•Methods: This study makes use of a unique data set of over 700 community dwelling adolescents (mean age=14, 51% male), from Santiago, Chile, who are part of a NIDA-funded study. Between 2008 and 2010, youth participants completed assessments that consisted of a 2-hour interviewer-administered questionnaire with substance use related questions including individual, familial, and neighborhood characteristics. Home addresses of study participants were mapped using Google Earth, and Google Maps. Additional data were then collected by mapping open spaces for recreational use (e.g. soccer fields, and “plazas”) using Google Earth satellite imagery from selected study neighborhoods in consultation with a faculty member from Catholic University in Chile, who was familiar with the study neighborhoods. Once data were mapped, and imported into ArcGIS, Thiessen polygons were generated to associate study participants with the nearest open space. The inverse of the size of the Thiessen polygon was used as a measure of open space accessibility. Mapping was done using respondents' actual addresses, however, only statistical summaries, and no identifying location information, are presented. Regression models, with smoking as a dependent variable were estimated in which age, sex, SES, peer usage, and accessibility were covariates. The autocorrelation tool in ArcGIS was utilized to analyze the distribution pattern of recreational space across the neighborhoods.
•Results: Age and peer use were associated with smoking. Participant gender and SES were not associated with smoking. Accessibility of open spaces and recreational facilities was associated with reductions in smoking despite the controls imposed for other possible confounding variables. Cluster analysis indicated that the distribution of open spaces was not homogenous.
•Implications: Further research is needed to supplement subtle understanding of neighborhood dynamics and characteristics of the youth populations in terms of how they utilize these open spaces. This study demonstrates: first, youth accessibility to recreational space is directly related to their smoking behavior controlling for other individual and peer factors; second, such open space for recreational activities is not evenly distributed across mid- to low income neighborhoods in Santiago, Chile. Communities with limited recreational space can consider developing such space as a neighborhood asset. Based on the study findings, it remains to be investigated the extent to which improving the physical environments at a neighborhood level may serve as a buffer to youth smoking onset.
