2025 Theses Doctoral

Individuals’ Exposure to Alcohol Environments: Measurement, Mobility, and Money

Mehranbod, Christina

Alcohol consumption is a significant issue both globally and domestically in the United States (US). Alcohol outlet density is a modifiable environmental factor that has been the focus of interventions to reduce alcohol consumption and alcohol-related harm. With almost 50 years of research on this topic, researchers have not reached a consensus on the true effect of outlet density on alcohol consumption, given the mixed evidence produced by the field in many contexts and populations (positive associations, negative associations, null associations). Despite publications of multiple systematic reviews synthesizing the research, significant empirical doubt remains on the association between outlet density and alcohol consumption. This dissertation proposes three main issues that may be reasons to produce mixed evidence: measurement, mobility, and money. That is, measurement is quite varied in the field, there exists a gap in the incorporation of geographic mobility knowledge in understanding and measuring exposure to alcohol outlets, and lastly, two potential effect modifiers, neighborhood income and individual income (i.e., money), have not been tested empirically as a reason for varying effects. This dissertation aims to explore each of these three concepts.

The three overarching objectives mentioned above are split among three dissertation aims. Aim 1 presents the results of a systematic review of all measurement techniques that have permeated the alcohol outlet density literature (measurement issue). Aim 2 quantifies the impact of various measurement methods using different neighborhood definitions on the estimates of the association of alcohol outlet density and alcohol consumption (measurement issue) and describes mobility in relation to factors related to alcohol consumption (mobility issue). Finally, Aim 3 explores income as an effect modifier of the alcohol outlet density and alcohol consumption relationship (money issue).

Chapter 2 presents the results of a systematic review that summarizes the state of alcohol outlet density measurement methods. Previous reviews focusing on alcohol outlet density summarized findings up to October 2014 without identifying differences in measurement methods that may drive the mixed effects permeating the literature. The aim of this systematic review was to describe the ways alcohol outlet density has been measured over the last 40 years and identify the theoretical justification of chosen measurement methods. In May 2024, literature searches of PubMed, Embase, Scopus, and Web of Science identified articles published after October 2014, the period following the last systematic review on alcohol outlet density. Papers synthesized in the three main systematic reviews prior were also included in this review due to the unique study focus. Studies must have presented alcohol outlet density as the exposure and any alcohol consumption measure or specific acute alcohol-related consequences (e.g., motor vehicle crashes, suicide, intimate partner violence, and violent crime) as outcomes. Three reviewers extracted information on research setting, study design, unit of analysis, measurement method, and whether justification for the neighborhood selection was provided, among other details.

This review identified 209 studies relating alcohol outlet density to alcohol consumption or alcohol-related harm. Container-based methods were the most used method of measuring alcohol outlet density, with 90% of studies using these methods. On average, ecological studies produced more positive associations than individual level studies and individual level studies produced more negative associations than ecological studies. Lastly, less than one-third of papers justified their method of measuring alcohol outlet density. This systematic review underscores wide variation in measurement method in the field, pointing to a potential reason for mixed findings. With the dominance of container-based methods, accompanied by a lack of theoretical justification for their use, and the well-documented spatial biases of container-based methods, future work in the field should look to establish methodological consistency and rigor by moving towards distance- and access-based measures.

Studies examining singular cities have documented the impact of using varying neighborhood definitions to calculate alcohol outlet density on alcohol consumption outcomes; however, it is unclear whether these impacts persist in multiple cities across the United States. Further in the identification of the appropriate neighborhood definition to study alcohol outlet density, understanding the mobility characteristics of populations in relation to factors related to alcohol consumption (e.g., income, age) remains imperative to understand the true exposure area for different individuals.

The study in Chapter 3 aimed to 1) compare residence-based and activity space-based measures of alcohol outlet density and identify differences in their associations with various alcohol consumption measures and 2) explore geographic mobility and neighborhood discordance by individual-level and neighborhood-level factors related to alcohol consumption. This cross-sectional study used data of participants SafeNights study that included GPS information and survey measures (n=147). I constructed on-premise alcohol outlet density measures using activity path buffers of 50m, 100m. and 200m, convex hulls, one standard deviational ellipses, home buffers of 50m, 100m, 200m, 400m, 800m, 805m, 1609m, and administrative units (census block groups, census tracts, ZIP codes, and counties). Friedman tests compared overall differences in measurements using varying neighborhood definitions.

Using negative binomial models, alcohol outlet density measurements were used to predict alcohol consumption, adjusting for individual age, income, education, race, and employment. Lastly, bivariate models reported differences in average daily distance traveled, proportion of activity space outside of the 400m and 800m buffer, and proportion of activity space outside of the 400m and 800m buffer. Lastly, this study quantified neighborhood discordance by comparing ZIP codes of bars visited and homes. This study found correlations between activity path buffers and other activity space-based measures, in addition to correlations between census tract and census block group, 800m, 805m and 1-mile buffers around home. On-premise outlet density was not consistently associated with average number of drinks on each drinking day, maximum number of drinks on each drinking day, drinking at bars or drinking at restaurants. This study found mobility to vary by income, age, and race, and strong evidence of neighborhood discordance between residential areas and bar locations. These findings suggest the use of different neighborhood definitions for different population subgroups in a study, especially when examining access to on-premise outlets that tend to be located away from home neighborhoods.

Chapter 4 explored another potential reason for the mixed findings found in the alcohol outlet density literature. To date, no studies had explored neighborhood income and individual income as effect modifiers of the relationship between outlet density and alcohol consumption. With mechanisms, such as the market potential mechanisms and environmental injustice mechanisms, that place alcohol outlets in the areas nearest to concentrations of individuals who have a greater propensity to drink due to higher personal incomes, and disparate relationships between income and alcohol consumption, I hypothesized that the relationship between on-premise outlet density and alcohol consumption would vary by strata of neighborhood income and personal income. This cross-sectional study of 3,762 SafeNights participants utilized information on local and lagged on-premise outlet density to predict drinking days at bars, restaurants, and home. If negative binomial models produced a significant effect, I then stratified by neighborhood income and individual income. Associations were tested on multiple spatial scales (census block group, census tract, ZIP code, city, and county).

This study did not find a significant association between lagged on-premise outlet density and drinking at bars in the high neighborhood income – high personal income stratum but did find a significant association when examining local on-premise outlet density multiple spatial scales. Further, there was preliminary evidence of a null association between local on-premise outlet density and drinking days at bars on most spatial scales for the low neighborhood income-low personal income group. These findings suggest potential varying effects of the relationship between outlet density and alcohol consumption stratified by different income levels. Future work should test effect modification through multi-scale models to also isolate the spatial scale with the most impact on this association.

This dissertation aimed to disentangle the mixed findings in the research base examining the impact of alcohol outlet density and alcohol consumption. Neighborhood definition varied widely in the literature, preventing comparison of studies and ultimately posing a barrier to understanding the effect of outlet density on consumption. Mobility characteristics also varied by individual-level characteristics that are also associated with alcohol consumption. Lastly, the dissertation produced preliminary evidence for effect modification by income, necessitating future multi-scale models to fully understand on which level the influence of outlets on alcohol consumption behaviors lies.