
Are Neighbourhood Social Capital and availability of sports facilities related to sports participation among Dutch adolescents?
Prins R.G., Mohnen S.M., Lenthe F.J. van, Brug J., Oenema A.
Submitted
Abstract
Background: The aim of this study is to explore whether availability of sports facilities, parks, and neighbourhood social capital (NSC) and their interaction are associated with leisure time sports participation among Dutch adolescents.
Methods: Cross-sectional analysis of complete data from the last wave of the YouRAction evaluation trial. Adolescents (n=852) completed a questionnaire asking for sports participation, perceived NSC and demographics. Ecometric methods were used to aggregate perceived NSC to zip code level. Availability of sports facilities and parks was assessed by means of geographic information systems within the zip-code area and within a 1600 meter buffer. Multilevel logistic regression analyses, with neighborhood and individual as levels, were conducted to examine associations between physical and social environmental factors and leisure time sports participation. Simple slopes analysis was conducted to decompose interaction effects.
Results: NSC was significantly associated with sports participation (OR: 3.51 (95%CI: 1.18;10.41)) after adjustment for potential confounders. Availability of sports facilities and availability of parks were not associated with sports participation. A significant interaction between NSC and density of parks within the neighbourhood area (OR: 1.22 (90%CI: 1.01;1.34)) was found. Decomposition of the interaction term showed that adolescents were most likely to engage in leisure time sports when both availability of parks and NSC were highest.
Conclusions: The results of this study indicate that leisure time sports participation is associated with levels of NSC; but not with availability of parks or sports facilities. In addition NSC and availability of parks in the zipcode area interacted in such a way that leisure time sports participation is most likely among adolescents living in zipcode areas with higher levels of NSC, and higher availability of parks. Hence, availability of parks appears only to be important for leisure time sports participation when NSC is high.
Introduction
Sports participation among adolescents is a public health priority 1-2 and increases the likelihood of being physically active in adulthood 3. Despite the fact that levels of sports participation are relatively high among Western adolescents, a steep decrease during adolescence has repeatedly been reported 4-8. For example, in the Netherlands it was found that 74% of the adolescents engaged in sports in the first year of secondary education, but this dropped to 48% two years later 4. In order to promote sport participation, deeper understanding of the factors that are associated with sports participation among adolescents is needed. In this respect, environmental factors, such as the availability of sports facilities or living in a supportive social environment, are of particular interest, as environmental factors may have an influence on the behaviour of large groups of people. Recently, there has been a call for studies to simultaneously study the social and the physical environment, in order to find their independent relation and potential synergy in facilitating physical activity (PA) 9-10. That is what the present study aims for.
For this purpose, the concept of behaviour setting, defined as “those social and physical situations in which behaviours take place, by promoting and sometimes demanding certain actions and by discouraging or prohibiting others” 11, is important. This implies that both social and physical environmental factors may influence PA, as also conceptualized in socio-ecological models 12-13. For example, the EnRG framework postulates that both, the physical and social environment may influence PA behaviour 12. In addition, a model suggested by Franzini et al. elaborates more on the interplay between social and physical environmental factors and suggests that the social environment can moderate the relation between factors in the physical environment and outdoor PA 13. In this view it is very well possible that, when facilities to be active are available, they will be used more often if the social environment is supportive instead of unsupportive.
Thus far, most empirical studies have focused on physical environmental factors, but studies have not yet been able to nail down which physical environmental factors are directly associated with PA. Some studies have found positive associations between the objectively assessed availability of facilities to be active and PA behaviour 14-16, whilst others did not find this association 4; 17-18. It may be that the physical environment facilitates behaviour to a certain extent, but is not sufficient to motivate or enable most people sufficiently to actually engage in PA 19. Various scholars have proposed to include other environmental factors and study their interaction in relation to PA 20-21.
Social capital is a social environmental factor that may have an influence on sports participation. The origin of social capital states that “social networks have value” 22 and that social capital is a resource that “inheres in the structure of relations between actors and among actors.” 23. In the present study, social capital is defined as the resources (e.g. norms, trust) that are available to all members of a community (in our case a neighbourhood) 24. Crucial to social capital on the community level is having common norms, behavioural reciprocity and mutual trust 22. Neighbourhood social capital (NSC) is thought to affect health related outcomes via maintenance of healthy norms and access to social support 25-26. Moreover, neighbours live close to each other, and therefore, it is likely that neighbours observe and learn from each other’s behaviour 27-28, especially if the individuals involved are strongly socially connected. Indeed, various studies confirmed that higher levels of perceived social capital (i.e. individual level) were associated with higher levels of PA among children 29 and adolescents 9-10, even after being adjusted for physical environmental factors 9; 30.
Despite the fact that social and physical environmental factors may determine sports participation in adolescents, they have thus far hardly been studied simultaneously. In the interaction between NSC and the physical environment, two mechanisms seem plausible. Firstly, it has been proposed that a supportive social environment may help to overcome an unsupportive physical environment 13. Hence, individuals may decide to be physically active even when some physical environmental variables are not favourable. Secondly, social capital may “add value” to the physical environment. For instance, when more people are physically active in their neighbourhood, parks and sport facilities may be more attractive places to go to. As formal amenities such as sports facilities may require memberships and informal amenities such as park do not require this, it is likely that the proposed interactions are stronger for regarding informal amenities. There are only few studies that have examined these interaction mechanisms; Broyles et al. 31 have found that parks with higher levels of social capital are visited more often and the total volume of energy expenditure was also significantly higher in those parks 31. Seaman and co-workers concluded that in order to promote access to green space in urban communities there is an interaction between physical availability of green space and urban community contexts 32.
The present study aims to explore the direct and adjusted associations of NSC (social environment), availability of sports facilities and parks (physical environment) with leisure time (LT) sports participation. In addition moderation of NSC on the physical environment- LT sports participation association will be explored.
We hypothesize that 1) NSC is positively and significantly associated with LT sports participation, 2) the availability of sports facilities and parks are positively and significantly associated with LT sports participation, 3) NSC and availability of sports facilities and parks interact in such a way that NSC is stronger related with LT sports participation when availability of parks and sports facilities are higher and that this effect will be more pronounced in freely available facilities such as parks.
Methods
Study design
The data used in this study are derived from the final post measurement of the YouRAction study (2009-2010, Rotterdam and surroundings, the Netherlands), because information on NSC was only collected in this measurement. YouRAction was a three-armed cluster randomized trial in which two versions of a computer-tailored PA promotion intervention were evaluated against a generic information control group. In the first arm, adolescents (12-13 years) received generic information about PA and diet. In the second and third arm of the trial adolescents received a computer-tailored advice to promote their PA levels. The interventions are extensively described elsewhere 33. In the trial, measurements were conducted at baseline, one month and six months post intervention (last measurement). School classes were randomly assigned to one of the study arms using block randomization. The evaluation study showed that the interventions were not effective in promoting moderate-to-vigorous PA among adolescents [see: Chapter 8] and additional analyses showed that there was no effect on changes in sports participation between baseline and final post intervention measurement.
The Medical Ethics committee of the Erasmus Medical Center issued a “declaration of no objection” for the YouRAction study.
Sampling and procedure
Schools were contacted to inform whether they were willing to participate in the trial. Classes which received general secondary education were eligible for inclusion in the study. Adolescents received information and passive consent forms for themselves and their parents (i.e. if adolescents and/or parents rejected to participate, they were removed from trial).
Of the 1240 adolescents invited for YouRAction, in total 27 (2.2%) adolescents or their parents declined study participation. In the final follow-up measurement, in total 1129 adolescents participated. Self-administered data was collected in classrooms, in the presence of a research assistant and a teacher.
Adolescents with complete data on the variables of interest and living in neighbourhoods in which at least 5 respondents lived were eligible for analyses. In total 852 adolescents met these criteria. Higher educated adolescents were more likely to be in the final sample.
Measures
LT sports participation
Sport participation was assessed using the sports participation questions from an adapted version of the Activity QUestionnaire for Adolescents & Adults (AQUAA) 34. The AQUAA showed moderate test-retest reproducibility, with an intra-class correlation of 0.59 for vigorous activities 34.
Adolescents could write down a maximum of 3 sports in which they had participated during the previous week, and indicate on how many days of the week (0-7 days) they had participated in each sport. Moreover, they could indicate the context in which this took place: school, neighbourhood, sports club and at home.
If sports only took place at school, the sports frequency for that particular sport was set to 0 (i.e. not participating in sports), because this does not add to LT sports participation. A dichotomized variable was created to indicate whether an adolescent participated in LT sports at least once a week (1) or not (0).
Neighbourhood Social Capital
Adolescents filled in two items in the questionnaire on social capital: 1) “the people in my neighbourhood get along with each other well”, and 2) “I live in a close-knit neighbourhood with a lot of solidarity”. Response categories ranged from ‘totally disagree’ (1) to ‘totally agree’ (5) (Cronbach’s Alpha: 0.87). A measure for NSC was constructed by aggregating the individual responses from the questionnaire to the neighbourhood (defined as 4 digit zip code) level. On average 15.2 adolescents per neighbourhood answered the two NSC questions.
An ecometrics approach was used for creating the aggregated measure of social capital (for extensive information see 35-37). In this approach, the two items measuring social capital were the dependent variables (i.e. a long dataset was created and a dummy variable indicates the item number). A linear three-level multilevel model (neighbourhoods, individuals, items), that accounts for the nesting of social capital items within individuals and neighbourhoods, was used. The model was adjusted for six individual characteristics that may influence the perception of NSC: gender, ethnicity, age, education, type of housing the adolescent lives in and years living in the current home. The residuals from this analysis (i.e. the part that cannot be attributed to individual response patterns) constitutes the NSC variable. Positive values indicate higher than average levels of NSC. The reliability of the ecometric scales depends on the variance at the three levels, i.e., items nested within respondents, and respondents nested within neighbourhoods 38. In our sample we found a reliability of the NSC variable (based on Hox, 2002 38) which was acceptable at 0.57 (see appendix A for more details on calculation). The construction of the NSC variable was done in MLwiN 2.02.
Availability of sports facilities and parks
Geographic information system (GIS) data on the availability of sports facilities and parks were retrieved from municipal databases. Addresses of adolescents’ homes were geocoded by using the centroid of the six-digit zip codes of their home address.
Two measures of availability of sports facilities and parks were constructed. Directly matching with the scale used for NSC, a variable indicating the density (i.e. the number of facilities per square kilometer) of sports facilities and parks within each neighbourhood (i.e. 4 digit zip code area) was constructed. In addition, using open source Quantum GIS software, crow-fly buffers were used to count the number of sports facilities and parks within 1600 meters from the participants’ home addresses. The chosen spatial scale is in line with previous research 4 and based on a study by Colabianchi et al. 39. This latter method of measuring the physical environment is more common in the study of physical environmental influences on PA than measuring the availability of facilities within the neighbourhood.
Covariates
Ethnicity was based on questionnaire information on country of birth of the adolescent and their parents, according to the standard definition of Statistics Netherlands 40. An adolescent was considered to be of Western descent if he or she and both parents were born in the Netherlands, another European country, Oceania, North America, Indonesia or Japan. If the adolescent or one of the parents was born in another country, he/she was considered to be of non-Western descent.
Adolescents could indicate which level of education they attended in the questionnaire, which was categorized into higher level education (preparatory education for university) and lower level education (vocational education).
Adjustment for neighbourhood wealth is useful to mitigate area-level confounding 41. Therefore neighbourhood wealth was considered to be a potential area-level confounder. Neighbourhood wealth was retrieved from the WoON ’09 database which is managed by the Dutch ministry of the Interior and Kingdom Relations. In this database 40.000 households are sampled and incomes per 4-digit zip code were averaged to create an overall variable of neighbourhood wealth. In addition, urbanity was considered to be an area-level confounder. Information about urbanity, measured on a zip code level was retrieved from Statistics Netherlands. The urbanity index ranges from 1 to 5, with 1 being the most rural areas and 5 being the most urban areas.
Analyses
Descriptive statistics were used to describe the study population. Univariate two-level random intercept multilevel logistic regression (with neighbourhood and individual as levels) analyses were conducted to assess associations of demographics, NSC, availability and density of parks and sports facilities, with LT sports participation.
In order to study the association of NSC and availability/density of parks and sports facilities with LT sports participation, multiple multilevel logistic regression analyses were conducted with neighbourhood and individual as levels. In an empty model (i.e. model 0) we found that LT sports participation clusters within neighbourhoods (Median Odds Ratio: 1.69, 95%CI: 1.42;2.2