Leisure-time walking and cycling may be appropriate types of physical activity (PA) to achieve current PA recommendations, since they are easy to implement in daily life with low cost and little risk of injury. The most effective ways to encourage these types of PA in adults are yet uncertain. Traditionally, focus has been on individual-level determinants of PA. Recently, a complementary ecological approach has been taken on which postulates that PA is also influenced by the individuals’ living environment .
To create neighbourhood environments that encourage leisure-time walking and cycling, it is important to understand which neighbourhood elements are most strongly related to these types of PA. According to a recent review, no firm conclusions can be drawn on environmental determinants of cycling due to lack of studies on this topic . Leisure-time walking has been studied more extensively. Pedestrian infrastructure and neighbourhood aesthetics have consistently been found to be associated with leisure-time walking . There is less consistent empirical support for other neighbourhood elements.
One of these elements is neighbourhood safety. Some studies have found a positive association between general neighbourhood safety and walking in leisure time [4, 5], but most studies did not find an association [6–13]. Many of the studies on neighbourhood safety have used a composite measure, in which various safety components (e.g. traffic, crime, disorder) are combined into one comprehensive safety variable [4, 6, 7, 9–12]. The use of such combined safety indicators might obscure the effects of the specific safety components.
Studies on specific safety components have most often explored the role of traffic safety. Inoue et al.  found that people who perceived good traffic safety were 1.5 times more likely to walk in leisure-time than those who perceived poor traffic safety. However, much other research has found inconsistent [5, 15, 16] or no associations [6, 7, 17–20] between traffic safety and walking. With regard to crime safety, again, results are inconsistent. Shigematsu et al.  found respondents’ perceptions to be associated with leisure-time walking, but only in some age groups. Other studies have found no association at all [6, 14, 17, 20, 21]. Only few studies have explored the role of neighbourhood disorder. Cleland et al.  found limited evidence of a positive association with leisure-time walking. Absence of physical neighbourhood disorder was positively associated with maintaining high levels of leisure-time walking between baseline and follow-up, but not with three other leisure-time walking outcomes. Other studies did not find any significant association of leisure-time walking with indicators of neighbourhood disorder such as garbage, graffiti or public drunkenness [6, 7].
Comparability of these results is limited, due to differences in settings, PA measures, and safety measures. Therefore, it is hard to determine the relative impact of each specific safety component on leisure-time walking behaviour. Studies are needed that simultaneously explore the association of various safety components with leisure-time walking. To our knowledge, there is only one American study that has explored multiple safety components . The authors found no association between leisure-time walking and objective measures of criminal offenses, traffic-related offenses, physical disorder and social disorder.
As the latter study , most studies on safety and leisure-time walking have been performed in America and Australia [2, 22]. European studies on this topic are rare. Yet, results may be different in Europe because PA patterns as well as the safety situation of neighbourhoods may differ [23, 24]. For example, the association of leisure-time PA with crime and traffic safety may be less strong, because of safer traffic and lower crime rates in deprived neighbourhoods in Europe compared to America . Within Europe, the Netherlands may be a particularly interesting country to explore the associations between neighbourhood safety and leisure-time PA. Due to high prevalence rates of walking and cycling , it offers the opportunity to explore the association of neighbourhood safety with both walking and the much less studied PA component cycling. The current evidence on the environmental correlates of cycling mainly comes from the transportation literature and has focused on cycling for transport rather than cycling in leisure time [17, 20, 26, 27]. The one study on leisure-time cycling found no significant association with general safety .
The aim of the current study is to explore the association of neighbourhood-level safety with leisure-time walking and cycling in a large sample of Dutch adults. First, the association of general safety with leisure-time walking and cycling will be explored. Next, we will explore whether these associations are different for specific safety components. Following McGinn et al. , the safety components will include physical disorder, social disorder, crime-related fear and traffic safety. In a last step, we will explore whether associations differ by subpopulation. A review of Foster et al.  has postulated that associations may differ according to age, gender, education, and residential density of the neighbourhood. Women, elderly and lower educated tend to feel more vulnerable which may manifest itself in a stronger association of safety with leisure-time walking and cycling. Further, we postulate that in the Netherlands, densely populated areas may provide increased natural surveillance from both houses and pedestrians, which may make people feel less vulnerable to unsafe situations. If so, this would result in a weaker effect of safety on leisure-time walking and cycling in densely populated areas. Little research has focused on these differences.