Race/ethnicity, community of residence, and DUI arrest after beginning treatment for an alcohol use disorder

Abstract

The purpose of this study was to examine whether racial/ethnic disparities in post-treatment arrests for driving under the influence (DUI) exist among clients receiving outpatient treatment for an alcohol use disorder (AUD), and to assess whether community characteristics were associated with this outcome. The sample included adults with an AUD entering publicly funded outpatient treatment in Washington State in 2012. Treatment data were linked with criminal justice and US Census data. Multilevel time-to-event analysis was employed to answer the research questions. Key independent variables included client race/ethnicity, community-level economic disadvantage, and racial/ethnic composition of the community. Latino clients and clients residing in communities with a higher proportion of Black residents had higher hazards of a DUI arrest post-treatment admission. Future research should examine whether disparities in DUI arrests are related to differences in treatment effectiveness or other factors (e.g., inequities in law enforcement) so that these disparities can be addressed.

Introduction

Alcohol-related crimes result in an estimated $73 billion in societal costs in the U.S., and nearly half of this is due to intoxicated driving-related vehicle accidents.¹ While concern over intoxicated driving due to marijuana use is growing as more states move to legalize recreational use, alcohol remains the most commonly used substance of intoxicated drivers.² Approximately one third of all motor vehicle fatalities are caused by an alcohol-impaired driver, and this proportion has remained unchanged since 2010.³ Ongoing efforts are necessary to further reduce the impact that alcohol impaired driving has on individuals and communities.

Many individuals who are arrested for a driving under the influence (DUI) charge enter the substance use treatment system. Several states require attendance at outpatient treatment to address an underlying alcohol problem after an arrest for a DUI charge. Nearly half of all referrals to outpatient treatment settings come from the criminal justice system, and in 2015, 11% of all criminal justice referrals to substance use services were for a DUI charge.⁴ Of those who were referred for a DUI charge, nearly one third reported alcohol as their primary substance.⁴ Therefore, reducing criminal justice involvement (e.g., arrests, convictions, incarceration) is frequently used as a treatment outcome to determine treatment effectiveness.

Most research examining DUI as a treatment outcome has focused specifically on populations who have already received a DUI charge. A review of court-mandated treatment interventions for individuals who had a DUI charge found that, overall, treatment is an effective tool for reducing future DUI charges.⁵ More recently, a systematic review found modest evidence of effectiveness for reducing future DUI arrest for interventions developed specifically for DUI-offenders.⁶ One study that examined a general sample of treatment participants found that treatment was associated with a reduction in rates of motor vehicle accidents and fatalities,⁷ which could indicate fewer incidences of driving under the influence. More research is needed that examines DUI as a treatment outcome for individuals receiving treatment for an alcohol use disorder.

Considering the well-documented disparity in access to treatment for racial/ethnic minority groups,^8,9 it is all the more essential that outcomes are equitable across all groups when treatment is utilized. However, studies suggest that this is not the case. For example, racial/ethnic disparities have been observed in rates of treatment completion. One study found that Black, Latino, and American Indian clients across treatment setting types were less likely than Whites to complete treatment.¹⁰ Disparities have also been observed in post-treatment outcomes. Another study found that post-treatment wages for individuals utilizing publicly funded outpatient treatment were significantly lower among Black and American Indian clients compared with White clients.¹¹ Post-treatment arrests and convictions for any charge have been shown to be significantly higher among Black clients compared with White clients utilizing publicly funded outpatient treatment.^12–14 To-date, no research has been conducted that examines whether racial/ethnic disparities exist using DUI arrest as the treatment outcome.

Much of the existing research on racial/ethnic disparities in DUI arrests focuses on individual self-report in the general population, not among those who are in treatment. According to 2016 unadjusted national estimates, 10% of Whites reported driving under the influence in the past year, while only 6% of Blacks, and 5% each for American Indians and Latinos reported doing so.¹⁵ Unadjusted national crime statistics from 2016 indicate that among all individuals arrested for a DUI, 82% were White, 23% were Latino, 14% were Black, and 2% were American Indian.¹⁶ Research that adjusted for individual characteristics found that American Indians in the general population were at greater risk of driving under the influence compared with other groups.¹⁷ Evidence for Latinos is mixed, with some finding that Latino men were less likely to report a DUI arrest compared with White and American Indian men¹⁸ while others found that Latino men were more likely to report a DUI arrest in their lifetime compared with White men.¹⁹ There is also some evidence that Latinos perceive the number of drinks that cause impairment to be greater than Whites, indicating they may be less likely to self-report driving under the influence, and may be more likely to drive while impaired compared with Whites.²⁰

In addition to individual characteristics, previous research suggests that the community in which an individual resides can also have important implications for alcohol use and treatment outcomes. One study found that individuals residing in neighborhoods characterized by economic disadvantage (e.g., high proportion of residents who did not graduate high school, were unemployed, had incomes that were below the poverty line or median income) or a higher proportion of Black residents had more negative consequences (e.g., fights with friends or family, problems with the criminal justice system or finding employment) related to drinking.²¹ Another study of individuals utilizing publicly funded outpatient treatment services found that clients residing in neighborhoods characterized by a high proportion of Black residents were less likely to meet the criteria of treatment engagement, a quality indicator of substance use treatment.²² In some cases community-level characteristics may be protective, as evidenced by one study that found that rates of past-year alcohol use disorders were lower for Latinos who lived in neighborhoods with a higher proportion of Latino residents.²³ Recent research has also found an association between community-level characteristics and treatment outcomes. Specifically, one prior study found that both community economic disadvantage and proportion of Black residents were associated with worse post-treatment employment outcomes.¹¹ More research that includes community-level factors is needed to better understand what might be driving disparities in treatment outcomes.

Given its importance as a treatment outcome, more research is also needed to better understand whether racial/ethnic disparities in post-treatment DUI arrest exist. Furthermore, accounting for community-level factors will provide a more complete understanding of what is associated with this treatment outcome. The present study seeks to address these gaps by using a large dataset that links administrative treatment and criminal justice data with data from the 2010 US census to examine DUI arrest outcomes for adults receiving outpatient treatment for an alcohol use disorder in Washington state. This study aims to answer two research questions: 1) Are there racial/ethnic differences in likelihood of post-treatment DUI arrest? and 2) Are community characteristics associated with a post-treatment DUI arrest?

Methods

Data sources

This study used client treatment, arrest, and incarceration data from the state of Washington, linked with data from the US Census Bureau. Washington state’s Behavioral Health Administration (BHA) maintained the Treatment Activity Report Generation Tool (TARGET), which collects, among other information, treatment for all individuals attending publicly funded substance use treatment services. Client data were collected by providers at admission and entered into the TARGET system. Incarceration and arrest data were obtained by BHA from the Department of Corrections and the Washington State Patrol. Data in each system were matched by BHA using a combined deterministic and probabilistic linkage technique^24–26 based on the client’s name, Social Security number, date of birth, and sex. For more detailed information on the linkage rules, please see Appendix D of the Link King manual.²⁷ These data were de-identified, and client addresses were converted to census tracts to enable linkage to data from the US Census and American Community Survey (2009–2013 5-year averages) by the research team.

Sample

The sample for this study included adults beginning a new outpatient treatment episode at a publicly funded outpatient substance use treatment setting (N= 12,506) in Washington State in 2012. Clients were excluded if they did not report their race/ethnicity as either White, Black, American Indian, or Latino (N= 1,011, 8%), or if they did not report alcohol as their primary, secondary, or tertiary substance at treatment admission (N= 3,189, 28%). Clients were also excluded if they had died at any point during the year after admission (N= 34, <1%), or had missing data on any of the independent variables (N= 111, 1%). Finally, clients were excluded if they were incarcerated post-index without an arrest (neither at pre- nor post- index; N= 25, <1%). This might occur if a client was charged more than a year before their index date or had any previous involvement in the criminal justice system prior to their index date. The final analytic sample consisted of 8,136 clients.

Dependent variable

The dependent variable for all analyses was time (in days) to an arrest for driving under the influence after admission to outpatient treatment.

Independent variables

Race/ethnicity:

Clients were asked whether or not they were Latino and were given 16 options separately to select the race or other ethnicity they may identify with, having the option to select all that applied. Clients who identified as Latino were categorized as Latino regardless of their race response. We restricted our analyses to clients who self-reported as being members of the four largest racial/ethnic groups in Washington state: non-Latino White, non-Latino American Indian, Latino, and non-Latino Black. Additional racial groups (e.g., Asian, Hawaiian, Pacific Islander, Multiracial) were excluded due to their small sample size which made it difficult to analyze them separately.

Community characteristics:

We defined communities at the census tract level as this has been recommended as the preferred unit for examining neighborhood-level effects on substance use.²⁸ Clients in our sample resided in a total of 1,269 census tracts representing 87% of all census tracts in Washington State. The first set of community characteristics we were interested in were indicators of community racial/ethnic composition (e.g., % American Indian residents, % Latino residents, % Black residents). The second set of community characteristics we were interested in were census tract percentages of the following: residents who were unemployed, living below the poverty level, those in management/professional occupations (recoded; for employed civilian population, ages 16+), female-headed households with children, and households with annual income > $75,000 (recoded). Prior studies found these to be associated with substance use, treatment continuity, or treatment completion, and thus might affect the outcomes of substance use treatment.^23,29–31 We conducted a factor analysis to reduce the number of variables in the second set of variables, and to address any highly correlated explanatory variables. The Kaiser criterion assists in identifying meaningful factors and indicates that factors should be retained only if they have an Eigenvalue of 1 or more.³² This led us to retain one factor with an Eigenvalue greater than 1 which was standardized and weighted, and loaded most heavily on income and occupation variables and more modestly on education and female head of household variables. We deemed this factor “community economic disadvantage” and included this in our analysis along with each of the measures of community racial/ethnic composition.

Furthermore, community variables were dichotomized by top quartile and bottom three quartiles based on all residents in each census tract (i.e., those in our sample as well as the general community residing in each census tract). The top quartile for the proportion of American Indians living in any given census tract (i.e., “Higher proportion of American Indians”) had a proportion of American Indians of 3.2% or more. Likewise, the proportion of Latino residents in the top quartile (i.e., “Higher proportion of Latino residents”) was 11.3% or more, and the top quartile for the proportion of Black residents (i.e., “Higher proportion of Black residents”) was 3.8% or more. For the economic disadvantage factor, the top quartile of census tracts deemed “Higher economic disadvantage” values ranged between .64 and 3.64.

Prior year arrest/incarceration:

Using arrest and incarceration data, we created a variable to indicate whether clients had been arrested or incarcerated (yes/no) during the year prior to the admission.

Covariates:

Client-level covariates were chosen based on prior findings of their association with arrest outcomes.^13,14 These included demographic characteristics (gender, age, marital status), socioeconomic characteristics at admission (education, homeless status, employment status), primary, secondary, and tertiary substances (alcohol only, alcohol and marijuana only, alcohol and other drugs), and an indicator of whether the client had been referred to treatment by the criminal justice system. In addition, we included a community-level covariate indicating the percent of households in a census tract that did not own a car. We did not have access to individual-level information on car-ownership, so this variable was used instead to adjust for access to a motor vehicle. Access to a motor vehicle would clearly associate with risk for a DUI arrest and therefore was important to include in our models.

Analyses

Descriptive analyses were performed for the overall sample characteristics and by client’s race/ethnicity. Differences by race/ethnicity were tested using Chi-Squares for categorical variables and ANOVA for continuous variables, followed with pairwise comparisons using a Bonferroni correction for multiple comparisons. Bivariate tests were conducted to compare the outcome variable by racial/ethnic groups using a Bonferroni correction for multiple comparisons. We also compared the outcome variable by whether a client did or did not have an arrest or period of incarceration in the year prior to treatment. We then used hierarchical time-to-event analyses with clustering of clients within communities.³³ The outcome variable in these models was time to DUI arrest, measured in days, and our key independent variables were race/ethnicity indicators and community characteristics. Models were conducted for the full sample, and then separately based on client’s prior year criminal justice involvement.

Clients’ times-to-event outcome values were censored at 365 days after the treatment admission if no arrest occurred before that time. When clients were in a controlled environment (i.e., incarcerated or in residential substance use treatment; N=47) during follow-up, their times-to-event were temporarily curtailed given that they would be significantly less likely to experience the outcome while in these settings. In these cases, the follow-up period was extended by the same number of days so that the maximum times-to-event for clients remained 365 days. As is the custom in times-to-event analyses, all observations were marked as ‘censored’ if they reached the maximum observation value without experiencing the outcome of interest. In the few cases when a client’s 365-day follow-up period went beyond December 31, 2013, the last day for which data are available, the time-to-event was the number of days between admission and December 31, 2013 excluding any days where the client was in a controlled setting.

Sensitivity Analysis

Previous studies have indicated that prior criminal justice involvement is a strong predictor of a future DUI arrest for the general population,³⁴ and future arrest for any charge among clients in treatment.¹⁴ Therefore, we also conducted sensitivity analyses using stratification to separate our sample into two groups: those who had a prior arrest or period of incarceration in the year prior to treatment admission, and those who did not. We ran the models this second time including all of the same variables as in the first model, except the dichotomous variable for whether there was an arrest or incarceration in the prior year.

Human subjects protection

This study was approved by the X and the X Institutional Review Boards.

Results

Client characteristics

Table 1 includes descriptions of the client characteristics at admission for the overall analytic sample, as well as results from bivariate analyses examining differences by client race/ethnicity. Most clients were male (64%), White (65%), and unemployed (80%). Nearly a quarter of clients (24%) reported their primary substances at admission as alcohol only, 23% as alcohol and marijuana only, and 53% reported alcohol and other drugs. About half (51%) of clients had been arrested or incarcerated in the year prior to treatment admission, and 11% had a prior DUI arrest. Client characteristics differed by race/ethnicity. For example, significantly more American Indian clients were female compared with the other racial/ethnic groups. Black clients tended to be older, while Latino clients were younger at admission. American Indian and Latino clients were more likely to be referred to treatment by the criminal justice system; however, Black clients were significantly more likely to have had an arrest or incarceration in the year prior to treatment admission. Black clients were significantly less likely to have had an arrest for a DUI in the year prior to treatment compared with their White, American Indian, and Latino counterparts. Latino clients were significantly less likely to have had a high school diploma, and Latino and American Indian clients were more likely to have reported alcohol only as their primary substance at admission compared with White and Black clients who were more likely to have reported alcohol and other drugs.

Table 1.

Client Characteristics at Treatment Admission

	Entire Sample (N =8,136)	White (W) (N= 5,279)	American Indian (AI) (N= 1,219)	Latino (L) (N= 912)	Black (B) (N= 726)	Differences between groups (p< 0.05)3
Client demographics	%
Female	35.7	37.4	42.8	24.5	25.5	AI > W> B, L
Age
18–20	6.6	5.1	7.3	16.8	3.7	L > W, AI, B
21–25	15.0	14.4	16.6	19.2	11.7	L > W, B AI > B
26–30	15.7	16.3	15.1	16.5	12.0	W > B
31–44	34.6	34.6	36.7	33.8	31.8	n.s.
45–54	21.2	22.2	19.2	9.8	31.5	B > W, AI > L
55+	6.9	7.4	5.2	4.1	9.2	W, B > AI, L
Married	22.7	20.3	31.3	30.6	16.2	AI, L >W > B
Arrest or incarceration in Prior   Year	50.6	49.3	49.6	54.0	57.3	B > W, AI
DUI arrest in Prior Year	10.9	10.9	11.0	13.3	7.6	W, AI, L > B
Socioeconomic status
Education
Less than H.S.	28.4	23.9	32.1	50.0	28.5	L > B, AI > W
H.S. Grad	56.6	58.8	58.7	39.7	58.7	W, AI, B > L
More than H.S.	8.3	9.7	5.9	4.9	5.9	W > B, L, AI
Vocational training	6.7	7.6	3.4	5.4	6.9	W, B > AI
Homeless/at risk of	16.8	17.8	9.8	10.6	28.4	B > W > L, AI
Unemployed	80.1	83.2	74.2	64.4	87.5	B > W > AI > L
Treatment referral and Substance use
Referral Source
Criminal Justice	60.1	56.4	72.1	68.2	57.2	AI, L > W, B
Substance use at admission1
Alcohol only	24.1	22.1	30.4	33.1	16.9	L, AI > W > B
Alcohol and Marijuana	23.3	20.9	26.7	29.5	27.7	B, L, AI > W
Alcohol and Other Drugs	52.6	57.0	43.0	37.4	55.4	W, B > L, AI
Age of first use2
≤ 10	14.5	15.2	13.5	10.8	15.6	B, W > L
11–14	38.4	39.6	39.9	32.9	34.3	W > B, L AI > L
15–17	32.6	32.5	31.0	36.0	31.8	n.s.
18–20	9.9	8.9	11.0	13.4	11.2	L > W
21+	4.6	3.9	4.7	7.0	7.2	B, L > W
Community Characteristics
Higher Economic Disadvantage	40.7	37.1	49.1	48.9	42.2	L, AI > W, B
Higher % American Indian	45.6	40.6	79.9	34.3	38.7	AI > W > L AI > B
Higher % Latino	35.6	29.6	44.7	60.3	32.5	L > AI > W, B
Higher % Black	30.0	26.7	18.9	25.7	77.6	B > W, L > AI
% households without a car – (Mean, SD)	10.1% (12.2%)	9.8% (11.6%)	8.2% (10.5%)	9.7% (11.6%)	17.9% (18.4%)	B > W, AI, L W > AI

Note:

¹Clients were asked what their primary, secondary, and tertiary substances were at admission. Mutually exclusive categories were created to identify the most common patterns of use, resulting in categories for individuals who reported using alcohol and no other illicit substances, those who reported using alcohol and marijuana only, and those who reported using alcohol and other illicit substances (e.g., cocaine, opiates, methamphetamines).

²Earliest age of first use of any of the substances reported as primary, secondary, or tertiary substance of abuse.

³Comparisons by race/ethnicity were done using Chi-Squares and ANOVA, followed by ad hoc pairwise comparisons. Differences at the p<.05 level using a Bonferroni correction for multiple comparisons

Regarding community characteristics, 41% of clients resided in communities characterized by a higher degree of economic disadvantage. Approximately half of clients (46%) resided in communities with a higher proportion of American Indian residents, while 36% resided in communities with a higher proportion of Latinos, and 30% resided in communities with a higher proportion of Black residents. The average proportion of households within a census tract that did not have a car was 10% (SD= 12.2%). Client community characteristics also differed by race/ethnicity. Latino and American Indian clients were more likely to live in communities characterized by a higher degree of economic disadvantage. Additionally, clients tended to reside in communities characterized by a higher proportion of their corresponding racial/ethnic identity. Black clients were significantly more likely to have lived in communities without access to a car compared with White, American Indian, and Latino clients.

Unadjusted DUI arrest outcomes

Prior to our multivariate analyses we examined rates of post-treatment arrest for a DUI overall, as well as by client race/ethnicity and by whether they had criminal justice involvement in the year prior to the treatment admission. Overall, approximately 9% of clients were arrested for a DUI-charge in the year following treatment admission. These rates varied by race/ethnicity; Latino (13%) and American Indian (11%) clients were significantly more likely to have had an arrest for a DUI compared with 8% of White clients and 6% of Black clients (all differences were significant at p< 0.01). These rates also varied based on whether clients had been arrested or incarcerated in the year prior to treatment. Clients who had been arrested/incarcerated prior to treatment were twice as likely to have been arrested for a DUI (12%) than those without an arrest or period of incarceration (6%) in the year prior to treatment (p< 0.01).

Multivariate results

Table 2 shows the results of our time-to-event analyses. For the full model, controlling for individual and community-level characteristics, Latino clients (HR: 1.43, p < 0.01) had higher hazards of a DUI arrest in the year following treatment admission compared to White clients. Clients residing in communities with a higher proportion of Black residents (HR: 1.46, p< 0.01) also had a higher hazard of a DUI arrest in the year following treatment admission. None of the other key community characteristics were significant.

Table 2.

Survival analysis of a DUI after beginning a new episode of outpatient treatment

	Full Sample (N= 8,136)		No Prior Arrest/Incarceration (N= 4,021)		Prior Arrest/Incarceration (N= 4,115)
Variable	HR	(95% CI)	HR	(95% CI)	HR	(95% CI)
Individual Characteristics
Race/Ethnicity (ref:White)
American Indian	1.22	(0.94, 1.57)	0.95	(0.60, 1.49)	1.32	(0.98, 1.78)
Latino	1.43**	(1.11, 1.85)	1.36	(0.86, 2.16)	1.43*	(1.06, 1.94)
Black	0.84	(0.59, 1.18)	0.85	(0.45, 1.63)	0.81	(0.54, 1.22)
Prior arrest or incarceration	2.07**	(1.72, 2.50)	-	-	-	-
Female	0.75**	(0.62, 0.91)	0.68*	(0.49, 0.94)	0.80	(0.64, 1.00)
Married	0.77*	(0.62, 0.94)	0.60*	(0.41, 0.89)	0.85	(0.67, 1.08)
Age (Ref: 45–54)
18–20	0.92	(0.62, 1.37)	0.50	(0.24, 1.05)	1.22	(0.76, 1.95)
21–25	1.07	(0.81, 1.42)	0.78	(0.46, 1.32)	1.20	(0.86, 1.68)
26–30	1.12	(0.85, 1.48)	1.01	(0.62, 1.64)	1.20	(0.86, 1.67)
31–44	1.04	(0.82, 1.31)	0.86	(0.57, 1.30)	1.13	(0.85, 1.51)
55+	1.03	(0.71, 1.50)	0.95	(0.52, 1.73)	1.07	(0.67, 1.72)
Education (Ref: Less than high school)
High School	0.91	(0.75, 1.10)	0.94	(0.66, 1.34)	0.90	(0.71, 1.12)
> High School	0.90	(0.65, 1.27)	0.57	(0.29, 1.09)	1.10	(0.74, 1.63)
Vocational	0.95	(0.66, 1.37)	1.34	(0.77, 2.33)	0.73	(0.44, 1.19)
Homeless	1.11	(0.86, 1.42)	1.00	(0.62, 1.60)	1.15	(0.86, 1.54)
Unemployed	0.93	(0.76, 1.13)	0.77	(0.55, 1.09)	1.01	(0.79, 1.30)
Referred to treatment by criminal justice system	2.05**	(1.64, 2.56)	3.08**	(2.17, 4.37)	1.47**	(1.12, 1.94)
Age of first use (Ref: 21+)
<10	1.40	(0.86, 2.29)	1.43	(0.61, 3.31)	1.38	(0.76, 2.54)
11–14	1.50	(0.96, 2.35)	1.44	(0.67, 3.10)	1.56	(0.90, 2.72)
15–17	1.40	(0.90, 2.18)	1.42	(0.67, 3.01)	1.43	(0.83, 2.47)
18–20	1.67*	(1.04, 2.66)	1.68	(0.75, 3.75)	1.72	(0.96, 3.06)
Substance use at admission† (Ref: Alcohol only)
Alcohol and Marijuana	0.83	(0.67, 1.04)	0.94	(0.63, 1.40)	0.81	(0.62, 1.07)
Alcohol and Other Drugs	0.69**	(0.56, 0.85)	0.71	(0.48, 1.04)	0.69**	(0.54, 0.89)
Community Variables
Higher Economic Disadvantage	0.87	(0.70, 1.10)	0.95	(0.66, 1.38)	0.81	(0.62, 1.06)
Higher % American Indian	0.90	(0.73, 1.10)	0.90	(0.63, 1.27)	0.92	(0.72, 1.18)
Higher % Latino	1.04	(0.84, 1.27)	1.06	(0.76, 1.48)	1.05	(0.82, 1.34)
Higher % Black	1.46**	(1.18, 1.81)	1.56*	(1.08, 2.25)	1.46**	(1.14, 1.87)
% Households without a car	0.08**	(0.02, 0.31)	0.01**	(0.00, 0.12)	0.16*	(0.04, 0.69)

Notes:

^*p<0.05;

^**p<0.01;

^†Clients were asked what their primary, secondary, and tertiary substances were at admission. Mutually exclusive categories were created to identify the most common patterns of use, resulting in categories for individuals who reported using alcohol and no other illicit substances, those who reported using alcohol and marijuana only, and those who reported using alcohol and other illicit substances (e.g., cocaine, opiates, methamphetamines).

Sensitivity analysis results

Results from the sensitivity analysis (Table 2) indicate that whether clients had been arrested or incarcerated prior to treatment is important for estimating DUI arrests post-treatment. For example, no racial/ethnic disparities were observed for clients without a prior arrest or incarceration. However, Latino clients did have a higher hazard of arrest for a DUI (HR: 1.43, p <0.05) among those with a prior arrest or incarceration in the year prior to treatment admission. Regardless of whether clients had an arrest or incarceration in the previous year, clients residing in communities with a higher proportion of Black residents were more likely to have experienced an arrest for a DUI post-treatment (No prior arrest/incarceration HR: 1.56, p< 0.05; Prior arrest/incarceration HR: 1.46, p< 0.01).

Discussion

This is the first study to examine racial/ethnic disparities in DUI arrests post-treatment, and to include community characteristics in the analysis. Findings from this study identified racial/ethnic disparities in DUI arrest outcomes. Specifically, these results indicate that Latino clients receiving outpatient substance use treatment were at greater risk than their White counterparts to be arrested for a DUI in the year following treatment. Furthermore, characteristics of clients’ residential community were also found to be important. Clients living in communities with a higher proportion of Black residents were significantly more likely to have a DUI arrest in the year after beginning treatment.

Very few studies have been conducted that examine post-treatment arrest or post-treatment DUI arrest as an outcome. One study that examined arrest as a post-treatment outcome for individuals receiving outpatient treatment in Washington state and found that, while American Indian clients had a higher risk of any arrest compared with Whites, Latinos did not.¹⁴ Similarly, no significant differences in risk of arrest were found between White and American Indian or Latino Social Security Insurance recipients utilizing outpatient treatment services in Washington.¹³ To the authors’ knowledge this is the first study to identify an increased risk for post-treatment DUI arrest among clients living in communities with higher proportion of Black residents. This finding is consistent with prior research examining other post-treatment outcomes, including one study that found that clients who lived in communities with a higher proportion of Black residents had lower wages and shorter length of employment.¹¹

There are several potential explanations for the study findings. First, individual client characteristics may influence their risk for post-treatment DUI arrest. It is possible that Latino clients enter treatment with greater symptom severity compared with their White counterparts and, therefore, they may have worse outcomes than other groups. Unadjusted patterns of drinking among the general population indicate that Latinos reported similar rates of binge drinking (29%) compared with American Indians (26%), Whites (26%), and Blacks (27%).¹⁵ Unadjusted rates of heavy drinking were also similar for American Indians (5%) and Latinos (5%), compared with Whites (7%) and Blacks (8%).¹⁵ However, after controlling for other individual characteristics, Latinos were found to have greater alcohol symptom severity compared with Whites in the general population.³⁵

Disparities in access to and use of treatment services may also account for unequal outcomes experienced by Latinos and for those living in communities with higher proportion of Black residents. Latinos are less likely to access treatment for substance use treatment compared to Whites,⁸ and they likely do not receive services that are tailored to their needs, leading to early drop-out and poorer outcomes compared with White clients. For example, one study found that Latinos traveled significantly farther than the average distance to an outpatient treatment site to attend treatment in settings with services offered in Spanish.³⁶ At the community level, there is some evidence that communities with a greater proportion of Black residents have more limited access to treatment sites that accept Medicaid (which represents a large proportion of addiction treatment settings).³⁷ Alternatively, some evidence suggests that as the density of providers increases, so too does the rate of past-year use of formal treatment services particularly among Blacks and Latinos.³⁸ Future research should examine how treatment availability is implicated in racial/ethnic disparities in treatment outcomes. Previous research identified disparities in rates of treatment completion for Latinos,^10,39 which could lead to worse alcohol outcomes and increased risk of arrest among this racial/ethnic group. Prior research also suggests that clients in communities characterized by a higher proportion of Black residents were also less likely to receive timely outpatient substance use treatment services after accessing treatment.²²

Neighborhood characteristics also may contribute to an increased risk of post-treatment DUI arrest. Previous research found a disproportionate concentration of alcohol outlets⁴⁰ and alcohol advertising⁴¹ in Black communities. This increased exposure and access to alcohol can result in worse alcohol use outcomes such as greater alcohol consumption⁴² and more alcohol-related vehicle crashes⁴³ in communities with a higher concentration of Black residents. Prior research also suggests that neighborhoods with a greater proportion of Black or Latino residents incur a greater number of traffic stop citations even after controlling for individual driver and other contextual factors.⁴⁴

Policing practices could also explain the observed arrest disparities for Latino clients. Nearly 20 years ago the ACLU produced an extensive report examining the phenomenon of “driving while Black or brown”, referring to the overrepresentation of Black and brown drivers in traffic stops, searches, citations, and arrests.⁴⁵ Empirical evidence suggests that, while Latino drivers may not be stopped for traffic violations more than Whites, they are more likely to be ticketed and twice as likely to be arrested compared with Whites once they have been stopped.⁴⁶ Research evidence indicating an increased police presence in Black communities⁴⁷ also indicates a greater risk of arrest in general. Future research should examine how policing practices may affect disproportionate arrest rates among racial/ethnic minority groups, as well as those aimed at deterring DUI (e.g., sobriety checkpoints) to determine whether they are being implemented equitably across communities.

If symptom severity is greater for some racial/ethnic minority groups at the point they enter treatment, making it more likely they will experience a DUI arrest post-treatment, future research should also focus on implementing and evaluating interventions in treatment programs that focus on these groups. Some evidence suggests that clients were retained in publicly funded outpatient treatment longer if they received care in settings that were more culturally responsive.⁴⁸ Research is also needed to establish the effectiveness of outpatient substance use treatment for addressing the needs of clients with a prior history of criminal justice involvement, particularly among Latino clients. Additionally, the adoption of evidence-based practices at substance use treatment sites overall is limited⁴⁹ and the extent to which adoption rates vary across treatment settings or communities may also contribute to disparities in treatment outcomes.

Several limitations are worth noting. Client treatment records were merged with arrest records using various identifiers and a well-established methodology and software. However, unmatched arrests could still have resulted if arrests took place in a different state, or due to errors in data entry or missing data of identifiers used in the linkage. These issues would have resulted in an underestimation of arrest rates. This study used information on the client’s residential census tract at the time of treatment admission. Some clients may have relocated after admission to an area with different characteristics that may influence the probability of their arrest. Our findings may be specific to Washington State, or to clients receiving public outpatient treatment. Given that the primary source of funding for treatment programs comes from public sources,^50,51 our findings while limited may still generalize to the large population of clients receiving publicly-funded services. Access to evidence-based effective treatment likely varies from site to site. Treatment service access, delivery system organization and financing, population and community characteristics, alcohol and drug control policies, and law enforcement policies and priorities also vary by state. These factors may affect treatment outcomes, arrest rates, as well as disparities in arrests, and future research should examine similar research questions in states with varying characteristics. Finally, we did not include treatment completion in our analyses. Future research should examine the association between community characteristics and racial/ethnic disparities in the likelihood of completing treatment.

Implications for Behavioral Health.

Latino clients with an alcohol use disorder in outpatient treatment were more likely to experience a DUI arrest compared with their White counterparts. Additionally, clients living in communities with a higher proportion of Black residents were more likely to have a DUI arrest after beginning treatment. Reducing criminal justice involvement among individuals attending treatment for a substance use disorder is an important treatment goal, and a reduction in post-treatment DUI may indicate improved drinking behaviors as a result of treatment attendance. While many factors outside of the treatment environment affect clients’ risk for DUI arrest post-treatment, providers should be culturally responsive to their client populations and implement interventions to the extent they are able. Future research should examine the interaction between treatment effectiveness, community characteristics, and inequities in law enforcement so that these disparities are better understood and can be addressed.