Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2026 Jan 1.
Published in final edited form as: J Crime Justice. 2024 Sep 12;48(2):176–189. doi: 10.1080/0735648X.2024.2396406

Using Risk Terrain Modeling and Geographically-Explicit Ecological Momentary Assessments to Examine Alcohol Use in Adults Experiencing Homelessness

Jamie M Gajos 1, Alejandro Gimenez-Santana 2, Jeffery T Walker 3, Karen L Cropsey 4, Scott T Walters 5, Michael S Businelle 6
PMCID: PMC11964400  NIHMSID: NIHMS2023593  PMID: 40183008

Abstract

Adults experiencing homelessness (AEH) are more likely to engage in risky alcohol use compared to adults in the general population. Nonetheless, relatively little is known about the naturalistic settings of alcohol use among AEH. We integrate risk terrain modeling (RTM) with geographically-explicit ecological momentary assessments (GEMAs) to identify the environmental characteristics of drinking locations among AEH in Dallas, Texas. Participants were given a study-provided smartphone pre-installed with the Insight™ application and instructed to self-initiate a GEMA when they were about to consume their first drink of the day over the duration of four weeks. Participants who used this app feature to report alcohol use were included in the analytic sample (N = 22). RTMs estimated the spatial influence of environmental features associated with alcohol availability and risky community features located in the city limits of Dallas, as well as in downtown Dallas. Across the City-Wide and Central Division analyses, proximity to food banks/shelters and public libraries are spatially associated with event locations of AEH alcohol use. Clusters of liquor stores and grocery stores are also significantly related to alcohol event locations. Findings provide insight into the geographical context of alcohol use among AEH and have implications for researchers and practitioners.

Keywords: Alcohol use, Homelessness, Risk terrain modeling, Geographically-explicit EMA (GEMA), Spatial influence

Introduction

The prevalence of adverse behavioral health outcomes and social conditions, such as food insecurity, poor mental health, stigma and discrimination, interpersonal violence, and substance use disorders in adults experiencing homelessness (AEH) is a major public health concern (Loftus et al., 2021; Mejia-Lancheros et al., 2020; Sullivan et al., 2019). One of the most significant contributors to increased morbidity and mortality rates in this population is alcohol misuse (Baggett et al., 2015; Collins et al., 2016). Approximately 38% of AEH engage in daily and/or weekly at-risk drinking behaviors in the past 30 days based on their responses from the Alcohol Quantity and Frequency Questionnaire (Neisler et al., 2019; Room, 1990). Moreover, compared to the general population, AEH are more likely to meet criteria for alcohol dependence (Fazel et al., 2008) and are 6 to 10 times more likely to experience alcohol-related mortality (Baggett et al., 2015). These substantial health disparities are further exacerbated by the fact that AEH are more likely to have contact with the criminal legal system than adults in the general population (Greenberg & Rosenheck, 2008). Relevant to the current study, it is well-established that substance use disorders are linked to incarceration disparities (Shearer et al., 2022). Moreover, drug possession and public drunkenness are among the most prevalent types of arrests among AEH (Gonzalez et al., 2018).

Prior research has found that spatial contexts or built environments are associated with substance use outcomes, such as opioid overdoses, at-risk alcohol consumption, and smoking cessation behaviors (Chichester et al., 2020; Kirchner et al., 2013; Theall et al., 2011). Rates of alcohol use tend to vary by geographical location and several factors associated with geographic areas (e.g., alcohol availability, drinking norms, economic factors) may also influence drinking behaviors (Dixon & Chartier, 2016). For instance, prior research suggests that alcohol outlet visibility, as well as outdoor drinking in public spaces (e.g., in parks, streets) is linked to permissive drinking norms and behaviors (Dimova et al., 2023). Nonetheless, relatively little is known about the naturalistic settings of alcohol use among AEH populations. Compared to housed individuals who typically consume alcohol in bars and restaurants, alcohol use among AEH is more likely to occur outside or in non-residential, non-food service locations (e.g., in parking garages, outside buildings, on sidewalks). Evidence for the relationship between the built environment and alcohol use is mixed, given the methodological differences across studies (Gmel et al., 2016) and that these relationships are likely to vary by population.

Moreover, much of the previous research on alcohol use has relied on retrospectively-reported, cross-sectional, or lab-based methodologies, which do not lend themselves well to estimating the environmental, cognitive, or behavioral antecedents of alcohol use in everyday life. Recall bias is important to consider among vulnerable populations, especially for AEH who may have chaotic daily schedules, as well as problems with memory and cognition. Ecological momentary assessment (EMA) captures the moments of everyday life by administering brief, repeated assessments in natural environments, typically collected through surveys administered on smartphones (Russell & Gajos, 2020; Trull & Ebner-Priemer, 2013). EMA has been successfully administered in studies with AEH (Gajos et al., 2023; Santa Maria et al., 2018; Semborski et al., 2022; Walters et al., 2021). Although EMA methodologies have traditionally relied on self-reports, many EMA studies capture self-reports along with wearable sensors, such as global positional system (GPS) sensors and actigraphy monitors for physical activity (Trull & Ebner-Priemer, 2013). Geographically-explicit ecological momentary assessment (GEMA) is a methodology that captures momentary experiences via brief surveys administered through smartphones, along with passively collected geolocation data (i.e., GPS) (Kirchner & Shiffman, 2016). Thus, each GEMA report is geo-tagged with GPS coordinates. Collecting GEMA data might inform understanding of the unfolding of drinking behaviors in AEH across spatial contexts in daily life.

Another method to study the associations between the built environment and health-risk outcomes is risk terrain modeling (RTM) (Caplan et al., 2011). RTM is a spatial analysis approach that can enhance understanding of the spatial risk factors related to the distribution of event locations (e.g., drinking locations) and identify which geographical areas have the greatest risk for future reported drinking based on the spatial risk factors present in the environment, such as bars, fast-food restaurants, or liquor stores (Szkola et al., 2021). Separate map layers of environmental risk features (i.e., independent variables such as those listed above) hypothesized to be associated with the event outcome (i.e., dependent variable) are combined into a composite risk terrain map with geographic information system (GIS) software (Caplan et al., 2015). RTM estimates the influence of each environmental feature on the event outcome, as well as the distance at which these features are significantly associated with the event outcome. Importantly, the estimated risk values reported in RTM do not suggest the inevitability of an event outcome (such as drinking event locations) but instead highlight locations where the risk for alcohol use is greater (Caplan et al., 2015). RTM is automated through the Risk Terrain Modeling Diagnostics (RTMDx) software (discussed in more detail below).

Current Study

The purpose of the current study is to examine whether features of the built environment are geographically associated with daily drinking event locations over 28 days (collected as GPS coordinates via geographically-explicit EMAs [GEMAs]) among AEH recruited from a homeless shelter in Dallas, Texas. Separate RTMs were analyzed for the City-Wide boundary, as well as for the Central Division of Dallas. We conducted a standalone RTM with the Central Division (i.e., downtown) of Dallas because over half of all the GPS coordinates of the drinking locations (53%) were mapped in the Central Division. Indeed, the clustering of GPS coordinates may be indicative of mobility characteristics among AEH populations compared to housed individuals, such that AEH might not travel long distances due to limited access to private transportation and the necessity to stay relatively close to services that are typically centrally located in city landscapes. To our knowledge, no prior research has examined the environmental features that are geographically associated with alcohol use locations (at the day-level) among a sample of AEH who self-initiated GEMAs.

Materials and Methods

Participants and Procedure

The sample includes homeless individuals who were enrolled in phase I of a three-phase treatment development study (see Businelle et al., 2020 for additional information on the study procedures). All phases of the study were conducted at a large homeless shelter in Dallas, Texas. AEH who were currently receiving services at the shelter (e.g., meals, mental health and substance use counseling, housing placement, and job readiness training) were recruited for the study. During phase I, AEH who reported hazardous drinking (n = 72), were loaned a study-provided smartphone installed with the Insight™ mHealth application and were prompted to complete five geographically-explicit ecological momentary assessments (GEMAs) each day over 28 days. These assessments were designed to measure psychosocial variables such as negative affect, stress, urge to drink and alcohol use, as well as their momentary relationships with social-ecological model constructs like social settings and social support. Relevant to the current study, the event sampling GEMAs were used, where participants were instructed to self-initiate a GEMA on the app when they were about to consume their first drink of the day (i.e., captures their first drinking episode of the day). In total, 22 participants (30% of sample; Mage=48, 86% male, 63% Black) utilized this feature at least once during the four weeks of the study in the Insight™ mHealth app (see more details below) and are included in the current analytical sample.

Outcome Events

We define the GPS coordinates of individuals’ first daily drinking episode locations as the outcome events in the RTM. The drinking event locations were assessed via participant-initiated GEMAs. Participants were instructed to self-initiate a GEMA when they were about to consume their first drink of the day by clicking a button on their study-provided smartphone. The corresponding assessment item was, “Are you about to drink alcohol?” Responses ranged from 1 to 3: 1 = I am about to drink for the first time today, 2 = I have already drunk alcohol today, 3 = I accidentally pressed the drink button. I have not and do not plan to drink today. The responses that indicated a score of “1” (i.e., I am about to drink for the first time today) were included in the RTM analyses in order to obtain the locations most likely to reliably represent the location of an alcohol event in real-time. Moreover, we chose to include all locations of the first drinking event of the day because this represents a unique alcohol event, which may influence subsequent alcohol use that day. Although the participants who selected “2” presumably consumed alcohol that day, we are unable to assess the true location of the drinking event with this option. The maximum number of days a participant used this feature was 6 days and on average, participants used this feature twice (2 days) over the four-week study period. The numbers of alcohol event-reported GEMAs per participant across the study period are presented in Figure 1. The tagged GPS coordinates associated with these GEMA responses were included in the analyses if they yielded a GPS accuracy of ≤30 m, following recommendations from previous research (Kirchner et al., 2013). A total of 45 GPS coordinates obtained from the self-initiated GEMAs during February 2019-January 2020 were used in estimating the RTMs. Figure 2 provides the distribution of the 45 de-identified GPS coordinates within the city of Dallas, where the blue boundary represents the Central Division boundary.

Figure 1:

Figure 1:

Open in a new tab

Alcohol Event-Reported GEMAs per Participant Across Study Duration

Figure 2:

Figure 2:

Open in a new tab

Distribution of GPS Coordinates of Drinking Locations within the City-Wide and Central Division Boundaries of Dallas, Texas

Note: Yellow dots represent GPS coordinates of drinking locations. The blue boundary represents the Central Division of Dallas, Texas

Environmental Risk Factors

Based on our knowledge of the area and locations AEH may visit frequently (Hodgetts et al., 2008; Reitzel et al., 2014; Rodriguez et al., 2009), as well as environmental features that are associated with alcohol use (Dimova et al., 2023; Halonen et al., 2013; Theall et al., 2011), we identified 15 environmental risk factors hypothesized to be associated with drinking event locations. Many of the risk factors were included because of their role in alcohol/drug availability (e.g., liquor stores, bars/nightclubs, smoke/vape shops). We anticipated that locations where alcohol is more readily available would attract opportunities for alcohol use among the sample. Several of the additional risk factors were included because of the likelihood for AEH to travel to community-based locations for food, shelter, and safety reasons (e.g., shelters, hospitals/urgent care centers), which may also be associated with alcohol use. All place feature data for risk locations (e.g., addresses and coordinates of entertainment places, restaurants, convenience stores, gas stations, and hospitals) in Dallas were obtained from Data Axle and were historically relevant for the 2019–2020 period the GEMA data were collected. The environmental risk factors are categorized by alcohol/drug availability features and community features in Table 1. The environmental features include: Bars/nightclubs, casinos, convenience stores, entertainment places, food banks/homeless shelters, gas stations, grocery stores, halls/auditoriums, hotels/motels, liquor stores, public libraries, parks, restaurants, smoke/vape shops/cigar lounges, hospitals/urgent care centers. The distribution of these risk factors within the City-Wide and Central Division boundaries are also presented in Table 1.

Table 1:

Environmental Features included in the City-Wide and Central Division Models

Risk Factor N (City-Wide) N (Central Division)
Alcohol/drug availability features
 Bars/nightclubs 235 102
 Casinos 1 0
 Entertainment places 36 9
 Liquor stores 97 35
 Restaurants 2115 628
 Smoke/vape shops/cigar lounges 67 4
Community features
 Convenience stores 201 46
 Food banks/shelters 10 4
 Gas stations 149 33
 Grocery stores 154 29
 Halls/auditoriums 12 6
 Hospitals/urgent cares 41 11
 Hotels/motels 213 75
 Parks 212 75
 Public libraries 31 10
Open in a new tab

Model Parameters

Models for the City-Wide boundary, as well as for the Central Division of Dallas were generated using RTMDx, a diagnostic spatial analysis platform developed by the Rutgers Center on Public Security (Caplan & Kennedy, 2013). Within the platform, the model presents the optimal spatial operationalization (OP), which examines whether being near risk factors (i.e., proximity) and/or within an area where risk factors cluster (i.e., density) is significantly related to alcohol use locations. For example, RTMDx can test whether the proximity to liquor stores is significantly related to the risk of being an alcohol use location and whether the density of liquor stores increases the risk of being an alcohol use location. Next, the maximum spatial influence (SI) at which environmental risk factors influence the risk of being an alcohol use location (e.g., one, two, or three block lengths) was tested, which was specified at half-block increments for up to three blocks. For the current study, the average block length was calculated at 500 feet, so approximately 250 feet was selected as the raster cell size for the model. This selection resulted in a total of 177,855 half-block cells in the City-Wide prediction map and 8,115 half-block cells in the Central Division prediction map.

Analytic Approach

After operationalizing and selecting a hypothesized spatial influence for each risk factor, RTMDx estimates the relationships between each risk factor (independent variables of the model) and alcohol use locations (our dependent variable). A cross-validation is performed with an elastic net penalization R package which selects variables for a Poisson regression and a negative binomial regression model. The negative binomial distribution includes a parameter to represent over-dispersion of counts, which can help to represent dependency between the alcohol event locations. All factors are selected based on nonzero coefficients in the penalized model. Next, bidirectional stepwise regression on the selected factors is performed using the GAMLSS R package. Risk factors are added and removed from the model one at a time to improve the “best” fit based on the Bayesian Information Criterion (BIC). The final model with the lowest BIC score is selected between the two distributions (i.e., a Poisson and a negative binomial distribution of events). This process results in the optimal model with the most appropriate environmental risk factors and spatial influences. For each risk factor that is significant, estimates of optimized spatial influence (i.e., the distance from the tested environmental risk factor having a greater concentration of drinking events) and a Relative Risk Value (i.e., the weight of the variable within the model) are created. The Relative Risk Value (RRV) is calculated based on the exponential of the regression coefficient (Szkola et al., 2021). The RRV represents the risk associated with a particular spatial feature compared to areas that do not contain that feature. Specific to the present study, RRV is a situational indicator for each cell in the map, where greater RRVs represent cells that have a greater risk of being an alcohol use location relative to other geographical areas without the significant risk factors present (Szkola et al., 2021). For additional information on the analytic approach used in RTMDx, see Caplan et al., 2013.

Results

City-Wide RTM

Results of the RTM for the City-Wide boundary showed the locational effect of food banks and homeless shelters were significantly related to the event locations of alcohol use among adults experiencing homelessness (AEH). The results presented in Table 2 suggest that locations within 250 feet of homeless shelters/food banks were at approximately 200 times higher risk than any other place within the City of Dallas to be an AEH drinking event location. As expected, locations with a concentration of liquor stores were also spatially associated with an increased likelihood of being an event-reported drinking location.

Table 2:

RTMDx Optimal Specifications for Significant, Risk-Predicting Features

Risk Factor City-Wide Central Division
OP SI RRV OP SI RRV
Alcohol/drug availability features
 Bars/nightclubs Proximity 1500 11.79 -- -- --
 Casinos -- -- -- -- -- --
 Entertainment places -- -- -- -- -- --
 Liquor stores Density 250 77.56 Density 250 80.59
 Restaurants -- -- -- -- -- --
 Smoke/vape shops/cigar lounges -- -- -- -- -- --
Community features
 Convenience stores -- -- -- -- -- --
 Food banks/shelters Proximity 250 226.60 Proximity 500 165.97
 Gas stations -- -- -- -- -- --
 Grocery stores Density 1000 3.89 Density 1000 10.33
 Halls/auditoriums -- -- -- -- -- --
 Hospitals/urgent cares -- -- -- -- -- --
 Hotels/motels -- -- -- Proximity 1500 10.38
 Parks -- -- -- -- -- --
 Public libraries Proximity 1000 8.35 Proximity 1000 15.27
Open in a new tab

Note: OP = Operationalization; SI = Spatial Influence (ft.); RRV = Relative Risk Value; Significant findings represent p ≤ 0.05.

In an RTM analysis, each environmental factor is given a relative risk value (RRV) which is the weight of the risk factor that can be compared across factors (Caplan & Kennedy, 2014). Therefore, it is worth noting that drinking locations were more than twice as likely to occur near shelter/food bank locations than near a clustering of liquor stores (see RRV difference of 226 vs 77). Other place features were also associated with drinking locations. These locations included proximity to bars and nightclubs, public libraries, and areas with a concentration of grocery stores and retail spaces. In sum, proximity to shelters and areas with a concentration of liquor stores were identified as the two main situational configurations where participants reported alcohol use. Other environmental factors, such as bars and nightclubs, public libraries, and grocery stores were also associated with drinking locations among AEH. Figure 3 (panel a) displays the high-risk places in the City-Wide model.

Figure 3:

Figure 3:

Open in a new tab

Composite Risk Terrain Maps of the City-Wide (a) and CentraL Division (b) Boundaries of Dallas, Texas

Central Division RTM

The second analysis of this study focused on the downtown district of the City of Dallas. This area is known as the central business district and is characterized by its dense urban environment with limited residential properties, which sets it apart from the rest of the city. Additionally, a small subset of highly clustered locations accounted for most instances where AEH reported drinking alcohol in the Central Division, thereby suggesting the significance of the area for characterizing activity patterns among our sample of AEH. This RTM analysis identified areas within 500 feet of homeless shelters or food banks as the primary locations where AEH reported drinking events in the downtown district of the City of Dallas (see Table 2). The second set of environmental features associated with AEH drinking locations were areas with a concentration of multiple liquor store establishments. These locations likely influence the locations where AEH drink due to greater access to alcohol compared to other areas. Analyses revealed that not all liquor store locations presented the same risk of increased likelihood of nearby drinking. Instead, a subset of highly clustered establishments accounted for a large majority of the risk of being a drinking event location (see Figure 3, panel b). In sum, these two environmental features, homeless shelters and liquor stores, had a strong situational effect on AEH drinking event locations. In addition, proximity to public libraries and hotels/motels, as well as locations with multiple grocery stores were spatially related to drinking event locations.

Discussion

The current study helps to understand the geographical context of drinking locations among AEH in Dallas, Texas. Results from the City-Wide and Central Division RTMs consistently showed that proximity to food banks and shelters presented an elevated environmental risk for being an alcohol use location among AEH. Although alcohol use is not allowed inside of shelters, these findings align with previous research suggesting that being near the shelter may increase the risk of substance use and craving (Businelle et al., 2013; Pratt et al., 2019). In one sense, these findings are intuitive; people are more likely to drink alcohol close to where they eat, sleep, and receive services. On the other hand, shelters may contribute a unique risk of alcohol use, as AEH are likely to interact with previous drinking partners while at the shelter, which has been shown to reduce positive affect when these interactions occur at the shelter versus when not at the shelter (Gajos et al., 2023). These findings have implications for the development of future mHealth interventions, as it may be important to deliver real-time alcohol reduction/abstinence messages via smartphones during moments when AEH are near the shelter. For example, a message might be phrased: “Try to avoid people or situations that remind you of drinking. Seek out people and places that make it easier for you to cut down on drinking” (Walters et al., 2022). Moreover, these findings may also be used to inform the development of future geographically-informed Just-in-Time Adaptive Interventions (JITAIs). Based on the current study findings, geofences could surround identified risk locations and passive sensing via smartphones could prompt real-time alcohol reduction/abstinence messages once participants are within 250–500 feet of a shelter.

In addition to shelter locations, the City-Wide and Central Division RTMs both suggest that proximity to public libraries increases the risk of alcohol use among our sample of AEH. This finding may also be indicative of the areas where AEH visit frequently, as public libraries likely represent a location where AEH spend a significant amount of time (e.g., many libraries provide free Wi-Fi, public restrooms, and may act as a temporary safe haven from exposure to severe weather conditions), as well as provide a place to consume alcohol. The findings from the current study suggest that once participants are within 1,000 feet of public libraries, receiving treatment messages that suggest alternatives to drinking (e.g., “You may be tempted to drink when you are bored or lonely. Do something to keep your mind off drinking. Listen to music, check-in with support staff, or take a walk”) may be beneficial.

Perhaps unsurprisingly, the City-Wide and Central Division models also suggest that high concentrations of liquor stores and grocery stores heighten the risk for drinking event locations. These spatial features are likely linked to greater alcohol availability as well, which may be an important mechanism explaining these findings. Echoing these findings, the City-Wide RTM suggests that proximity to bars/nightclubs is also associated with greater risk for being an alcohol use location, but this finding may not differ from what we might expect with a sample of housed individuals (e.g., housed individuals drink near bars/nightclubs but then travel home). The mechanisms linking AEH alcohol use locations to liquor stores and bars/nightclubs likely represent a different kind of risk than the mechanisms linking shelter proximity to alcohol use locations. For instance, AEH may need reminders to move to different locations or find other people who are not drinking when they are near alcohol availability place features. Future research may seek to examine these unique mechanisms more closely. Finally, the Central Division RTM suggests that proximity (i.e., 1,500 ft) to hotels/motels significantly increases the risk of being a drinking location among AEH. It may be important to consider whether these hotels/motels have bars inside them. Future work may benefit from this distinction as this was outside the scope of the current study. Nonetheless, it might be reasonable to suspect the availability of alcohol at these locations. Proximity to hotels/motels may warrant motivation-themed messages that remind AEH about their health, relationships, and/or the benefits of sobriety.

Limitations

Although the current study focused on aggregate patterns of drinking locations among AEH, a focus on individual-specific patterns in drinking locations would assist practitioners and researchers in understanding spatial behaviors among high-risk populations (Rossmo et al., 2012) and perhaps inform the development of tailored intervention messages for individuals. Future research is needed in this area. First, we are not able to generalize the findings of drinking locations to all AEH who used alcohol during the study, as only a subset of the sample self-initiated the GEMAs in real-time. In turn, those participants who did not report their drinking locations may exhibit other environmental risks associated with drinking. The findings may also capture bias towards the few participants who submitted the alcohol event-reported GEMAs more frequently (see Figure 1). We did not separately incentivize participants to complete the event-reported drinking GEMAs. A primary limitation of the current study is the small percentage of AEH who initiated the event-reported GEMA on the mobile app. However, we take caution in drawing the conclusion that our sample of AEH with risky alcohol use did not engage in alcohol use over the 28-day period. Instead, we hypothesize that the small sample size is more likely to be an indication that many respondents did not use the event-report GEMA on the mobile app or forgot to. Indeed, many AEH have additional cognitive impairments/deficits (Stone et al., 2019) which may decrease their likelihood of self-initiating an event-reported GEMA for alcohol use. Second, it may be possible that the participants who self-initiated the GEMAs did not actually drink alcohol after indicating their intention to consume their first drink of the day, but we are unable to confidently confirm this. Another limitation concerns the relatively small number of GPS coordinates that were analyzed in the RTMs. Many GPS coordinates did not yield high accuracy levels—especially if respondents were located inside buildings at the time GEMAs were administered. We tried to address this limitation by selecting only the GPS coordinates with an accuracy of ≤30 m based on prior work (Kirchner et al., 2013) to yield more reliable estimates of location. Moreover, any GPS coordinates of drinking locations that were identified outside of the City-Wide boundary and/or the Central Division of Dallas were not included in the RTMs because the models will only include locations of outcome events that are located within the defined boundary layers. Therefore, we are unable to generalize the findings to drinking locations reported outside of the boundary layer of the city of Dallas. Finally, the possibility that the influence of the identified risk factors on drinking locations might vary temporarily (across time of day or seasonally) is another area for future research. The timing of the drinking event locations was limited to the “first drink of the day”; thus, the average reported time of the first drink among the sample was approximately 1:00 PM. Future work may wish to examine whether different environmental risk factors are associated with daytime versus nighttime drinking locations.

Conclusions

Utilizing GEMA methods via data collected through mobile devices—such as GPS enabled smartphones—has the potential to inform prevention and intervention efforts targeting mental and behavioral health by linking momentary experiences to objective measures of socio-ecological contexts in real-time (Kirchner & Shiffman, 2016). The current study suggests that the application of risk terrain modeling could increase our understanding of the locations where alcohol is used among AEH, and aid in the development of tailored interventions that address the spatial dynamics of alcohol use for vulnerable populations (Caplan et al., 2017).

Acknowledgements:

This work was primarily supported by R34AA024584 to MSB and STW. JMG’s work on this study was supported by K01DA054262. Programming and technological support were provided through the mobile health shared resource of the Stephenson Cancer Center via an NCI Cancer Center Support Grant (P30CA225520) and through the Oklahoma Tobacco Settlement Endowment Trust grant 092-016-0002.

Biography

Jamie M. Gajos, PhD, is an Assistant Professor in the Department of Family and Community Medicine in the Heersink School of Medicine at the University of Alabama at Birmingham. Her research program utilizes innovative statistical methodologies to understand and predict substance use and related health-risk behaviors and to disseminate the research findings to inform future mHealth interventions. Much of her work uses longitudinal data to understand the etiology and developmental patterns of health-risk outcomes, with a particular interest in testing dynamic person-environment interactions with geographically-explicit ecological momentary assessment (GEMA) data.

Alejandro Giménez-Santana, PhD, is an Assistant Professor of Professional Practice in the School of Criminal Justice at Rutgers University, where he serves as Co-Executive Director of the Newark Public Safety Collaborative (NPSC). He has worked extensively in researching the association between unique contexts of social disorganization and crime risk on the spatial distribution of violence across various urban settings.

Jeffery T. Walker, PhD, is University Professor and the J. Frank Barefield Endowed Chair of Communities and Crime in the Department of Criminal Justice at the University of Alabama at Birmingham. Dr. Walker has expertise in the study of crime and neighborhoods, community determinants of health, and spatial and crime analysis.

Karen L. Cropsey, PsyD, is the Conatser Turner Endowed Professor in the Department of Psychiatry and Behavioral Neurobiology at the Heersink School of Medicine at the University of Alabama at Birmingham (UAB) and is the Director of the Center for Addiction and Pain Prevention and Intervention (CAPPI) at UAB. Dr. Cropsey’s career has focused on treating addiction in vulnerable and stigmatized populations, such as individuals in the criminal justice system.

Scott T. Walters, PhD, is Regents Professor in the Department of Population & Community Health in the School of Public Health at the University of North Texas Health Science Center. His experience ranges from brief interventions for underage drinking, to adults in the criminal justice system, to heavy drinkers in hospital settings, to cancer risk screening interventions, to community-based health navigation.

Michael S. Businelle, PhD, is the Peggy and Charles Stephenson Endowed Chair in Cancer, Co-Director of the TSET Health Promotion Research Center, and Professor in the Department of Family and Preventive Medicine at the University of Oklahoma Health Sciences Center. His research focuses on smoking cessation, alcohol cessation, case management, anxiety/depression, and COVID-19 in cancer patients. His primary goal is to improve understanding of the causes of health disparities and to create and disseminate effective smartphone based just-in-time adaptive interventions (JITAIs) that reduce health disparities.

Appendix:

Poisson Analysis Estimates for Significant Risk-Predicting Features

Risk Factor City-Wide Central Division
Estimate SE t-value Estimate SE t-value
Alcohol/drug availability features
 Bars/nightclubs 2.47 0.41 5.96 -- -- --
 Casinos -- -- -- -- -- --
 Entertainment places -- -- -- -- -- --
 Liquor stores 4.35 0.56 7.77 4.39 0.71 6.19
 Restaurants -- -- -- -- -- --
 Smoke/vape shops/cigar lounges -- -- -- -- -- --
Community features
 Convenience stores -- -- -- -- -- --
 Food banks/shelters 5.42 0.46 11.75 5.11 0.56 9.11
 Gas stations -- -- -- -- -- --
 Grocery stores 1.36 0.36 3.80 2.34 0.50 4.69
 Halls/auditoriums -- -- -- -- -- --
 Hospitals/urgent cares -- -- -- -- -- --
 Hotels/motels -- -- -- 2.34 0.64 3.67
 Parks -- -- -- -- -- --
 Public libraries 2.12 0.46 4.65 2.73 0.57 4.79
Open in a new tab

Note: SE = Standard Error; Significant findings represent p ≤ 0.05.

Footnotes

Declaration of Interests

Dr. Businelle is the primary inventor of the Insight mHealth Platform, which was used in the current study. He receives royalties related to its use, but since he was a PI on R34AA024584, he did not receive royalties in this case.

References

  1. Baggett TP, Chang Y, Singer DE, Porneala BC, Gaeta JM, O’Connell JJ, & Rigotti NA (2015). Tobacco-, alcohol-, and drug-attributable deaths and their contribution to mortality disparities in a cohort of homeless adults in Boston. American Journal of Public Health, 105(6), 1189–1197. 10.2105/AJPH.2014.302248 [DOI] [PMC free article] [PubMed] [Google Scholar]
  2. Businelle MS, Cuate EL, Kesh A, Poonawalla IB, & Kendzor DE (2013). Comparing homeless smokers to economically disadvantaged domiciled smokers. American Journal of Public Health, 103(S2), S218–S220. 10.2105/AJPH.2013.301336 [DOI] [PMC free article] [PubMed] [Google Scholar]
  3. Businelle MS, Walters ST, Mun E-Y, Kirchner TR, Hébert ET, & Li X (2020). Reducing drinking among people experiencing homelessness: Protocol for the development and testing of a just-in-time adaptive intervention. JMIR Research Protocols, 9(4), e15610. 10.2196/15610 [DOI] [PMC free article] [PubMed] [Google Scholar]
  4. Caplan JM, & Kennedy LW (2013). Risk terrain modeling diagnostics utility (Version 1.0). Newark, NJ: Rutgers Center on Public Security [Google Scholar]
  5. Caplan JM, Kennedy LW, Barnum JD, & Piza EL (2015). Risk terrain modeling for spatial risk assessment. Cityscape 17(1), 7–16. [Google Scholar]
  6. Caplan JM, Kennedy LW, Barnum JD, & Piza EL (2017). Crime in context: Utilizing risk terrain modeling and conjunctive analysis of case configurations to explore the dynamics of criminogenic behavior settings. Journal of contemporary criminal justice, 33(2), 133–151. 10.1177/10439862166888 [DOI] [Google Scholar]
  7. Caplan JM, Kennedy LW, & Miller J (2011). Risk terrain modeling: Brokering criminological theory and GIS methods for crime forecasting. Justice Quarterly 28(2), 360–381. 10.1080/07418825.2010.486037 [DOI] [Google Scholar]
  8. Caplan JM, Kennedy LW, & Piza EL (2013). Risk Terrain Modeling Diagnostics Utility User Manual (Version 1.0). Newark, NJ: Rutgers Center on Public Security. https://www.rutgerscps.org/uploads/2/7/3/7/27370595/rtmdxusermanual_final_caplankennedypiza.pdf [Google Scholar]
  9. Chichester K, Drawve G, Giménez-Santana A, Sisson M, McCleskey B, Dye DW, Walker J, Mrug S, & Cropsey K (2020). Pharmacies and features of the built environment associated with opioid overdose: A geospatial comparison of rural and urban regions in Alabama, USA. International Journal of Drug Policy, 79, 1027–1036. 10.1016/j.drugpo.2020.102736. [DOI] [PubMed] [Google Scholar]
  10. Collins SE, Jones CB, Hoffmann G, Nelson LA, Hawes SM, Grazioli VS, Mackelprang JL, Holttum J, Kaese G, & Lenert J (2016). In their own words: Content analysis of pathways to recovery among individuals with the lived experience of homelessness and alcohol use disorders. International Journal of Drug Policy, 27, 89–96. 10.1016/j.drugpo.2015.08.003 [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Dimova ED, Lekkas P, Maxwell K, Clemens TL, Pearce JR, Mitchell R, et al. (2023). Exploring the influence of local alcohol availability on drinking norms and practices: A qualitative scoping review. Drug Alcohol Rev., 42(3): 691–703. 10.1111/dar.13596 [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Dixon MA, & Chartier KG (2016). Alcohol use patterns among urban and rural residents: Demographic and social influences. Alcohol Research: Current Reviews 38(1), 69–77. [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. Fazel S, Khosla V, Doll H, & Geddes J (2008). The prevalence of mental disorders among the homeless in western countries: systematic review and meta-regression analysis. PLoS medicine, 5(12), e225. 10.1371/journal.pmed.0050225 [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Gajos JM, Oliver JA, Hébert ET, Walters ST, & Businelle MS (2023). Does the relationship between affect and social interactions among adults experiencing homelessness differ during moments when at a shelter versus not? Mental Health Science, 1–6. 10.1002/mhs2.47 [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Gmel G, Holmes J, & Studer J (2016). Are alcohol outlet densities strongly associated with alcohol‐related outcomes? A critical review of recent evidence. Drug and Alcohol Review 35(1), 40–54. 10.1111/dar.12304 [DOI] [PubMed] [Google Scholar]
  16. Gonzalez JR, Jetelina KK, Roberts M et al. (2018). Criminal Justice System Involvement Among Homeless Adults. Am J Crim Just 43, 158–166. 10.1007/s12103-017-9413-7 [DOI] [Google Scholar]
  17. Greenberg GA, & Rosenheck RA (2008). Jail incarceration, homelessness, and mental health: A national study. Psychiatric Services, 59(2), 170–177. [DOI] [PubMed] [Google Scholar]
  18. Halonen JI, Kivimäki M, Virtanen M, Pentti J, Subramanian SV, Kawachi I, & Vahtera J (2013). Living in proximity of a bar and risky alcohol behaviours: a longitudinal study. Addiction, 108(2), 320–328. 10.1111/j.1360-0443.2012.04053.x [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. Hodgetts D, Stolte O, Chamberlain K, Radley A, Nikora L, Nabalarua E, & Groot S (2008). A trip to the library: Homelessness and social inclusion. Social & Cultural Geography, 9(8), 933–953. 10.1080/14649360802441432 [DOI] [Google Scholar]
  20. Kirchner TR, Cantrell J, Anesetti-Rothermel A, Ganz O, Vallone DM, & Abrams DB (2013). Geospatial exposure to point-of-sale tobacco: Real-time craving and smoking-cessation outcomes. American Journal of Preventive Medicine 45(4), 379–385. 10.1016/j.amepre.2013.05.016. [DOI] [PMC free article] [PubMed] [Google Scholar]
  21. Kirchner TR, & Shiffman S (2016). Spatio-temporal determinants of mental health and well-being: Advances in geographically-explicit ecological momentary assessment (GEMA). Social psychiatry and psychiatric epidemiology, 51, 1211–1223. 10.1007/s00127-016-1277-5 [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Loftus EI, Lachaud J, Hwang SW, & Mejia-Lancheros C (2021). Food insecurity and mental health outcomes among homeless adults: A scoping review. Public Health Nutrition 24(7), 1766–1777. 10.1017/S1368980020001998 [DOI] [PMC free article] [PubMed] [Google Scholar]
  23. Mejia-Lancheros C, Lachaud J, O’Campo P, Wiens K, Nisenbaum R, Wang R, Hwang SW, & Stergiopoulos V (2020). Trajectories and mental health-related predictors of perceived discrimination and stigma among homeless adults with mental illness. PloS ONE 15(2), e0229385. 10.1371/journal.pone.0229385 [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. Neisler J, Shree S, Reitzel LR, Chen T-A, Kendzor DE, Obasi EM, Wrighting Q, & Businelle MS (2019). Characterizing alcohol use behaviors among homeless men and women. American Journal of Health Behavior 43(1), 37–49. 10.5993/AJHB.43.1.4 [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. Pratt R, Pernat C, Kerandi L, Kmiecik A, Strobel-Ayres C, Joseph A, Everson Rose SA, Luo X, Cooney N, & Thomas J (2019). “It’s a hard thing to manage when you’re homeless”: the impact of the social environment on smoking cessation for smokers experiencing homelessness. BMC Public Health, 19(1), 1–8. 10.1186/s12889-019-6987-7 [DOI] [PMC free article] [PubMed] [Google Scholar]
  26. Reitzel LR, Kendzor DE, Nguyen N, Regan SD, Okuyemi KS, Castro Y, Wetter DW, & Businelle MS (2014). Shelter proximity and affect among homeless smokers making a quit attempt. American Journal of Health Behavior, 38(2), 161–169. 10.5993/AJHB.38.2.1 [DOI] [PMC free article] [PubMed] [Google Scholar]
  27. Rodriguez RM, Fortman J, Chee C, Ng V, & Poon D (2009). Food, shelter and safety needs motivating homeless persons’ visits to an urban emergency department. Annals of Emergency Medicine 53(5), 598–602. e591. 10.1016/j.annemergmed.2008.07.046. [DOI] [PubMed] [Google Scholar]
  28. Room R (1990). Measuring Alcohol Consumption in the United States. In: Kozlowski LT, et al. Research Advances in Alcohol and Drug Problems. Springer, Boston, MA. 10.1007/978-1-4899-1669-3_3 [DOI] [Google Scholar]
  29. Rossmo DK, Lu Y, & Fang TB (2012). Spatial-temporal crime paths. In Andresen MA & Kinney JB (Eds.), Patterns, prevention, and geometry of crime (pp. 16–42). Routledge. [Google Scholar]
  30. Russell MA, & Gajos JM (2020). Annual Research Review: Ecological momentary assessment studies in child psychology and psychiatry. Journal of Child Psychology and Psychiatry, 61(3), 376–394. 10.1111/jcpp.13216 [DOI] [PMC free article] [PubMed] [Google Scholar]
  31. Santa Maria D, Padhye N, Yang Y, Gallardo K, Santos G-M, Jung J, & Businelle M (2018). Drug use patterns and predictors among homeless youth: Results of an ecological momentary assessment. The American Journal of Drug and Alcohol Abuse, 44(5), 551–560. 10.1080/00952990.2017.1407328 [DOI] [PMC free article] [PubMed] [Google Scholar]
  32. Semborski S, Winn JG, Rhoades H, Petry L, & Henwood BF (2022). The application of GIS in homelessness research and service delivery: A qualitative systematic review. Health & Place, 75, 102776. 10.1016/j.healthplace.2022.102776 [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. Shearer RD, Shippee ND, Vickery KD, Stevens MA, & Winkelman TN (2022). A longitudinal cross-sectional analysis of substance use treatment trends for individuals experiencing homelessness, criminal justice involvement, both, or neither-United States, 2006–2018. The Lancet Regional Health–Americas, 7, 1–13. 10.1016/j.lana.2021.100174 [DOI] [PMC free article] [PubMed] [Google Scholar]
  34. Stone B, Dowling S, & Cameron A (2019). Cognitive impairment and homelessness: A scoping review. Health & Social Care in the Community, 27(4), e125–e142. 10.1111/hsc.12682 [DOI] [PMC free article] [PubMed] [Google Scholar]
  35. Sullivan CM, Bomsta HD, & Hacskaylo MA (2019). Flexible funding as a promising strategy to prevent homelessness for survivors of intimate partner violence. Journal of interpersonal violence, 34(14), 3017–3033. 10.1177/0886260516664318 [DOI] [PubMed] [Google Scholar]
  36. Szkola J, Piza EL, & Drawve G (2021). Risk terrain modeling: Seasonality and predictive validity. Justice Quarterly, 38(2), 322–343. 10.1080/07418825.2019.1630472 [DOI] [Google Scholar]
  37. Theall KP, Lancaster BP, Lynch S, Haines RT, Scribner S, Scribner R, & Kishore V (2011). The neighborhood alcohol environment and at‐risk drinking among African‐Americans. Alcoholism: Clinical and Experimental Research, 35(5), 996–1003. 10.1111/j.1530-0277.2010.01430.x [DOI] [PMC free article] [PubMed] [Google Scholar]
  38. Trull TJ, & Ebner-Priemer U (2013). Ambulatory assessment. Annual review of clinical psychology, 9, 151–176. 10.1146/annurev-clinpsy-050212-185510 [DOI] [PMC free article] [PubMed] [Google Scholar]
  39. Walters ST, Businelle MS, Suchting R, Li X, Hébert ET, & Mun E-Y (2021). Using machine learning to identify predictors of imminent drinking and create tailored messages for at-risk drinkers experiencing homelessness. Journal of Substance Abuse Treatment 127, 108417. 10.1016/j.jsat.2021.108417 [DOI] [PMC free article] [PubMed] [Google Scholar]
  40. Walters ST, Mun EY, Tan Z, Luningham JM, Hébert ET, Oliver JA, & Businelle MS (2022). Development and preliminary effectiveness of a smartphone‐based, just‐in‐time adaptive intervention for adults with alcohol misuse who are experiencing homelessness. Alcoholism: Clinical and Experimental Research, 46(9), 1732–1741. 10.1111/acer.14908 [DOI] [PMC free article] [PubMed] [Google Scholar]

RESOURCES