Tobacco retail environment and smoking: a systematic review of geographic exposure measures and implications for future studies

Introduction
To review the geographical exposure measures used to characterize the tobacco environment in terms of density of and proximity to tobacco outlets, and its association with smoking-related outcomes.

Methods
We used PubMed and Google Scholar to find articles published until December 2019. The search was restricted to studies which 1) measured the density of and/or proximity to tobacco outlets and 2) included associations with smoking outcomes. The extraction was coordinated by several observers. We gathered data on the place of exposure, methodological approaches, and smoking outcomes.

Results
Forty articles were eligible out of 3,002 screened papers. Different density and proximity measures were described. 47.4% density calculations were based on simple counts (number of outlets within an area). Kernel Density Estimations and other measures weighted by the size of the area (outlets/sq km), population, and road length were identified. 81.3% of the articles which assessed proximity to tobacco outlets used length distances estimated through the street network. Higher density values were mostly associated with higher smoking prevalence (76.2%), greater tobacco use and smoking initiation (64.3%); and lower cessation outcomes (84.6%). Proximity measures were not associated with any smoking outcome except with cessation (62.5%).

Conclusion
Associations between the density of tobacco outlets and smoking outcomes were found regardless of the exposure measure applied. Further research is warranted to better understand how proximity to tobacco outlets may influence on smoking outcomes. This systematic review discusses methodological gaps in the literature and provides insights for future studies exploring the tobacco environment.

Implications
Our findings pose some methodological lessons to improve the exposure measures on the tobacco outlet environment. To solve these methodological gaps is crucial to understanding the influence of the tobacco environment on the smoking outcomes. Activity spaces should be considered in further analyses since individuals are exposed to tobacco beyond their residence or school neighbourhood. Further studies in this research area demand density estimations weighted by the size of the area, population, or road length; or measured using Kernel Density Estimations. Proximity calculations should be measured through the street network and should consider travel times apart from the length-distance.