Key Points
Question
What is the association of a state-mandated reduction of the number of hours when bars/taverns may sell or serve alcohol with prevention of violent crime near these outlets?
Findings
This interrupted time series analysis of 26 bars/taverns in Baltimore, Maryland, found that reducing hours of alcohol sales from 6 am to 2 am to 9 am to 10 pm was associated with a 23% annual decrease in all violent crime compared with control areas. Furthermore, homicide rates dropped by 51% in the first month post intervention and 40% annually thereafter.
Meaning
These findings suggest that statutory restriction of hours of alcohol sales for bars/taverns may serve as a model for other cities looking to create safer neighborhoods.
This interrupted time series analysis assesses the association of reduced hours of alcohol sales at bars/taverns with violent crime rates in Baltimore.
Abstract
Importance
It is well established that alcohol outlets (ie, places that sell alcohol) attract crime, particularly during late-night hours.
Objective
To evaluate the association of Maryland Senate Bill 571 (SB571), which reduced the hours of sale for bars/taverns in 1 Baltimore neighborhood from 6 am to 2 am to 9 am to 10 pm, with violent crime within that neighborhood.
Design, Setting, and Participants
This controlled interrupted time series analysis compared the change in violent crime density within an 800-ft buffer around bars/taverns in the treatment neighborhood (ie, subject to SB571) and 2 control areas with a similar mean baseline crime rate, alcohol outlet density, and neighborhood disadvantage score in the City of Baltimore between May 1, 2018, and December 31, 2022. The interrupted time series using Poisson regression with overdispersion adjustment tested whether the violent crime density differed before vs after the policy change in the treatment neighborhood and whether this difference was localized to the treatment neighborhood.
Exposure
Statutory reduction of bar/tavern selling hours from 20 to 13 hours per day in the treatment neighborhood.
Main Outcomes and Measures
The primary outcome was all violent crime, including homicide, robbery, aggravated and common assault, and forcible rape. Secondary outcomes were homicides and assaults. All violent crime measures summed the monthly incidents within 800 ft of bars/taverns from 8 pm to 4 am. For each outcome, a level change estimated the immediate change (first month after implementation), and a slope change estimated the sustained change after implementation (percent reduction after the first month). These level and slope changes were then compared between the treatment and control neighborhoods.
Results
The treatment neighborhood included 26 bars/taverns (mean [SD] population, 524.6 [234.6] residents), and the control neighborhoods included 41 bars/taverns (mean [SD] population per census block, 570.4 [217.4] residents). There was no immediate level change in density of all violent crimes the month after implementation of SB571; however, compared with the control neighborhoods, the slope of all violent crime density decreased by 23% per year in the treatment neighborhood after SB571 implementation (annualized incidence rate ratio, 0.77; 95% CI, 0.60-0.98; P = .04). Similar results were seen for homicides and assaults. Several sensitivity analyses supported the robustness of these results.
Conclusions and Relevance
This study’s findings suggest that alcohol policies that reduce hours of sale could be associated with a reduction in violent crimes. Given these findings, SB571 may serve as a model for other cities looking to create safer neighborhoods.
Introduction
In 2020, there were nearly 900 000 emergency department visits for assaults in the US.1 Systematic reviews have established a link between alcohol outlets and violent crime.2,3,4 Natural experiments have suggested that violent crime rises and falls with the number of alcohol outlets.5,6 However, to our knowledge, there has only been 1 published evaluation of an intentional effort to reduce violent crime in the US by influencing outlet-level bar availability. This natural experiment occurred in a relatively high-income neighborhood,5 whereas numerous studies have documented that alcohol outlets that sell for off-site consumption (eg, bars/taverns) cluster in low-income and historically marginalized communities.7,8
Related strategies to reduce violence include limiting bar/tavern density3 or operating hours9 or increasing monitoring and compliance checks regarding sales to intoxicated patrons.10 Limiting late-night hours of sale may be effective because later closing hours enable patrons to drink longer, promoting higher intoxication levels and ultimately facilitating alcohol-related violence,11 and when bars/taverns close after other businesses, there are few bystanders or potential guardians to prevent violence.
A 2010 systematic review suggested that reducing alcohol sales by 2 hours may mitigate alcohol-related harms but did not find evaluations of such US policies or any policies restricting hours of sale.9 Evaluations abroad have demonstrated alcohol sales policies’ potential to reduce violent crime. A July 2002 policy in Diadema, Brazil, which closed bars at 11 pm instead of operating for 24 hours was associated with 9 fewer monthly homicides between January 1999 and July 2005.12 An evaluation of a COVID-19 policy changing closing hours from 5 am to 12 am concluded that the change in hours was associated with 28% fewer violent crimes per square kilometer in Copenhagen, Denmark.13 However, these evaluations studied crime throughout the day, despite late-night crimes being the most likely to be alcohol related.11,12,13
Maryland Senate Bill 571 (SB571), effective July 1, 2020, which reduced the late-night hours for bars/taverns in 1 Baltimore neighborhood, presented a unique opportunity for the first US evaluation of a natural experiment limiting alcohol outlet hours of sale. Before SB571, bars/taverns operated for 20 hours a day (from 6 am to 2 am). The bill reduced the hours of sale in this neighborhood by 7 hours, permitting sales only from 9 am to 10 pm.
This study assessed whether SB571 was associated with reduced late-night violent crime near bars/taverns in the bill catchment (henceforth referred to as treatment) neighborhood compared with control neighborhoods. We hypothesized that SB571 would be associated with decreases in late-night violent crime in the treatment neighborhood but not the control neighborhood.
Methods
Study Overview
The aim of SB571 was to reduce violent crime in 1 high-priority neighborhood (treatment area) by reducing operating hours for alcohol outlets holding an LBD-7 license, a unique bar/tavern license that permits both on-premise and off-premise sales. It passed on March 15, 2020, and went into effect July 1, 2020.14 This interrupted time series (ITS) analysis used data for 3752 census block (CB)–months spanning May 1, 2018, to December 31, 2022. We compared late-night monthly change in crime density near bars/taverns in the treatment area to similar Baltimore neighborhoods. This study does not meet the criteria for human participants research and did not require ethical review under the Common Rule. This study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline.
Treatment Area
Maryland SB571 provided cross streets for the treatment area.14 We drew a polygon around the boundary streets using the ArcGIS create polygon function (Esri). The treatment area was 1.01 square miles and contained approximately 9000 residents in 2019.
Control Areas
We defined control locations using community statistical areas (CSAs). Unique to Baltimore, CSAs are demarcated neighborhood units established by the city’s planning department.15 The CSAs were selected as control locations for their geographic similarity to the treatment area and alignment with Baltimore neighborhoods.16 We ranked CSAs according to their similarity to the treatment area’s baseline alcohol outlet density (per 1000 population), total violent crime rate (per 1000 population), and neighborhood disadvantage (eTable 1 in Supplement 1). To accomplish this ranking, we calculated the differences between each CSA and treatment area value, averaging them across the 3 measures to obtain 1 score for each CSA. We excluded 4 CSAs that intersected the treatment area, resulting in 51 potential control locations.
We defined 2 sets of control areas. The primary control areas were used in the main analyses and included the 2 neighborhoods ranked first (Southwest Baltimore) and second (Midway/Coldstream) in similarity. The secondary control areas used in sensitivity analyses added CSAs ranked third to fifth in similarity (Sandtown-Winchester/Harlem Park, Poppleton/The Terraces/Hollins Market, and Greater Charles Village/Barclay).
Unit of Analysis
An ITS was performed at the CB level, the smallest geographic area for which the US Census Bureau collects data. Census blocks approximate city blocks and contain 15 to 75 people.17 We only included CBs with a bars/tavern because bars/taverns were the reference point for violent crimes. There were 31 bars/taverns in the treatment area and 136 in the 5 control areas. After excluding 56 bars/taverns that closed or moved during the study period, there were 26 in the treatment area and 41 in the primary control areas, resulting in a total of 67 CBs for each of the 56 months.
Measures
Outcome: Violent Crime
Violent crime incident data were obtained through Open Baltimore,18 which includes crime date, time, type, and coordinates. We had 1 primary outcome (total violent crime) and 2 secondary outcomes (homicide and assault). All violent crime measures summed the monthly number of incidents near bars/taverns between 8 pm and 4 am, encompassing the late-night hours (from 10 pm to 2 am) when bars/taverns were prohibited from selling alcohol under SB571. While SB571 also restricted early-morning hours of sale, late-night crimes are more likely to be related to alcohol.11
We drew network (road-based) buffers around each bar/tavern and selected the buffer radius using a 3-step process.19 First, we summed the number of violent crimes that fell within nonoverlapping buffers around each bar/tavern in 400-ft increments. Second, we averaged the number of violent crimes within each buffer. Third, we identified the point at which mean violent crime counts began to decline.19 This point occurred in the 801- to 1200-ft buffer, so we used a 0- to 800-ft buffer radius.
The primary outcome used the Federal Bureau of Investigation’s definition of violent crime, which includes homicide, robbery, aggravated assault, and forcible rape.20 We also included common assault because SB571 aimed to reduce homicides and assaults.14 Our secondary outcomes were homicide and assault (common and aggravated).
SB571 Implementation
We created 3 time variables: (1) intervention period, a binary variable identifying whether a CB-month was before (May 2018 to June 2020) or after implementation (July 2020 to December 2022); (2) total time from baseline, a continuous variable that incremented the months (range, 1-56); and (3) time after the intervention, a continuous variable that identified whether a CB-month was exposed to SB571. For the treatment area, this variable was coded as 0 during the preintervention period and counted months after policy implementation (range, 1-30). For control areas, this variable equaled 0 throughout the entire study period.
Covariates
We adjusted for population, on- and off-premise alcohol outlet density (outlets per square mile), percentage of Black and White residents, neighborhood disadvantage, and the number of convenience stores (eFigure in Supplement 1). Because approximately 97% and 92% of residents in the treatment and control areas, respectively, reported either Black or White race, we considered only these racial groups in the analyses to ensure stable estimates in the models. American Community Survey 5-year estimates provided the CB group (CBG)–level population, racial composition, renter-occupied housing, adults aged 25 years or older with a college degree, female head of households, and poverty data. We used the 5-year estimates for 2018, 2019, 2020, and 2021, assigning the annual estimates to each CB-month that occurred within a CBG and year.21 We operationalized neighborhood disadvantage using the Ross and Mirowsky Index of Neighborhood Disadvantage.22 Alcohol outlet density and convenience stores were measured at the CBG level. Open Baltimore provided liquor license data,18 and the Center for a Livable Future provided convenience store count data.23
Statistical Analysis
Prior to analyses, the data underwent multiple preparation and linkage steps for ITS analyses. Details are provided in the eMethods in Supplement 1.
t Tests were used to compare mean violent crime rates and covariate values in the treatment area vs primary control areas during the first 12 months of the preintervention period to assess baseline comparability between the treatment and control areas. Violent crimes were log transformed before performing the t tests. For easier interpretation, we report nontransformed violent crime results.
The ITS was used to evaluate violent crime density before vs after SB571 in the treatment area vs control area. We used a sequence of observations taken at equal time intervals to establish an underlying trend and determined how this trend changed post intervention.24 We estimated (1) a level change in the month after SB571 implementation, estimated using the intervention period variable, and (2) a slope change, which used the time after the intervention variable to model whether changes in violent crime density after SB571 implementation persisted over time.
We performed the ITS using Poisson regression with overdispersion adjustment through a scale parameter and used log-transformed land area (square miles) as an offset.25 The modeling sequence followed 3 steps. Model 1 was an unadjusted base model. Model 2 was the main adjusted model controlling for sociodemographic covariates. Model 3 and sensitivity analysis 1 further controlled for CB-level fixed effects to adjust for unmeasured location-specific factors.26
We conducted sensitivity analyses for robustness checking. Sensitivity analysis 2 added an interaction between the intervention period and a treatment indicator for each CB. The main models did not include this interaction because it was not significant and worsened model fit. Still, this adjustment may affect parameter estimates. Sensitivity analysis 3 adjusted for CB-level population, a key covariate that may alter results. Sensitivity analysis 4 included a COVID-19 period indicator (0 before January 1, 2020; 1 on or after January 1, 2020) to examine whether pandemic-related changes in violent crime explained associations seen in the main models. Sensitivity analysis 5 included all 5 secondary control areas.
Sensitivity analyses 6 through 8 investigated whether SB571 displaced crime temporally and spatially. Sensitivity analysis 6 summed crime from 12 pm to 8 pm to determine whether SB571 was associated with a shift in violent crimes to earlier in the day. Sensitivity analyses 7 and 8 changed the buffer radius.
In post hoc analyses, level and slope changes from the main model were used to calculate the cost of crime saved from part I violent crimes in the first year of the bill’s implementation following methods described by McCollister et al.27 This method uses information about victim costs, criminal justice system costs, and crime career costs to calculate a tangible cost for each violent crime type. We updated these estimates to 2021 inflation-adjusted dollars using a Consumer Price Index of 1.259.
The data analyses were performed using SAS, version 9.4 (SAS Institute Inc) software. A 2-sided P < .05 was considered significant.
Results
Main Models
The treatment and control areas had similar baseline populations (mean [SD] number of residents across 67 CBs: treatment, 524.6 [234.6], including 92.16% [11.38%] Black residents and 4.69% [8.10%] White residents; control, 570.4 [217.4], including 77.96% [23.08%] Black residents and 13.96% [15.39%] White residents), violent crime counts within 800 ft of bars/taverns, neighborhood disadvantage, and convenience store availability (Table 1). The Figure shows that the aggregate trend for all late-night violent crimes per square mile in the treatment area from May 1, 2018, to June 30, 2020 (before SB571 went into effect) and control areas from May 2018 to December 2022 was similar, but after June 30, 2020 (when SB571 went into effect), the slope in the treatment area decreased relative to the control areas. The results of these analyses are summarized in Table 2 (sensitivity analysis 1). While there was no immediate change in all violent crime density in the treatment area after SB571 implementation (incidence rate ratio [IRR], 1.06; 95% CI, 0.89-1.26; P = .31), there was a significant decrease in slope in the treatment area compared with control areas over time (IRR, 0.76; 95% CI, 0.58-0.99; P = .04). Compared with control areas, violent crime density near bars/taverns in the treatment area fell by an additional 23% per year (annualized IRR, 0.77; 95% CI, 0.60-0.98; P = .04), which translates to an estimated 67 fewer violent crimes, equating to $18.2 million in tangible costs saved on violent crime in the first year after SB571. This slope change remained significant after adjusting for CB-level fixed effects (annualized IRR, 0.79; 95% CI, 0.64-0.97; P = .03).
Table 1. Comparison of Characteristics Across 67 Census Blocks During the Preintervention Period in Baltimore, by Intervention Status.
| Characteristic | Mean (SD) | P valuea | |
|---|---|---|---|
| Treatmentb | Controlc | ||
| Primary outcomed | |||
| All violent crimes | 23.13 (13.01) | 21.91 (10.43) | .42 |
| Covariates | |||
| Neighborhood disadvantagee | 0.04 (0.48) | 0.06 (0.23) | .23 |
| Alcohol outlet density | 60.62 (26.10) | 60.31 (37.94) | .97 |
| Black residents, % | 92.16 (11.38) | 77.96 (23.08) | .002 |
| White residents, % | 4.69 (8.10) | 13.96 (15.39) | .003 |
| Population | 524.6 (234.6) | 570.4 (217.4) | .42 |
| No. of convenience stores | 3.86 (3.32) | 3.24 (2.35) | .38 |
P values were obtained using log-transformed values due to nonnormality.
The treatment neighborhood included 26 census blocks.
The 2 control neighborhoods included 41 census blocks.
Values represent the mean number of annual crimes within 800 ft of bars/taverns in the treatment and control areas.
Measured using the Ross and Mirowsky Neighborhood Disadvantage Index,22 which combines indicators of poverty, female head of households, people aged 25 years or older with a college degree, and home ownership.
Figure. Interrupted Time Series of Monthly Late-Night Violent Crimes per Square Mile Before and After Implementation of Hours of Sale Alcohol Policy Among 67 Census Blocks in Baltimore, May 2018-December 2022.
Models adjusted for census block group–level covariates (ie, neighborhood disadvantage, alcohol outlet density, percentage of Black residents, percentage of White residents, and number of convenience stores) as well as seasonality. Violent crimes were counted in 0- to 800-ft buffers around the bars/taverns and from 8 pm to 4 am. Vertical dashed line indicates the policy effective date of July 2020.
Table 2. Differences in Violent Crimea Rates After Implementation of Hours of Sales Alcohol Policy, Comparing Census Blocks in Treatment and Control Areas in Baltimore, May 2018 to December 2022.
| Time estimate | Base modelb | Main modelc | Sensitivity analysis 1: adding CB fixed effects | |||
|---|---|---|---|---|---|---|
| IRR (95% CI)d | P value | IRR (95% CI) | P value | IRR (95% CI) | P value | |
| Level change | 1.06 (0.89-1.26) | .31 | 1.02 (0.86-1.20) | .84 | 0.97 (0.85-1.11) | .67 |
| Slope change | 0.76 (0.58-0.99) | .04 | 0.77 (0.60-0.98) | .04 | 0.79 (0.64-0.97) | .03 |
Abbreviations: CB, census block; IRR, incidence rate ratio.
Violent crimes were counted in 0- to 800-ft buffers around the bars/taverns and from 8 pm to 4 am.
The base model is a crude model of the level and slope change without any adjustment variables. For the base model, the baseline levels (P = .71) and slopes (P = .91) between intervention and control sites were similar. Therefore, we removed these 2 terms in the models to yield a more parsimonious model.
The main model is adjusted for CB group–level covariates (ie, neighborhood disadvantage, alcohol outlet density, percentage of Black residents, percentage of White residents, and number of convenience stores), as well as seasonality. We did not adjust for total population in these models because we used land area as an offset, and land area correlates with population.
Estimates shown are annualized, representing the percent reduction in violent crimes per year.
The secondary outcomes of homicide and assault showed similar trends in decline (eTable 2 in Supplement 1). In adjusted models, homicide density dropped by 51% in the first month after SB571 implementation (IRR, 0.49; 95% CI, 0.33-0.72) and by an additional 40% per year after the first month (IRR, 0.60; 95% CI, 0.43-0.82). While there was no significant level change in assault density after SB571 implementation, assault density within 800 ft of treatment area bars/taverns dropped by 23% per year (IRR, 0.77; 95% CI, 0.60-0.99). After additionally adjusting for CB fixed effects, the slope change for assault density persisted, with an annual decrease of 22% (IRR, 0.78; 95% CI, 0.64-0.96), showing a faster decline in crimes around bars/taverns in the treatment area. Model 3 for homicide density did not converge, possibly because these crimes are rare. When looking at common and aggravated assaults separately, a similar magnitude of associations with respect to slope decrease was detected (eTable 3 in Supplement 1).
Sensitivity Analyses
Sensitivity analysis 2 included the treatment indicator and time-by-treatment interactions. Sensitivity analyses 3 and 4 adjusted for population and the COVID-19 period, respectively. The associations from the main models persisted. While there were no significant level changes in these models, all violent crime density declined in the treatment area by 24% (IRR, 0.76; 95% CI, 0.59-0.97) to 27% (IRR, 0.73; 95% CI, 0.59-0.91) per year (Table 3; eTable 4 in Supplement 1). Additional sensitivity analyses 5 to 8 are provided in eTables 5 and 6 in Supplement 1.
Table 3. Sensitivity Analyses Using All Violent Crimes as the Primary Outcome.
| Time estimate | IRR (95% CI) | ||
|---|---|---|---|
| Sensitivity analysis 2a,b | Sensitivity analysis 3c | Sensitivity analysis 4d | |
| Level change | 1.02 (0.86-1.20) | 0.99 (0.83-1.18) | 1.07 (0.89-1.28) |
| Slope change | 0.73 (0.59-0.91)e | 0.74 (0.59-0.94)a | 0.76 (0.59-0.97)a |
Abbreviation: IRR, incidence rate ratio.
P < .05.
Adjusted for treatment vs control indicator and time-treatment interaction.
Adjusted for population.
Adjusted for a COVID-19 period indicator.
P < .01.
Discussion
The study findings revealed that implementing SB571, which reduced hours of alcohol sales in bars/taverns in 1 Baltimore neighborhood, was associated with a substantial reduction in violent crimes over time. Sensitivity analyses supported these results, suggesting that the policy did not shift crime earlier in the day or further away from the bars/taverns. During this same period, the violent crime rate remained constant across the City of Baltimore, falling a small relative 1.5% (from 20.3 to 20.0 violent crimes per 1000 residents).28
Our findings on reduced temporal alcohol availability align with prior studies evaluating the association between reduced physical alcohol availability and violent crimes.29,30 A community-based longitudinal study in Atlanta, Georgia, found that a modest reduction in alcohol outlet density substantially reduced violent crime exposure.5 Furthermore, alcohol outlet hours of sale modified the association between alcohol outlet density and violence. A panel study of entertainment zones in Queensland, Australia, that found higher assault rates in areas with greater concentrations of high-risk alcohol outlets (eg, bars, pubs) also found that this association was stronger when those outlets closed later at night.31 Unlike previous studies, our investigation addressed a crucial gap highlighted by the Task Force on Community Preventive Services by providing a US-based evaluation of a change of more than 2 hours of sale.32 We used a rigorous design to examine how a decrease in hours of sales may be linked to a decline in violent crimes. The inclusion of comparable control areas and further community-level covariate adjustments and CB-level fixed effects enhanced the validity of these findings.
While the slope decreases of all violent crimes were consistently detected in the main model and several sensitivity analyses, we also detected a statistically significant level decrease in several models concerning homicide, which contributed to a small fraction of the overall violent crimes. Senate Bill 571 was enacted to address violent crime in 1 high-priority neighborhood33 by reducing the hours of operation from 6 am to 2 AM to 9 am to 10 pm for bars/taverns. It is possible that establishment owners recognized that selling alcohol after 10 pm was believed to be the source of violent crimes. In compliance with the policy effective in July 2020, establishment owners may have taken other voluntary measures on safety monitoring and surveillance and/or increased their level of collaboration with local law enforcement agencies, which may explain the immediate level decrease in homicide, the most severe form of violent crime. Due to small counts of homicide, a decrease in the number of homicide incidents may also have resulted in a large percent change. We attempted to empirically test whether changes in police patrolling in the treatment neighborhood explained the lower violent crime rate using arrest data for alcohol-related nonviolent offenses. However, there were high levels of missingness for arrest locations, and arrests for alcohol-related nonviolent offenses were rare even before excluding missing data. Thus, we are unable to rule out the possibility that our findings, at least in part, are due to changes in police presence in the treatment neighborhood.
Some violent offenders, particularly those committing robbery, tend to commit crimes on the boundaries of 2 neighborhoods.34 In addition, it is possible that people in the treatment area may have chosen to go to bars/taverns in adjacent neighborhoods, even though bars (on-premise only) in the treatment neighborhood remained open until 2 am. Future research should investigate whether SB571 was associated with violent crime density in adjacent neighborhoods.
Limitations
This study has some limitations. First, it evaluated a policy change in 1 Baltimore neighborhood, making generalization to other communities challenging. Second, the 18-month postenactment period limited the assessment of the policy’s longer-term impact. Third, despite comparable control areas and CBG-level covariates, residual confounding may have influenced associations. Namely, the policy was implemented during the COVID-19 pandemic. It is possible that the treatment and control neighborhoods were differentially affected by pandemic-era lockdowns due to the number and types of businesses within each neighborhood. While a sensitivity analysis showed that the association of SB571 with violent crime reductions persisted after adjusting for a COVID-19 period indicator, this covariate may not have completely accounted for between-neighborhood differences in responding to the pandemic. Fourth, small numbers of homicides resulted in model nonconvergence when controlling for CB-level fixed effects; however, estimates from other models remained stable and consistent. Finally, the study outcomes were late-night violent crimes, as they are more likely to be affected by the policy and related to alcohol.11 However, there was no explicit information about whether these incidents were alcohol related.
Conclusions
This study is, to our knowledge, the first US evaluation of a natural experiment suggesting that alcohol policies that reduce hours of sale could substantially reduce violent crimes. While regulating alcohol outlet density is more commonly recommended to reduce the availability of alcohol,35 our findings suggest that reducing late-night hours of sale may be an effective tool in addressing problem drinking and associated assault and other crimes around these outlets. Future studies in varied urban contexts and with longer observation periods are needed.
Public health and medicine often share complementary goals. Emergency physicians can bring their frontline experience and research to help prevent injury through a population-based public health approach.36 Other areas in medicine, including preventive medicine, can also play a role in improving population health by broadening a traditionally individual-oriented clinical approach to advocate for effective public health policies that address upstream sociocontextual factors that often manifest themselves in clinical encounters.37 Specifically, this analysis investigated whether community-level alcohol policy changes may also hold promise for reducing the burden of violence-related injuries placed on the health care system. Our study has important policy implications for guiding local and state legislators, establishment owners, law enforcement personnel, physicians, and community members in addressing violent crime, especially in low-income and historically marginalized neighborhoods where alcohol outlets often cluster.
eTable 1. Rankings of the Baltimore Community Statistical Areas as Potential Control Areas Based on Alcohol Outlet Rate, Violent Crime Rate, and Neighborhood Disadvantage
eMethods. Data Preparation and Linkage
eFigure. Conceptual Framework Summarizing Variables, Level of Measurement, and Key Design Features of Main Models
eTable 2. Differences in Level and Slope Changes of Homicide and Assault Rates After Implementation of Hours of Sales Alcohol Policy Comparing Census Blocks in the Treatment Area to Controls in Baltimore, May 2018 to December 2022
eTable 3. Changes in Level and Slope of disaggregated Assault Rates After Implementation of Hours of Sales Alcohol Policy Among 67 Census Blocks in Baltimore, May 2018 to December 2022
eTable 4. Sensitivity Analyses Using Homicide and Assault as Secondary Outcomes
eTable 5. Sensitivity Analyses Using All Violent Crimes as the Primary Outcome
eTable 6. Sensitivity Analyses for Buffer Radius Using All Crime as the Primary Outcome
Data Sharing Statement
References
- 1.Healthcare Cost and Utilization Project (HCUPnet). Agency for Healthcare Research and Quality. 2020. Accessed December 17, 2023. https://datatools.ahrq.gov/hcupnet/ [PubMed]
- 2.Popova S, Giesbrecht N, Bekmuradov D, Patra J. Hours and days of sale and density of alcohol outlets: impacts on alcohol consumption and damage: a systematic review. Alcohol Alcohol. 2009;44(5):500-516. doi: 10.1093/alcalc/agp054 [DOI] [PubMed] [Google Scholar]
- 3.Campbell CA, Hahn RA, Elder R, et al. ; Task Force on Community Preventive Services . The effectiveness of limiting alcohol outlet density as a means of reducing excessive alcohol consumption and alcohol-related harms. Am J Prev Med. 2009;37(6):556-569. doi: 10.1016/j.amepre.2009.09.028 [DOI] [PubMed] [Google Scholar]
- 4.Sherk A, Stockwell T, Chikritzhs T, et al. Alcohol consumption and the physical availability of take-away alcohol: systematic reviews and meta-analyses of the days and hours of sale and outlet density. J Stud Alcohol Drugs. 2018;79(1):58-67. doi: 10.15288/jsad.2018.79.58 [DOI] [PubMed] [Google Scholar]
- 5.Zhang X, Hatcher B, Clarkson L, et al. Changes in density of on-premises alcohol outlets and impact on violent crime, Atlanta, Georgia, 1997-2007. Prev Chronic Dis. 2015;12:E84. doi: 10.5888/pcd12.140317 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Yu Q, Scribner R, Carlin B, et al. Multilevel spatio-temporal dual changepoint models for relating alcohol outlet destruction and changes in neighbourhood rates of assaultive violence. Geospat Health. 2008;2(2):161-172. doi: 10.4081/gh.2008.240 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Fliss MD, Cox ME, Wallace JW, Simon MC, Knuth KB, Proescholdbell S. Measuring and mapping alcohol outlet environment density, clusters, and racial and ethnic disparities in Durham, North Carolina, 2017. Prev Chronic Dis. 2021;18:E89. doi: 10.5888/pcd18.210127 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Trangenstein P, Gray C, Rossheim M, Sadler R, Jernigan D. Alcohol outlet clusters and population disparities. J Urban Health. 2020;97(1):123-136. doi: 10.1007/s11524-019-00372-2 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Hahn RA, Kuzara JL, Elder R, et al. ; Task Force on Community Preventive Services . Effectiveness of policies restricting hours of alcohol sales in preventing excessive alcohol consumption and related harms. Am J Prev Med. 2010;39(6):590-604. doi: 10.1016/j.amepre.2010.09.016 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Rammohan V, Hahn RA, Elder R, et al. ; Task Force on Community Preventive Services . Effects of dram shop liability and enhanced overservice law enforcement initiatives on excessive alcohol consumption and related harms: two community guide systematic reviews. Am J Prev Med. 2011;41(3):334-343. doi: 10.1016/j.amepre.2011.06.027 [DOI] [PubMed] [Google Scholar]
- 11.Mäkelä P, Warpenius K. Night-time is the right time? late-night drinking and assaults in Finnish public and private settings. Drug Alcohol Rev. 2020;39(4):321-329. doi: 10.1111/dar.13068 [DOI] [PubMed] [Google Scholar]
- 12.Duailibi S, Ponicki W, Grube J, Pinsky I, Laranjeira R, Raw M. The effect of restricting opening hours on alcohol-related violence. Am J Public Health. 2007;97(12):2276-2280. doi: 10.2105/AJPH.2006.092684 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Ejrnæs A, Scherg RH. Nightlife activity and crime: the impact of COVID-19 related nightlife restrictions on violent crime. J Crim Justice. 2022;79:101884. doi: 10.1016/j.jcrimjus.2022.101884 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Baltimore City–Alcohol Beverages–Licenses–Hours of Sale, S 571, Reg Sess (2020). Accessed July 20, 2023. https://mgaleg.maryland.gov/2020RS/bills/sb/sb0571t.pdf
- 15.Neighborhood health profiles–frequently asked questions. Baltimore City Health Department . Updated May 27, 2014. Accessed December 20, 2023. https://health.baltimorecity.gov/node/231
- 16.Baltimore Neighborhood Indicators Alliance . Accessed December 20, 2023. https://bniajfi.org/
- 17.Bureau of the Census; Economics and Statistics Administration; US Department of Commerce . Census blocks and block groups. Geographic Areas Reference Manual. Bureau of the Census; 1994:11-1-11-19. [Google Scholar]
- 18.Open Baltimore. City of Baltimore. Accessed December 17, 2023. https://data.baltimorecity.gov/
- 19.Groff E. Exploring ‘near’: characterizing the spatial extent of drinking place influence on crime. Aust N Z J Criminol. 2011;44(2):156-179. doi: 10.1177/0004865811405253 [DOI] [Google Scholar]
- 20.Violent crime. Federal Bureau of Investigation . 2018. Accessed August 20, 2023. https://ucr.fbi.gov/crime-in-the-u.s/2018/crime-in-the-u.s.-2018/topic-pages/violent-crime
- 21.American Community Survey 2016-2020 5-year data release. US Census Bureau. 2022. Accessed June 12, 2023. https://www.census.gov/newsroom/press-kits/2021/acs-5-year.html
- 22.Ross CE, Mirowsky J. Neighborhood disadvantage, disorder, and health. J Health Soc Behav. 2001;42(3):258-276. doi: 10.2307/3090214 [DOI] [PubMed] [Google Scholar]
- 23.Center for a Livable Future. Johns Hopkins University. Accessed January 15, 2023. https://clf.jhsph.edu/node/1
- 24.Bernal JL, Cummins S, Gasparrini A. Interrupted time series regression for the evaluation of public health interventions: a tutorial. Int J Epidemiol. 2017;46(1):348-355. doi: 10.1093/ije/dyw098 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Xuan Z, Yan S, Formica SW, et al. Association of implementation of postoverdose outreach programs with subsequent opioid overdose deaths among Massachusetts municipalities. JAMA Psychiatry. 2023;80(5):468-477. doi: 10.1001/jamapsychiatry.2023.0109 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Hallberg K, Williams R, Swanlund A, Eno J. Short comparative interrupted time series using aggregate school-level data in education research. Educ Res. 2018;47(5)295-306. doi: 10.3102/0013189X18769302 [DOI] [Google Scholar]
- 27.McCollister KE, French MT, Fang H. The cost of crime to society: new crime-specific estimates for policy and program evaluation. Drug Alcohol Depend. 2010;108(1-2):98-109. doi: 10.1016/j.drugalcdep.2009.12.002 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Baltimore City Public Safety Accountability Dashboard. City of Baltimore. Accessed December 18, 2023. https://monse.baltimorecity.gov/baltimore-public-safety-accountability-dashboard
- 29.Gruenewald PJ, Remer L. Changes in outlet densities affect violence rates. Alcohol Clin Exp Res. 2006;30(7):1184-1193. doi: 10.1111/j.1530-0277.2006.00141.x [DOI] [PubMed] [Google Scholar]
- 30.Xu Y, Yu Q, Scribner R, Theall K, Scribner S, Simonsen N. Multilevel spatiotemporal change-point models for evaluating the effect of an alcohol outlet control policy on changes in neighborhood assaultive violence rates. Spat Spatiotemporal Epidemiol. 2012;3(2):121-128. doi: 10.1016/j.sste.2012.04.005 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Taylor N, Livingston M, Coomber K, et al. The combined impact of higher-risk on-license venue outlet density and trading hours on serious assaults in night-time entertainment precincts. Drug Alcohol Depend. 2021;223:108720. doi: 10.1016/j.drugalcdep.2021.108720 [DOI] [PubMed] [Google Scholar]
- 32.Task Force on Community Preventive Services . Recommendations for reducing excessive alcohol consumption and alcohol-related harms by limiting alcohol outlet density. Am J Prev Med. 2009;37(6):570-571. doi: 10.1016/j.amepre.2009.09.021 [DOI] [PubMed] [Google Scholar]
- 33.McCray C. Commentary: want to reduce crime? challenge the status quo. AFRO News. November 3, 2020. Accessed September 4, 2022. https://afro.com/commentary-want-to-reduce-crime-challenge-the-status-quo/
- 34.Brantingham PL, Brantingham PJ. Nodes, Paths and Edges: Considerations on the Complexity of Crime and the Physical Environment. In: Walker JT, ed. Social, Ecological and Environmental Theories of Crime. Routledge; 2017:363-388. [Google Scholar]
- 35.Jernigan DH, Sparks M, Yang E, Schwartz R. Using public health and community partnerships to reduce density of alcohol outlets. Prev Chronic Dis. 2013;10:E53. doi: 10.5888/pcd10.120090 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36.Anderson E, Bernstein E, Xuan Z, Alter HJ. Inventing social emergency medicine: summary of common and critical research themes using a modified Haddon matrix. Ann Emerg Med. 2019;74(5S):S74-S77. doi: 10.1016/j.annemergmed.2019.08.464 [DOI] [PubMed] [Google Scholar]
- 37.Hunter DJ. The complementarity of public health and medicine - achieving “the highest attainable standard of health”. N Engl J Med. 2021;385(6):481-484. doi: 10.1056/NEJMp2102550 [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
eTable 1. Rankings of the Baltimore Community Statistical Areas as Potential Control Areas Based on Alcohol Outlet Rate, Violent Crime Rate, and Neighborhood Disadvantage
eMethods. Data Preparation and Linkage
eFigure. Conceptual Framework Summarizing Variables, Level of Measurement, and Key Design Features of Main Models
eTable 2. Differences in Level and Slope Changes of Homicide and Assault Rates After Implementation of Hours of Sales Alcohol Policy Comparing Census Blocks in the Treatment Area to Controls in Baltimore, May 2018 to December 2022
eTable 3. Changes in Level and Slope of disaggregated Assault Rates After Implementation of Hours of Sales Alcohol Policy Among 67 Census Blocks in Baltimore, May 2018 to December 2022
eTable 4. Sensitivity Analyses Using Homicide and Assault as Secondary Outcomes
eTable 5. Sensitivity Analyses Using All Violent Crimes as the Primary Outcome
eTable 6. Sensitivity Analyses for Buffer Radius Using All Crime as the Primary Outcome
Data Sharing Statement
