Many approaches are available to researchers who wish to measure individuals’ exposure to environmental conditions. Different approaches may yield different estimates of associations with health outcomes. Taking adolescents’ exposure to alcohol outlets as an example, we aimed to (1) compare exposure measures and (2) assess whether exposure measures were differentially associated with alcohol consumption.
We tracked 231 adolescents 14–16 years of age from the San Francisco Bay Area for 4 weeks in 2015/2016 using global positioning systems (GPS). Participants were texted ecologic momentary assessment surveys six times per week, including assessment of alcohol consumption. We used GPS data to calculate exposure to alcohol outlets using three approach types: residence-based (e.g., within the home census tract), activity location–based (e.g., within buffer distances of frequently attended places), and activity path–based (e.g., average outlets per hour within buffer distances of GPS route lines). Spearman correlations compared exposure measures, and separate Tobit models assessed associations with the proportion of ecologic momentary assessment responses positive for alcohol consumption.
Measures were mostly strongly correlated within approach types (ρ ≥ 0.7), but weakly (ρ < 0.3) to moderately (0.3 ≤ ρ < 0.7) correlated between approach types. Associations with alcohol consumption were mostly inconsistent within and between approach types. Some of the residence-based measures (e.g., census tract: β = 8.3, 95% CI = 2.8, 13.8), none of the activity location–based approaches, and most of the activity path–based approaches (e.g., outlet–hours per hour, 100 m buffer: β = 8.3, 95% CI = 3.3, 13.3) were associated with alcohol consumption.
Methodologic decisions regarding measurement of exposure to environmental conditions may affect study results.
From the aDepartment of Epidemiology, Mailman School of Public Health, Columbia University, New York, NY
bDepartment of Epidemiology and Preventive Medicine, Monash University, Melbourne, VIC, Australia
cPrevention Research Center, Pacific Institute for Research and Evaluation, Berkeley, CA
dDepartment of Pediatrics, Indiana University School of Medicine, Indianapolis, IN
eDepartment of Biostatistics, Epidemiology and Informatics, Perelman School of Medicine, University of Pennsylvania, Philadelphia, PA.
Submitted June 1, 2018; accepted October 17, 2018.
This study was funded by the National Institute for Child Health and Human Development (R01HD078415-01A1)
The authors report no conflicts of interest.
Supplemental digital content is available through direct URL citations in the HTML and PDF versions of this article (www.epidem.com).
Data are not publicly available to protect human subjects’ confidentiality. Code for the statistical analysis is available from the authors by request.
Correspondence: Christopher N. Morrison, Department of Epidemiology, Mailman School of Public Health, Columbia University, 722 West 168th Street, R505 New York, NY 10032. E-mail: email@example.com.