Synthesis of findings
Most of the studies we reviewed either explicitly or implicitly adopted a social-ecological framework, recognizing the significance of interactions between intrapersonal, social, physical and cultural environments for perceptions of physical and social urban environments and, ultimately, physical activity behaviors [56, 57]. By highlighting whether the influences of dogs on physical and social environments were positive or negative, who was impacted by these influences, and what characterized the physical and social contexts in which this was taking place, our attention was directed to assessing the overall impact dogs had on the sharing of public spaces. In particular, we considered processes that could render dogs into influencers of physical activity among dog-owners and non-owners of different ages and in different life stages.
The direction of influence was not uniformly or consistently positive. Women [4, 34–39], ethnic minorities [34–40], and older adults [29, 31, 39] seemed to be most susceptible to experiencing other people's dogs as barriers to being physically active, although there were exceptions [45, 50]. In contrast, studies focusing upon middle-class, predominantly Caucasian adult populations positioned other peoples' dogs as either having no impact on the physical activity patterns of women  or older adults , or as facilitators of physical activity through the positive contributions made to the social environment [5–7, 9, 33, 41, 54].
Notably, in King et al.'s intervention research on physical activity, loose and unattended dogs were identified as a barrier in ethnic-minority neighborhoods in Atlanta, Georgia and Memphis, Tennessee, while participants from Stanford, California; Eugene, Oregon; and Kingston, Rhode Island - primarily White populations - did not identify dogs as barriers . This difference was not explained by household income levels, as the Oregon and Rhode Island samples had lower average household incomes than did the Georgia and Tennessee samples. Sallis and colleagues noted similar findings, stating the lowered levels of physical activity reported by women in neighborhoods where loose and unattended dogs were more likely to be reported were not explained by variations in socioeconomic status . This suggests that social disadvantages associated with gender, age, and race/ethnicity may be more pertinent than income when predicting the direction of influence dogs might have on surrounding social and physical environments, which has implications for future interventions and evaluation of interventions.
Clearly, urban-dwelling dogs have the potential to impact the health of human populations in both positive and negative directions. The negative aspects of negotiating the use of shared spaces with dogs are modifiable, however, and could be leveraged in ways that can overcome not just the direct, dog-related outcomes, but other environmental factors that were often reported alongside nuisance from and fear of dogs. For example, many African American women who participated in Griffin et al.'s study  listed fear of uncontrolled dogs among the factors inhibiting them from being more active, but also proposed that increasing a sense of connectedness among neighbors was a potential solution to increasing physical activity. A campaign targeting responsible dog-ownership could potentially minimize the role dogs play as barriers and reposition them as facilitators of population-level physical activity. Similarly, in Sanderson and colleagues' study , African American adults identified seeing no one else out being active in the neighborhood as a barrier to being active themselves. These participants also suggested that living in 'a good neighborhood,' where neighbors looked out for one another, would facilitate exercising outdoors. An increased presence of responsible dog-owners - who walked their dogs regularly, kept them on-leash and under control, and picked up dog litter - in these neighborhoods could address both of these conditions, as was shown in studies by Wood and colleagues [5, 9].
Lacking the motivation to be physically active was also identified by some study participants as contributing to physical inactivity [35, 43]. And yet, dog-owners were able to overcome their own personal barriers, including minor illness and depression as well as inclement weather, to walk their dogs on a daily basis . While we acknowledge that reasons for choosing not to own a dog may range from financial limitations, housing situations, and time-constraints to personal preferences, some of these could be addressed by physical activity interventions that leverage, to the benefit of non-dog-owners, the motivation dogs seem to provide to walk regularly. Peel and colleague's account of a type 2 diabetic who was able to access her neighbor's dog-walking regime to incorporate regular physical activity into her own life supports the supposition that ownership is not a requirement for someone to benefit directly from a dog's need to be walked. Many of the studies in this review offered evidence that dog-owners are more likely than non-owners to attain and to maintain recommended levels of physical activity, yet interventions could help increase the likelihood of non-owners achieving this important public health goal. In this regard, the results of Johnson & Meadow's 'loaner dogs' intervention among public housing tower residents are encouraging.
Strengths and limitations of this review
Realist reviews have, to date, been used to evaluate the effectiveness of planned interventions [60, 61] and, as far as we know, ours is the first to apply this method to a scoping review. Given the disparate literatures involved, as well as the observation that dog-relevant evidence is at times included in results, but not highlighted or discussed by the authors, it is possible that we did not identify all studies that met our inclusion criteria. Nevertheless, we did not exclude any study based on methodological grounds, as our aim was to be as inclusive as possible. This bias toward inclusion was helpful, given that the studies appeared in journals with divergent audiences and disciplinary orientations. Still, the total number of studies meeting our inclusion criteria remained manageable. It is also to be expected that previous experience in research and practice influenced our inclusion decisions and interpretations of the data, and so other researchers could adopt our search strategy and yet end up with different inclusion decisions and analytic interpretations .
Implications for research and practice
Our analysis supports the need for a dedicated literature exploring the influence of dogs on population health via physical and social urban environments. Dogs, as our analysis highlights, do not have uniform effects and we need to understand more about these differences, so as to maximize benefits and minimize inconveniences and harms. In conducting this review, we wish to invite further intervention-oriented research on dogs as potential contributors to population health via physical activity.
When planning population health interventions that depend upon the social and physical qualities of neighborhood environments to achieve success, we strongly suggest looking beyond human populations. Responsible dog-ownership practices, as well as provision of dog-supportive amenities, are the key to enabling dogs to reach their potential as promoters of population health through their positive impacts on physical and social environments. Bjerke & Ostdahl's identification of the prevalence of problems with neighbors dogs, using a random sample rather than targeting disadvantaged communities , underscores the importance of designing, implementing and enforcing policy to reduce the negative and increase the positive benefits of co-existing with dogs. The findings of both Christian (neé Cutt) et al.  and Lee et al.  emphasized the importance of accessible, dog-supportive parks as arenas for becoming aware of and putting into practice responsible dog-ownership behaviors, in addition to encouraging walking (for both dogs and their owners) and facilitating social encounters with neighbors and other park-users. The presence of wildlife can encourage people to get outdoors and walk, which adds a layer of complexity in managing dog populations in urban environments and in large park settings .
It will be crucial, however, to consider strategies for ensuring that disadvantaged communities are not excluded from these efforts. These neighborhoods will likely require the greatest investment of time and other resources. Animal restraint ordinances, the most commonly suggested means of eliminating stray or uncontrolled dogs as barriers to physical activity, may be of benefit but should be stringently evaluated. Options that could be implemented and evaluated include: spay/neuter programs, subsidized veterinary services, pet food donations, responsible dog-ownership campaigns and training classes, dog-sharing and dog-fostering programs, amenities that support dog-walking, and enforcement of ordinances regarding dog litter and dog control.