool for observational cohort and cross-sectional studies [70]. The details of this assessment tool are shown in Supplementary Material 2. Each article was assessed against 14 criteria, including research aim, participants, measurements, and statistical analysis. Based on the guidance for the assessment tool [70], the research aim was assessed based on its importance and explicit description. Detailed information such as the demographic, location, and inclusion criteria provided showing a higher probability to replicate the study. Studies using a reliable and valid measurement of variables, employing multiple measurements, and considering the potential covariates typically receive higher quality scores. Each article was given a quality rating of good, fair, or poor according to the scoring guidelines. Disagreements were discussed between the two raters until consensus was reached. Most of the reviewed studies were of good (58.2%) or fair (40.0%) quality; therefore, we did not weight study findings based on their rigour. There was moderate agreement between the two independent raters on the quality assessment of the included studies; the percentage of overall agreement was 85.5%, and Cohen's Kappa coefficient was 0.69.

Synthesis of research findings
The associations of workplace built environment attributes with physical activity, or sedentary behaviour were coded into "+" (significant positive association), "−" (significant negative association), and "N" (non-significant association). If a study reported findings from several models, only the results of the most adjusted models were used. If a study reported findings from a composite score as well as its components, the results of each component were extracted. Furthermore, if a study showed results for both the overall sample and subsamples, the results from the subsamples were primarily extracted. We summarised the findings for each domain of physical activity and sedentary behaviour separately. This review considered an association to be significant if the p-value of an observed effect was < 0.05.

Study characteristics
Cross-sectional studies accounted for 85.5% of the 55 included articles ( Table 1). The period of follow-up implemented in the eight longitudinal studies ranged from 1 month [20,67,68] to 7 years [59]. Around half of the studies reviewed (n = 26) were included in the recent review by Zhu et al. [10], with the other half of the studies (n = 29) being novel to our review. Most of the studies were undertaken in the USA (n = 25, 45.5%) and the UK (n = 13, 23.6%), accounting for around 70% of all studies. All of the studies reviewed sampled working adults while some studies examined the associations in question for particular population subgroups such as women [38,63], parents [24,50], specific nationalities [19,36,51,67,68], commuters [34,46], and workers working in an university [28] and local governments [65]. Sample sizes ranged from 26 to 111,808, of which three had a sample size lower than 100 [23,53,67] and five with a sample size larger than 10, 000 [36,51,58,61,65].

Built environment correlates of physical activity and sedentary behaviour
Detailed syntheses of the findings are shown in Table 2. We reported the findings according to "instances" rather than "studies" as most of the studies reported associations of different built environment attributes with multiple domains of physical activity and sedentary behaviours. Overall, there were 455 instances in our analysis, nearly half of instances involved destination-related attributes (193 out of 455), followed by safety (111 out of 455) and route-related attributes (105 out of 455).
Additionally, most of them examined physical activity (431 out of 455), particularly within the transport domain (325 out of 431). Only 24 out of 455 examined sedentary behaviour, and all of them focused on transport settings.

Composite environmental indices
There were 14 instances where an association was estimated between a workplace neighbourhood composite index and a physical activity outcome. All measured walkability (primarily based on density, land use mix, and street connectivity), with the only exception measuring the presence of facilities and routes for walking through an audit tool [37]. Most of the instances regarding walkability were measured objectively by geographic information systems based on established indices (n = 7) and Walk Score® (n = 3) while the remainder measured perceived walkability (n = 3). Of 14 instances where composite indexes were examined, only four showed positive associations with physical activity [14,63]; all of which were in relation to the occupational domain. More than 70% of the instances reported were nonsignificant.
Fifteen instances used composite indices mixing other attributes (e.g., organised sports teams and classes within the workplace) in addition to workplace neighbourhood built environment attributes. The majority of the Composite environmental indices Physical activity instances (n = 9) found positive associations with physical activity, mostly about the recreational domain. The remainder of instances found either nonsignificant [28,46,61] or negative associations [46] in relation to physical activity. There were no instances of associations between composite indices and sedentary behaviour.

Route-related attributes
Most of the instances examined routes for pedestrians or cyclists and street connectivity, as route-related attributes to investigate their associations with physical activity, especially in the transport settings, which accounted for 81% of the instances. Around 16% of the 102 instances reported positive associations with physical activity [18,21,26,35,44,49,62]. The majority of instances (n = 83) were non-significant. However, three instances were negatively associated with physical activity; all of which were in relation to the transport domain [27,59].
All three instances of examining an association between route-related attributes and sedentary behaviour found that routes for pedestrians or cyclists were not associated with sedentary transport behaviour [29,43].

Destination-related attributes
The majority of the instances used the presence, density, and diversity of destinations as destination-related attributes in the workplace neighbourhood to examine their associations with physical activity, especially during transport-related contexts, which accounted for 77% of the instances. Almost 40% of the 178 instances which examined destination-related attributes were found to be positively associated with physical activity, mostly in relation to the transport domain. Over 60% of the instances were non-significant. Additionally, four instances were found to be negatively associated with transport physical activity [22,52,57,65]; of which one negative association examined car parking with transport physical activity [22]. The features of these destination-related attributes identified in the positive and negative associations were different. The presence or density of shops, transport stops, and recreational facilities were more identified in the positive associations; by contrast, all of the negative associations identified that longer distance between workplace and home and car parking around workplace were associated with lower levels of transport physical activity.
The majority of instances (11 out of 15) reported negative associations between destinations-related  attributes and sedentary behaviour; all of which were regarding the transport domain [24,29,50,51,56,66]. Most of these destination-related attributes examined were the distance between workplace and home. Furthermore, the only two instances examining car parking showed positive associations with sedentary transport behaviour [24,56]. The remainder of instances showed non-significant associations [24,43]. The diversity of destination-related attributes examined in sedentary behaviour was less than in physical activity; a higher proportion of instances examined the distance between workplace and home with sedentary behaviour.

Safety
The instances in relation to workplace neighbourhood safety mainly measured the traffic (e.g., low volume of traffic for pedestrians and bicyclists) and crime safety (e.g., low crime rates). Less than 10% of the 105 instances of estimates between safety and physical activity showed positive associations [21,44,50,59,62]. Over 85% of the instances were not significant. Also, there were four instances where safety was negatively associated with transport-related physical activity [26,43,62].
Most of the instances (4 out of 6) regarding associations of safety with sedentary behaviour showed nonsignificant associations. The remainder of the instances found that perceiving it to be safer to cross the road and cycle was associated with a lower likelihood of car-only trips [43]. All of the instances were examined in the transport settings.

Aesthetics
All but 3 of the 17 instances examining associations between workplace neighbourhood aesthetics and physical activity were not significant. There was one instance of a positive association between perceptions that streets were free from litter and transport physical activity [21] and two instances of negative associations between objectively measured greenness and occupational physical activity [63].
None of the instances reported indicators of aesthetics with any domains of sedentary behaviour.

Discussion
This review examined studies of the associations of workplace neighbourhood built environment attributes with active and sedentary behaviours among adults working in occupations involving prolonged indoor sitting. We found most of the instances examining associations between the workplace neighbourhood built environment and physical activity to be non-significant, irrespective of the study quality. Our findings are consistent with a previous review on workplace neighbourhood built environments [10]. This previous review also showed mostly non-significant associations of workplace neighbourhood built environment attributes with physical activity [10]. However, another systematic review on the residential neighbourhood reported consistently positive associations between accessibility to destinations and transport-related walking [12]. These findings suggest that compared with the residential environment, the workplace environment seems to be a less important context for supporting physically-active behaviours. A potential explanation for the weaker associations of workplace neighbourhood built environment attributes with physical activity in comparison with residential neighbourhoods may be due to workers having less autonomy over behavioural decisions during working hours, especially for those who perceive their managers disapprove of absenteeism from desks for walking [71]. Additionally, a lack of time for exercising during typical working hours may be another explanation for the weaker associations [15].
We found that destination-related attributes, notably longer distances between workplace and home, as well as better access to car parking around the workplace, were positively associated with transport-related sedentary behaviour; the sedentary transport behaviour in all the reviewed studies was travelling by car. A previous review on residential neighbourhood built environments found that better access to recreational facilities and public open spaces were negatively associated with transport-related sedentary behaviour [11]. Although varied destination-related attributes were measured in these two reviews, these findings suggest that better access to destinations, except for car parking, may be disincentives for transport-related sedentary behaviour, in both workplace and residential neighbourhoods. A short travelling distance to the workplace may make it more likely that workers undertake active commuting [72] which can replace time spent in car travel. For this reason, it may not matter whether there are wellmaintained pathways or safe routes in or around the workplace neighbourhood if workers live too far from their workplaces to commute actively. Urban design policies on co-locating residential and workplaces together and decreasing car parking spaces around the workplace or moving car parking further away from workplaces may be effective strategies to reduce workers' time spent in car travel.
The difference in the number of included studies between the previous review [10] and our review was mainly explained by the novel studies published after 2018 and additional studies examining specific neighbourhood built environment attributes. Our findings may suggest an authentic lack of an association between workplace built environments and physical activity, or the null findings may be attributable to the misperceptions of neighbourhood environments. A previous study reported that those who were less physically active for transport purposes perceived their high walkable residential neighbourhood as low walkable [73]. Such a misperception of neighbourhood environments may attenuate the associations of perceived workplace neighbourhood built environment attributes with active behaviours because most of the reviewed studies measured perceived environmental attributes. Based on the review findings, there are some research priorities suggested for improving the quality of future relevant studies, as follows.

Conducting research on workplace neighbourhood built environment and sedentary behaviour
Most of the previous studies examined the relationships of the workplace neighbourhood built environment with workers' physical activity, rather than their sedentary behaviour. Considering the increased proportion of workers in desk-based occupations [74] and the high proportion of sitting time occurring in the workplace [1,2], more attention is needed in investigating sedentary behaviour and workplace neighbourhood built environment correlates. A recent review showed that previous studies on workplace environments and sedentary behaviour mainly focused on the workplace interior environment, such as workstations [10]; however, our findings suggest that some destination-related attributes surrounding workplaces were associated with sitting time among workers. For developing effective approaches to improving workers' health through reducing sitting time, studies on workplace neighbourhood built environments and sedentary behaviour, particularly the domains most likely to be affected (e.g., occupational and transport-related sedentary behaviour), are needed.

Improving measurement and diversity of workplace neighbourhood built environment
Most previous studies assessed the workplace neighbourhood built environment using perceived measures; however, perceptions of the attributes of neighbourhoods could vary markedly between individuals, regardless of the objective environmental attributes that exist in the workplace neighbourhood. Furthermore, all the reviewed studies identified the shortest commuting route between workplace and home [27,32,33] irrespective of workers' transport modes. The shortest commuting route may not represent the routes taken by the individuals. Future research included both perceived, and objective measures of the workplace neighbourhood built environment and monitoring the actual commuting routes (e.g., by global positioning system) for workers is encouraged to clarify their associations with active and sedentary behaviours. Additionally, some of the reviewed studies measured the variety of public open space in the workplace neighbourhood without considering the quality of such attributes. Previous research has shown that the quality of built environment attributes, including destinations such as parks, may be an essential factor to influence individuals' active behaviours [75,76] and thus should be considered. Future research on diverse varieties of public open space around the workplace and multiple measures for accessibility (e.g., the number of public transport stops as well as its frequency) may provide insights that will be relevant to developing effective strategies to promoting workers' physical activity and reducing sedentary behaviour.
Developing a framework for defining the influential buffers of workplace neighbourhood built environments for active and sedentary behaviours Many of the reviewed studies did not clearly define the locations or areas of environmental attributes. Some studies used ambiguous terms (e.g., at or around the workplace) without specifying the neighbourhood's boundary [22,36,77]. The core rationale for distinguishing interior and neighbourhood contexts of workplaces is its implications for where responsibility for improvements liesi.e., land/property owners or governments. Furthermore, re-examining the influential buffers of the workplace is essential as it may be smaller than the frequently used buffers (e.g., 400-and 800-m) in research around residences [69] due to limited free time [15] and less autonomy over their behavioural decisions [71] for workers during working hours. Some other studies combined interior facilities and workplace policies [36,37,46] while assessing environmental attributes. These additional variables may contribute to stronger associations. Future studies are recommended to develop a framework in specifying the size of the workplace neighbourhood with clear definitions when examining their associations with physical activity or sedentary behaviours.

Enhancing the correspondence between where built environments and behaviours are assessed
Most of the previous studies did not precisely designate 'where' the active or sedentary behaviours occurred, whereas they assessed built environment attributes surrounding the workplace only. The disparity between these variables may lead to a misinterpretation of workplace neighbourhood built environments due to the contribution of active or sedentary behaviours in non-work contexts to total physical activity. For example, workers who engaged in more light-intensity physical activity during working hours do compensate by doing less active during non-working hours [78]. Therefore, distinguishing the venue and different intensities for these behaviours in different contexts could inform detailed information to examine whether the compensation of physical activity or sedentary behaviour occurred. For ascertaining the associations of the workplace neighbourhood on physical activity or sedentary behaviour that occurs during work hours, or during commuting, studies which could identify specific venues (e.g., global positioning system) and timing-specific behaviours (e.g., accelerometer) are suggested.

Considering potential confounding factors
When considering the workplace environmental correlates of physical activity or sedentary behaviour, some potential covariates should be examined in future research. For example, some geographic attributes attached to locations play an important role when individuals choose where to reside and work [79]. The self-selection of the residence and workplace may moderate the association of environmental attributes with active and sedentary behaviours. Referring to the ecological model, there may be an accumulative effect across different levels of factors. Individual motivations and attitudes, lifestyle preferences, social supports, interior workplace facilities, and workplace health promotion programs, may all contribute to the associations of workplace built environment attributes with workers' physical activity and sedentary behaviour to some extent [45]. However, few of the reviewed studies accounted for factors such as the preference of the workplace while examining the associations in question. Studies considering these additional factors will provide additional evidence for the independent associations of environmental correlates with active or sedentary behaviours.

Implementing research in diverse settings with prospective designs
Previous studies on workplace environmental correlates of active or sedentary behaviours have been mainly conducted in Western countries such as the USA and the UK. More relevant studies from non-Western countries should be encouraged because different countries or areas have varied behaviour patterns and neighbourhood built environments. For example, there are marked differences in the prevalence of active commuting across regions [80]. The low prevalence of physically-active commuting in Western countries leads to most of these studies investigating how to promote active commuting to or from work (i.e., transport physical activity) but less into other domains. However, studies giving weight to other domains of physical activity and sedentary behaviour may have more contributions to increase the total amount of physical activity and reduce the total time of sedentary behaviour, especially in countries or areas with a relatively high prevalence of active commuting to work. Additionally, future research should use prospective or experimental designs to evaluate whether changes in workplace neighbourhood built environments affect active and sedentary behaviours, rather than crosssectional designs, which form the majority of the existing evidence base.

Conclusions
Desk-based workers can spend around 80% of their working hours sedentary and can have limited opportunities for physical activity in and around the workplace. Synthesizing the current research evidence, we found that workers who lived further from their work and who could easily access car parking surrounding the workplace had a higher likelihood of transport-related sedentary behaviour. However, we found that workplace neighbourhood built environments such as route-related attributes, safety, and aesthetics did not appear to be influential for workers' physically-active and sedentary behaviours. Designing mixed-use neighbourhoods where there are opportunities to live close to workplaces as well as have access to a high density of shops, services, and recreational facilities may be useful for reducing workers' sedentary time. Future investigations with improvements in research design and measurements are needed to more deeply understand the impacts of workplace neighbourhood environments on workers' physically-active and sedentary behaviours.