Lack of regular physical activity and prolonged sedentary time have independently been associated with numerous physical and mental health risks in all age groups (children, adolescents, adults and older adults) [1]. Nonetheless, large proportions of populations in developed countries do not engage in sufficient physical activity to gain health benefits [2]. Consequently, interventions to promote healthy and active lifestyles need to be improved and implemented. Ecological models of health behaviors emphasize the importance of taking into account multiple levels of influence when developing interventions. In addition to individual and social attributes, built environment and policy factors are expected to affect physical activity in multiple domains (e.g. leisure or transport) and settings (e.g. neighborhoods, parks) [3].

Public parks can be an important setting for physical activity promotion. Public parks are among the most common places for physical activity [4]. Parks are available in most communities, are usually free to access, can serve diverse populations including low-income and minority groups, and their provision can be influenced by local policies [5]. Thus, enhancing access to parks and optimizing their design seem to be promising strategies to increase physical activity.

To increase park use and the activity levels of park users, the crucial attributes related to active park use need to be defined. According to the conceptual model of Bedimo-Rung and colleagues [6] it is important to understand the associations of park user characteristics and structural park characteristics with visitation and physical activity within parks. Studies in the USA showed that park users were predominantly male and that adults and children were more likely to visit parks than adolescents and seniors [7–9]. Males and children were more likely to be active in parks than respectively, females and adults [7, 8]. Furthermore, lower-income families were less likely to visit parks, while the evidence on racial/ethnic minorities was inconsistent [8–10]. Concerning the associations between specific structural park characteristics and park use, studies conducted in Australia and the USA showed that park availability, park size, and the availability and quality of amenities in parks were related with physical activity in parks among children, adolescents and adults [11–13].

In addition to examining park user and structural characteristics, it is necessary to examine neighborhoods and built environments where parks are located [6, 14]. Since individuals must travel through the surrounding neighborhood before entering a park, neighborhood characteristics are likely to have an influence on whether and how a park is used [6]. Some neighborhood characteristics have been examined in previous research: low traffic safety, negative neighborhood aesthetics (e.g. abandoned housing) and low crime safety have been related to less park use [15–17]. Another potentially important neighborhood characteristic that has not been examined is neighborhood walkability (index including residential density, land use mix and street connectivity). High neighborhood walkability has been consistently related with physical activity in adults [18–20], but little is known about the availability and use of parks in high- and low-walkable neighborhoods. Possibly, parks are more prevalent, of higher quality and more easily accessible in high-walkable neighborhoods. Consequently, the overall higher amount of physical activity among high-walkable neighborhood inhabitants might be partially due to more frequent and active park use.

Neighborhood income is also an important environmental factor to take into account. Environmental characteristics like aesthetics, traffic safety infrastructure, crime safety and access to recreation facilities are less favorable in low-income neighborhoods [21, 22], creating barriers to physical activity. It is not known whether park quality or physical activity in parks differ by neighborhood income.

Almost all studies examining the correlates of park use have been conducted in the USA or Australia. Because European built environments differ from American environments, it is important to conduct comparable studies across continents to determine whether similar study methods are applicable to different regions, if park use is comparable across regions and if similar factors are related with park use.

The present study addressed some of the shortcomings in the literature. The first aim was to describe structural park characteristics and socio-demographic characteristics of park visitors, and to examine if these attributes differed between Ghent, Belgium and San Diego, California, USA, between high- and low-walkable neighborhoods and between high- and low-income neighborhoods. Data of these two cities were used because of the need to compare novel European evidence with results of regions with more evidence on correlates of park use. As a second aim, we examined if study site, neighborhood walkability and neighborhood income were associated with the overall number of park visitors and their activity levels in parks.

Overall, park users were more likely to be male, and Non-Hispanic White, than female or Black, Latino or of another ethnicity. These results are in agreement with previous findings showing that men are more likely to visit parks than women [8, 9]. The higher prevalence of Non-Hispanic Whites seems to be a reflection the overall composition of the populations in Ghent and San Diego. Previous results also showed that adults and children are more likely to visit parks than adolescents and older adults [8, 9]: this was the case in San Diego, but in Ghent, mainly adults and adolescents visited the parks. A large proportion (44.9%) of the park users was observed being sedentary. This is an important observation, pointing out that park visits do not necessarily induce or stimulate active park use.

This study revealed large differences in park and user characteristics across the two study sites. When interpreting these differences, it should be taken into account that the mean temperature during data collection was higher in San Diego than in Ghent, possibly affecting the findings. The regression analyses showed that after taking into account the covariates, park users in San Diego were more likely to be vigorously active, and the average intensity level of activities (defined as METs/observation) was higher than in Ghent. On the other hand, the total number of park visitors was higher in Ghent, as well as the number of visitors walking. The greater number of vigorously active park visitors in San Diego might be explained by more activity-supportive park features in San Diego parks than in parks in Ghent. For example, swimming pools, play areas and sports field were more prevalent in San Diego than in Ghent. Additionally, the higher activity-related intensity level of park users in San Diego could possibly be explained by the fact that USA cities usually are more car-oriented than European cities [27]. In San Diego, parks probably play a more significant role as environments to be physically active, rather than in Ghent, where many people are active on the safer streets for both leisure and transport purposes (e.g. jogging, cycling). In Ghent, parks are possibly perceived more as a place to relax or walk instead of a place to be vigorously active.

After taking into account the covariates (e.g. park size, darkness), the activity-related intensity level (METs/observation), total number of visitors, and number of visitors sedentary and walking were higher in high-walkable neighborhoods than in low-walkable neighborhoods. Previous studies showed that living in a high-walkable neighborhood was associated with more active transportation in adults [18–20] and the present findings show that neighborhood walkability also contributes to more (active) park use. Perhaps easier access to parks in high-walkable neighborhoods facilitates more visits to both sedentary and walking pursuits. Higher walkability may be related to more visitors walking to parks from their homes. Until now, no studies examined neighborhood walkability as a correlate of park use, but other environmental characteristics like high traffic safety, high crime safety and high aesthetics have been associated with higher park use [15–17]. These features are usually more favorable in high-walkable neighborhoods [28, 29] so present results confirm these previous findings.

The higher number of sedentary park users in high-walkable neighborhood parks might be explained by the fact that fewer park features were present in these parks. Furthermore, results from a previous study conducted in Ghent showed that high neighborhood walkability was not only related to more physical activity, but also (and independently) to more sedentary time [30]. Similarly, high neighborhood walkability might stimulate active (i.e. walking), as well as sedentary park use. Possibly, high-walkable neighborhood characteristics stimulate inhabitants to go outside, not only to be active but also to be sedentary in parks.

After controlling for the covariates, neighborhood income was positively related to the total number of observed visitors, but negatively to the number of vigorously active visitors. Interestingly, the average intensity level of park users did not differ between high- and low-income neighborhood parks, neither did the number of activity-supportive park features. This is encouraging, as low-income individuals are often found to be at higher risk of insufficient physical activity and overweight/obesity [31]. It seems that in Ghent and San Diego, both parks in high- and low-income neighborhoods have features (e.g. open spaces, play areas, sports fields) that can encourage physical activity, and parks are a promising environment to stimulate (vigorous) physical activity. Parks in high-income neighborhoods had more park amenities, possibly leading to increased park attractiveness and a higher overall number of park visitors. Creating parks and improving the quality (features and amenities) of available parks might be an effective approach to reach population subgroups at risk for overweight and obesity.

In accordance with the conceptual model of Bedimo-Rung and colleagues [6], not only user and park characteristics, but also characteristics of the surrounding neighborhood are important to understand park use. Nonetheless, the present study could not confirm if park users live in the neighborhoods surrounding the parks they use or not. This remains to be explored in further studies, possibly by including interviews with park users. It would also be useful to collect information on the transportation modes used to visit parks. It might be that individuals are more likely to use active transportation modes to visit parks that are located in high-walkable neighborhoods compared with parks in low-walkable neighborhoods. This can be important to fully understand the role parks play to stimulate physical activity. Moreover, future studies should examine how active park use can be encouraged in low-walkable neighborhoods, because the quality (i.e. park features and amenities) of parks in the low-walkable neighborhoods was generally high, but (active) park use was still lower than in the high-walkable neighborhoods.

The present study has several strengths. First, observational data, measured with valid and reliable observation tools, were used to assess park use and park characteristics. Second, an identical study protocol was used in Ghent and San Diego, making it possible to compare the findings across two cities in different countries. These comparisons revealed large and surprising differences. Third, the same trained observers collected SOPARC data in Ghent and San Diego. For EAPRS data collection, the observers received a standard training offered by the team who collected the data in San Diego. Applying these procedures increased the comparability of data between cities. However, study limitations also need to be acknowledged. First, no inter-rater reliability data were collected and although all observers were trained and certified during the training, some discrepancies in observations between different raters may still have been present. Second, because of the limited number of parks (n = 20), the associations between park features/amenities and park use could not be analyzed statistically in the present study. Third, because of practical limitations, SOPARC data were only collected over three days, while Cohen and colleagues [32] recommend four days of data collection. Fourth, no background information of park users was collected, making it impossible to know if park users lived in the neighborhood around the parks or not.

In conclusion, the present study showed that in addition to park and user characteristics, neighborhood characteristics are important to explain park use. The results showed that the presence of parks did not necessarily induce active park use; a large proportion of the park users were observed being sedentary. Furthermore, (active) park use was higher in San Diego and in high-walkable neighborhood parks. The overall number of park visitors was lower in low-income neighborhoods, but interestingly, visitors of low-income neighborhood parks were more vigorously active and the overall activity-related intensity did not differ between high- and low-income neighborhoods. So, it seems that the promotion of active use of existing parks or creating new parks can be a promising strategy to increase physical activity in population subgroups at risk for overweight and obesity.