Treatment outcome disparities for opioid users: Are there racial and ethnic differences in treatment completion across large US metropolitan areas?

Drug and Alcohol Dependence 190 (2018) 170–178

Contents lists available at ScienceDirect

Drug and Alcohol Dependence journal homepage: www.elsevier.com/locate/drugalcdep

Full length article

Treatment outcome disparities for opioid users: Are there racial and ethnic differences in treatment completion across large US metropolitan areas?

T

⁎

Gerald J. Stahler , Jeremy Mennis Department of Geography and Urban Studies, Temple University, (025-27), 309 Gladfelter Hall, Philadelphia, PA 19122, United States

A R T I C LE I N FO

A B S T R A C T

Keywords: Treatment completion Disparities Treatment outcomes Metropolitan areas Opioid users

Background: The present study examined racial/ethnic disparities in initial treatment episode completion for adult clients reporting opioids as their primary problem substance in large US metropolitan areas. Methods: Data were extracted from the 2013 TEDS-D dataset (Treatment Episode Dataset-Discharge) for the 42 largest US metropolitan statistical areas (MSAs). Fixed effects logistic regression controlling for MSA was used to estimate the effect of race/ethnicity on the likelihood of treatment completion. The model was repeated for each individual MSA in a stratified design to compare the geographic variation in racial/ethnic disparities, controlling for gender, age, education, employment, living arrangement, treatment setting, medication-assisted treatment, referral source, route of administration, and number of substances used at admission. Results: Only 28% of clients completed treatment, and the results from the fixed effects model indicate that blacks and Hispanics are less likely to complete treatment compared to whites. However, the stratified analysis of individual MSAs found only three of the 42 MSAs had racial/ethnic disparities in treatment completion, with the New York City (NYC) MSA largely responsible for the disparities in the combined sample. Supplementary analyses suggest that there are greater differences between whites and minority clients in the NYC MSA vs. other cities on characteristics associated with treatment completion (e.g., residential treatment setting). Conclusion: This study underscores the need for improving treatment retention for all opioid using clients in large metropolitan areas in the US, particularly for minority clients in those localities where disparities exist, and for better understanding the geographic context for treatment outcomes.

1. Introduction 1.1. Background Illicit opioid use represents one of the most harmful drug problems globally, responsible for an estimated 70% of the world’s burden of disease attributable to drug use disorders as well as 66% of the 63,632 US drug overdose deaths in 2016 (Seth et al., 2018; United Nations Office on Drugs and Crime (UNODC, 2017). Although the US is the global leader in both absolute numbers (one quarter of the world’s total) and rates of overdose deaths, other nations such as Canada, Australia, Ireland, Turkey, England, Wales, and Scotland have all seen recent substantial increases in overdose mortality, primarily due to opioids (European Monitoring Centre for Drugs and Drug Addiction (EMCDDA, 2017a; United Nations Office on Drugs and Crime (UNODC, 2017). An important pillar for policy strategies to address the current opioid overdose crisis has been to provide increased access and capacity for treatment for those with opioid use disorders (OUDs) (Evans and

⁎

Farrelly, 2017; Franklin et al., 2015; Murphy et al., 2016). One of the most widely used proximal measures of treatment effectiveness for substance use disorders (SUDs) is treatment completion (Brorson et al., 2013), generally defined as successfully completing treatment goals (Greenfield et al., 2007). Despite evidence showing sustained recovery may involve multiple episodes over time (Guerrero, 2013; McKay and Weiss, 2001), individual treatment completion episodes can serve as an important indicator associated with longer term abstinence, fewer relapses, higher levels of employment, higher wages, fewer readmissions, less future criminal involvement, and better health (Brorson et al., 2013). Black and Hispanic people in the US tend to have lower treatment utilization rates, greater barriers to receiving treatment, and poorer outcomes, including treatment completion, compared to white clients (Alegría et al., 2006, 2011; Arndt et al., 2013; Guerrero et al., 2013a; National Research Council, 2003; Saloner and Le Cook, 2013). Similarly, in Europe, there has been a recent recognition that ethnic minorities, migrants, refugees, and asylum seekers who have substance

Corresponding author. E-mail addresses: [email protected] (G.J. Stahler), [email protected] (J. Mennis).

https://doi.org/10.1016/j.drugalcdep.2018.06.006 Received 10 January 2018; Received in revised form 1 June 2018; Accepted 4 June 2018 Available online 11 July 2018 0376-8716/ © 2018 Elsevier B.V. All rights reserved.

Drug and Alcohol Dependence 190 (2018) 170–178

G.J. Stahler, J. Mennis

(SAMHSA, 2015). This dataset is compiled from an annual survey conducted by the federal government of state agencies concerning SUD treatment programs. It includes most publicly funded treatment admissions in the US during a given year and is considered generally representative of a national sample (Substance Abuse and Mental Health Services Administration (SAMHSA, 2015). For the present study, we limited our sample to adults (18 and over) whose primary substance use was opioids (including heroin, non-prescription methadone, and synthetic and other opioids) living in a metropolitan statistical area (MSA) with a population greater than one million (according to the US Census Bureau). We included residential and outpatient treatment settings, but detoxification was excluded since it is generally regarded as only an initial stage of treatment (NIDA, 2018). This exclusion is consistent with other studies on treatment completion using the TEDS-D dataset (e.g., Krawczyk et al., 2017a; Sahker et al., 2015; Saloner et al., 2014; Stahler et al., 2016). Because the dataset is based on discharges and not individuals, we limited our sample to only those records that had “no prior admissions,” ensuring that each case is a unique individual, an approach consistent with our own and others’ prior research (Arndt et al., 2013; Krawczyk et al., 2017a; Sahker et al., 2015; Stahler et al., 2016). Our final sample comprised 34,380 cases located in 42 MSAs.

use problems may be particularly vulnerable to barriers in accessing needed treatment services (European Monitoring Centre for Drugs and Drug Addiction (EMCDDA, 2017b). Evidence suggests that race/ethnicity also interacts with other factors, such as drug of choice and treatment modality, to produce disparities in treatment completion (Mennis and Stahler, 2016; Stahler et al., 2016), and these differences in outcomes vary considerably across different areas of the country (Arndt et al., 2013; Cummings et al., 2011; Guerrero et al., 2013a). The use of aggregate national level data may therefore mask geographic variations that may have important implications for SUD treatment policies that are generally implemented at the regional or municipal level in the US. This is particularly important when examining treatment outcomes for OUDs because of geographic variations in demographic characteristics, prevalence, overdose incidence, treatment systems, illicit opioid availability, and insurance coverage that exist in the US (Cummings et al., 2014; Hand et al., 2017; Martins et al., 2017; Rossen et al., 2014; Substance Abuse and Mental Health Services Administration (SAMHSA, 2017). Even in nations with more centralized national treatment systems, geographic variations in prevalence patterns and treatment access may be substantial (Morley et al., 2017; United Nations Office on Drugs and Crime (UNODC, 2017). While epidemiologic research has found important geographic and demographic differences in patterns of opioid use cross nationally (United Nations Office on Drugs and Crime (UNODC, 2017), among rural and urban users (Rigg and Monnat, 2015), across regions (Hand et al., 2017; Martins et al., 2017), and across different metropolitan areas (Roberts et al., 2010), few studies have focused on the geographic variation in racial/ethnic treatment outcome disparities (Arndt et al., 2013).

2.2. Independent and dependent variables Our primary dependent variable was treatment episode completion, defined in the dataset as “all parts of the treatment plan or program were completed,” and where non-completion included the designations of “left against professional advice” and “terminated by the facility” (Substance Abuse and Mental Health Services Administration (SAMHSA, 2015). We use the term “treatment completion” here to refer to completion of a single treatment episode. As in previous research (Mennis and Stahler, 2016), we excluded cases with other discharge outcomes including “transferred to another treatment program or facility,” “incarcerated,” “death,” “other,” and “unknown” to create the strongest possible contrast between successful and unsuccessful treatment completion outcomes. In addition, we note that the dataset includes only clients who are discharged from treatment during the particular year; data for clients admitted but currently receiving treatment would not be included in the dataset. Our primary independent variable was race/ethnicity, coded using five mutually exclusive categories: non-Hispanic white, non-Hispanic black, Hispanic, Asian (including Pacific Islanders), and Other race/ethnicity, where Hispanics of any race or ethnic origin were included in the Hispanic category (Mennis and Stahler, 2016; Sahker et al., 2015; Saloner and Le Cook, 2013). Control variables were derived from prior research (Mennis and Stahler, 2016; Stahler et al., 2016) and included: gender, age, education, employment, living arrangement, treatment setting, medicationassisted treatment, referral source, primary route of administration, and number of substances used at admission.

1.2. Present study The present study examined racial and ethnic disparities in successfully completed treatment episodes for first time adult clients reporting opioids as a primary substance of use across the largest US metropolitan areas (populations greater than one million) using data extracted from the 2013 Treatment Episode Dataset-Discharge (TEDSD) (Substance Abuse and Mental Health Services Administration (SAMHSA, 2015). We limited our sample to large metropolitan areas because there are considerable differences between urban and rural drug use patterns, treatment systems, and demographic characteristics, especially regarding opioid use (Wang et al., 2013). In addition, there are particularly large concentrations of minority groups in the largest metropolitan areas, and many smaller metropolitan areas have few OUD treatment discharges for certain racial/ethnic groups in the TEDSD dataset, limiting statistical analyses for these areas. Our major research questions were: (1) Are there racial/ethnic disparities in first time treatment episode completion for clients reporting opioid use as their primary substance of use in large metropolitan areas in the US? Based on prior research demonstrating racial/ethnic disparities (e.g., Guerrero et al., 2013a, b; Mennis and Stahler, 2016), we hypothesize that black and Hispanic opioid users will have a lower likelihood of treatment completion compared to whites. (2) If there are racial/ethnic disparities in first time treatment episode completion for clients reporting opioid use as their primary substance of use, do these disparities vary across large metropolitan areas in the US? Based on prior research showing geographic variation in disparities in treatment completion (Arndt et al., 2013; Saloner et al., 2014), we hypothesize that metropolitan areas will vary significantly in their level of racial/ethnic disparity.

2.3. Analytic plan To investigate the unadjusted effect of each of our independent variables on treatment completion, we used the chi-squared statistic to test for significant differences for each independent variable categorical value. We employed logistic regression to estimate the effect of race/ ethnicity on the probability of treatment completion while controlling for the confounding variables listed above. We then tested a second analogous model using a fixed effects logistic regression to control for the effect of the MSA within which the treatment discharge occurred. To investigate whether the effect of race/ethnicity on treatment completion differs by MSA, we employed a stratified design to calculate the treatment completion percentage for each race/ethnicity category within each MSA. We then estimated separate logistic regression equations of race/ethnicity on the likelihood of treatment completion

2. Methods 2.1. Data source and sample The data used for this study were extracted from the 2013 TEDS-D dataset (Substance Abuse and Mental Health Services Administration 171

Drug and Alcohol Dependence 190 (2018) 170–178

G.J. Stahler, J. Mennis

3. Results

Table 1 Frequencies, treatment completion percentages, and results of chi-squared tests for variables used in the analysis (N = 34,380; ***=p < 0.005). Variable

Entire Sample Age

Sex Race/Ethnicity

Education

Employment

Housing Status

Treatment Setting Referral Source

MAT Route of Administration

Number of Substances

Values

18-20 21-24 25-29 30-34 35-39 40-44 45-49 50-54 > 55 Male Female White Black Hispanic Asian Other 8 Years or Less 9-11 Years 12 Years 13-15 Years 16 Or More Years Full Time Part Time Unemployed Not in Labor Force Homeless Dependent Living Independent Living Residential Outpatient Individual (Self) Drug/Alc Provider Health Provider School Employer Other Community Criminal Justice YES NO Oral Smoking Inhalation Injection Other 1 2 3 or more

N

%

% Completed Treatment

3.1. Sample characteristics

Pearson chisquared

34,380 2414 6811 7524 5473 3134 2574 2550 1950 1950 20181 14194 24,218 4742 3995 325 1100 1464

100.0 7.0 19.8 21.9 15.9 9.1 7.5 7.4 5.7 5.7 58.7 41.3 70.4 13.8 11.6 0.9 3.2 4.3

27.8 28.8 29.4 29.0 27.9 27.3 26.3 25.0 24.0 25.3 29.0 26.0 29.4 20.9 26.8 27.7 23.7 22.5

8089 15185 7462 1780

23.8 44.7 22.0 5.2

23.4 28.5 29.3 37.3

4662 2582 14545 12155

13.7 7.6 42.8 35.8

28.8 23.9 25.7 30.0

82.41***

3099 7444

9.1 21.8

28.6 35.3

290.74***

23610

69.1

25.2

9164 25216 16562

26.7 73.3 48.8

46.9 20.8 21.7

3308

9.7

32.5

1990

5.9

24.6

19 121 3759

0.1 0.4 11.1

10.5 51.2 29.6

8188

24.1

37.7

9102 24113 9168 2422 8253 14002 396 12750 12359 9271

27.4 72.6 26.8 7.1 24.1 40.9 1.2 37.1 35.9 27.0

13.0 31.8 30.6 27.6 25.4 27.3 29.0 28.8 28.1 26.0

Table 1 displays the descriptive statistics for the outcome and explanatory variables. For the total sample of 34,380 discharges, only 27.8% completed treatment. The majority of the sample (70%) was white, with the remainder almost exclusively comprised of black (14%) and Hispanic (12%) people. About one quarter of the sample had less than a high school education (28%), most were not employed (79%), two thirds (69%) lived in independent housing, and the majority received treatment in outpatient settings (73%). Additionally, about one quarter (24%) were referred by the criminal justice system, 41% were injection users, and about a quarter received medication assisted treatment (MAT) (27%). Treatment completion varied significantly by race/ethnicity with white clients having the highest completion rate (29%) and black clients having the lowest (21%).

48.45***

36.17*** 156.73***

3.2. Results of the logistic regression of treatment completion for the combined total sample 189.88***

Results of the logistic regression of treatment completion for the entire sample of 34,380 cases are shown in Table 2, where Models 1 and 2 report the odds ratios for regressions with no fixed effects and with metropolitan area fixed effects, respectively. Model 2 indicates that after controlling for the effects of the confounding variables, black (OR = 0.82) and Hispanic clients (OR = 0.81) were significantly less likely to complete treatment compared to whites. Homelessness and using more than one substance at admission were also associated with a decreased likelihood of treatment completion. Greater likelihood for treatment completion was associated with older age (over 50 years old); 12 or more years of education; full-time employment; residential treatment; and referrals from the criminal justice system, other drug or alcohol providers, and employers.

2282.50***

3.3. Results of the stratified analysis by metropolitan area 800.72***

Table 3 reports the count and treatment completion percentages for each MSA for all cases, as well as for each of the three primary race/ ethnicity groups (white, black, and Hispanic). Also reported are adjusted odds ratios for the black and Hispanic groups (compared to whites) in the MSA-specific models. Because some metropolitan areas have very low numbers of discharges for black and/or Hispanic clients, we report only the adjusted odds ratios for race/ethnicity categories with at least 30 discharges in an MSA. Numbers of cases within individual MSAs ranged from 110 in San Antonio to 7853 in NYC, the latter accounting for 23% of all discharges in the sample. Treatment completion rates varied widely across MSAs, ranging from 9% in Baltimore to 52% in Hartford. In MSAs with more than 30 black discharged clients, completion percentages varied from 0% (Kansas City) to 60% (Denver), with a similar range for Hispanics–4% (Baltimore) to 79% (Nashville). Results of the stratified logistic regressions indicate that only three of the 42 MSAs showed significant racial/ethnic disparities in treatment completion, after controlling for the other explanatory variables in the model. Blacks were significantly less likely to complete treatment than whites in the Buffalo, NYC, and Riverside MSAs. Similarly, only in the Riverside and NYC MSAs were Hispanics significantly less likely to complete treatment than whites.

1194.32*** 6297***

22.11***

for the subset of discharges within each of the 42 MSAs, while controlling for the confounding variables listed above. Regression models were implemented in IBM SPSS Statistics Version 24 using the Logistic Regression function. Because we found relatively unique results for the New York City (NYC) MSA (see below), which also contained the largest number of cases by far, we performed a supplementary analysis comparing treatment completion and other characteristics of NYC MSA discharges to those from the other MSAs using descriptive statistics and chi-squared tests.

3.4. Supplementary analysis of New York City versus other metropolitan areas To further investigate the results of the stratified analyses of individual MSAs, we conducted a series of supplementary analyses by repeating the fixed effects logistic regression analysis (Model 2) 172

Drug and Alcohol Dependence 190 (2018) 170–178

G.J. Stahler, J. Mennis

Table 2 Results of logistic regression and fixed effects logistic regression; Odds ratios are reported with confidence intervals in parentheses (N = 31,939; *=p < 0.05, **=p < 0.01, ***=p < 0.005). Variable

Values

Model 1 (No Fixed Effects)

Model 2 (Fixed Effects)

Age

18-20 21-24 25-29 30-34 35-39 40-44 45-49 50-54 > 55 Male Female White Black Hispanic Asian Other 8 Years or Less 9-11 Years 12 Years 13-15 Years 16 Or More Years Full Time Part Time Unemployed Not in Labor Force Homeless Dependent Living Independent Living Residential Outpatient Individual (Self) Drug/Alc Provider Health Provider School Employer Other Community Criminal Justice YES NO Oral Smoking Inhalation Injection Other 1 2 3 or more

– 1.02 1.02 0.99 1.02 1.08 1.10 1.15 1.41 – 0.93 – 0.68 0.95 1.00 0.82 – 0.96 1.24 1.31 1.89 – 0.80 0.76 0.83 – 1.43 1.09 – 0.37 – 1.14 1.00 0.36 2.69 1.28 1.77 – 1.89 – 0.79 0.78 0.72 0.77 – 0.88 0.81

– 1.04 1.04 1.02 1.03 1.11 1.14 1.20 1.46 – 0.95 – 0.82 0.81 1.04 0.82 – 1.00 1.24 1.27 1.69 – 0.80 0.80 0.83 – 1.47 1.19

Sex Race/Ethnicity

Education

Employment

Housing Status

Treatment Setting Referral Source

MAT Route of Administration

Number of Substances

(0.91-1.14) (0.91-1.15) (0.88-1.12) (0.89-1.17) (0.93-1.24) (0.95-1.27) (0.99-1.35) (1.20-1.65)*** (0.88-0.99)* (0.62-0.74)*** (0.87-1.03) (0.76-1.31) (0.70-0.95)* (0.83-1.11) (1.08-1.43)*** (1.13-1.52)*** (1.59-2.25)*** (0.71-0.90)*** (0.70-0.83)*** (0.76-0.91)*** (1.29-1.58)*** (0.99-1.19) (0.34-0.39)*** (1.04-1.25)** (0.89-1.23) (0.08-1.63) (1.86-3.90)*** (1.17-1.40)*** (1.65-1.89)*** (1.74-2.04)*** (0.71-0.89)*** (0.72-0.85)*** (0.67-0.77)*** (0.60-0.98)* (0.83-0.93)*** (0.75-0.86)***

0.35 – 1.26 1.00 0.35 2.70 1.26 1.78 – 1.95 – 0.79 0.77 0.74 0.74 – 0.90 0.86

(0.92-1.17) (0.93-1.17) (0.90-1.15) (0.89-1.18) (0.96-1.29) (0.98-1.32) (1.03-1.41* (1.25-1.72)*** (0.90-1.01) (0.74-0.91)*** (0.74-.89)*** (0.79-1.37) (0.70-0.96)* (0.86-1.16) (1.07-1.43)*** (1.09-1.48)*** (1.41-2.02)*** (0.70-0.90)*** (0.72-0.86)*** (0.76-0.91)*** (1.32-1.63)*** (1.07-1.31)*** (0.33-0.38)*** (1.14-1.40)*** (0.89-1.13) (0.07-1.66) (1.84-3.96)*** (1.15-1.38)*** (1.65-1.91)*** (1.79-2.12)*** (0.70-0.89)*** (0.71-0.84)*** (0.69-0.79)*** (0.57-0.95)* (0.85-0.96)*** (0.80-0.92)***

which is similar to that of white clients in the other MSAs (58%). For black and Hispanic users in the NYC MSA, the rate is 88% and 84%, respectively, which is higher than for black and Hispanic users in other MSAs (80% and 73%, respectively). Across the entire sample heroin users completed treatment at a lower rate (26%) compared to other opioid users (31%), and this difference is even greater for black (20% and 27%, respectively) and Hispanic (25% and 34%, respectively) users. The magnitude of this pattern appears to be even stronger in the NYC MSA, where treatment completion rates among white heroin versus other opioid users are about the same (41% and 40%, respectively), but differ substantially among blacks (26% and 43%, respectively) and Hispanics (24% and 33%, respectively).

displayed in Table 3, but excluding the NYC and Buffalo MSAs. None of the adjusted coefficients for any of the race/ethnicity categories were statistically significant at P < 0.05. Given the large number of cases from the NYC MSA, we then focused our supplementary analysis on comparing the NYC MSA to the combined sample of all other MSAs. Table 4 reports descriptive statistics for the NYC vs. all other MSAs for all variables in our analysis broken down by race and ethnicity. Notably, the NYC MSA has a substantially higher overall completion percentage (36%) compared to the other MSAs (26%), and within the NYC MSA, the disparity in treatment completion is considerable – 41%, 28%, and 25% for whites, blacks and Hispanics, respectively. White clients in the NYC MSA sample were more likely to have more years of formal education, more full time employment, and to receive residential treatment, suggesting that the white clients in the NYC MSA tend to have a higher socioeconomic status (SES) than those in the other MSAs, whereas the SES of black and Hispanic clients in the NYC MSA were similar to those in other MSAs. Another issue concerns the proportion of cases using heroin rather than non-prescription methadone and other opiates and synthetics. In the NYC MSA, the percentage of heroin use among whites is 60%,

4. Discussion To the best of our knowledge, this is the first analysis to examine racial/ethnic disparities in treatment episode completion for opioid using clients across and within the largest US metropolitan areas. Our results concerning racial/ethnic disparities in OUD treatment completion is consistent with a large body of research that has examined 173

Drug and Alcohol Dependence 190 (2018) 170–178

G.J. Stahler, J. Mennis

Table 3 Count and treatment completion percentages for whites, blacks, and Hispanics, for each metropolitan area. Adjusted odds ratios reported for blacks and Hispanics. All Discharges

White

Black

Hispanic

Metropolitan Area

N

% of Total

% Com.

N

% Com.

N

% Com.

OR

N

% Com.

OR

Baltimore Boston Buffalo Charlotte Chicago Cincinnati Cleveland Columbus Dallas Denver Detroit Hartford Houston Kansas City Las Vegas Los Angeles Louisville Memphis Minneapolis Nashville New York Philadelphia Pittsburgh Portland Providence Raleigh Richmond Riverside Rochester Sacramento Salt Lake City San Antonio San Diego San Francisco San Jose Seattle-Tacoma St. Louis Virginia Beach Washington

2298 497 835 683 1190 916 315 487 286 482 1168 239 183 189 346 1404 963 153 1032 142 7853 1136 420 1733 657 407 262 1095 487 498 546 110 929 951 119 1457 591 203 1118

7% 1% 2% 2% 4% 3% 1% 1% 1% 1% 3% 1% 1% 1% 1% 4% 3% < 1% 3% < 1% 23% 3% 1% 5% 2% 1% 1% 3% 1% 1% 2% < 1% 3% 3% < 1% 4% 2% 1% 3%

9% 26% 23% 40% 21% 34% 30% 16% 50% 44% 18% 52% 43% 14% 26% 22% 18% 49% 38% 51% 35% 37% 34% 22% 24% 17% 35% 30% 25% 24% 38% 39% 28% 26% 34% 21% 14% 29% 20%

1312 409 702 639 449 897 277 457 196 344 931 152 114 155 252 734 868 82 793 106 4951 984 374 1515 612 376 142 622 422 337 425 56 620 483 56 1119 355 133 767

8% 26% 24% 39% 26% 34% 31% 17% 52% 45% 18% 60% 42% 17% 27% 23% 18% 51% 42% 43% 41% 37% 34% 22% 24% 16% 39% 34% 24% 25% 39% 48% 30% 27% 32% 22% 17% 27% 19%

891 23 45 23 623 6 22 24 39 20 175 20 38 27 13 101 49 20 96 6 1167 70 28 58 17 18 105 37 17 53 9 3 32 230 6 78 212 60 281

10% 26% 11% 48% 17% 33% 27% 4% 46% 60% 22% 25% 50% 0% 23% 13% 10% 35% 21% 50% 28% 39% 21% 17% 18% 22% 31% 16% 24% 23% 11% 33% 25% 24% 50% 18% 10% 27% 25%

0.98 – 0.36* – 0.86 – – – 1.71 – 1.17 – 1.45 – – 0.66 0.79 – 1.04 – 0.63*** 0.94 – 0.81 – – 0.91 0.21*** – 1.52 – – 0.56 0.90 – 0.73 0.69 0.63 1.57

53 46 54 2 94 2 12 1 50 102 20 60 25 3 69 471 3 51 41 29 1521 62 14 16 12 10 9 368 35 64 93 49 219 154 43 94 4 4 36

4% 20% 22% 50% 22% 50% 17% 0% 44% 42% 30% 40% 44% 0% 22% 21% 67% 51% 44% 79% 25% 42% 50% 13% 25% 40% 33% 26% 26% 23% 39% 29% 24% 23% 35% 22% 25% 100% 22%

0.30 0.65 0.86 – 0.96 – – – 1.09 0.94 – 0.46 – – 0.58 1.12 – 0.84 1.50 – 0.64*** 0.79 – – – – – 0.65* 1.32 1.20 0.75 1.00 0.84 0.79 2.08 1.15 – – 1.42

among minority populations (Schneider, 2008). This would be consistent with our findings that the differences within the NYC MSA among client groups on characteristics associated with treatment completion (e.g., SES, drug of choice, residential setting) (Guerrero, et al., 2013b; Jacobsen et al., 2007; Saloner and Le Cook, 2013) seem to be magnified compared to the other MSAs. In Europe, for example, one of the most urbanized regions of the world, even though drug policies and treatment systems are organized at the national level, there are still considerable variations across metropolitan areas within the same countries in terms of problem drug use and treatment services, and local urban authorities are often involved in developing city-level drug policies and in implementing policy strategies and treatment services (European Monitoring Centre for Drugs and Drug Addiction (EMCDDA, 2015). Given that treatment outcomes are likely a result of a complex interplay of factors at the individual client, program, service system, and policy levels that may have differential effects on ethnic and racial groups, it is important that future research take into account local contextual variables, even if place is in fact a proxy for these various factors.

national level SUD treatment completion data (Arndt et al., 2013; Mennis and Stahler, 2016; Saloner et al., 2014; Stahler et al., 2016) as well as regional and local data (Cooper et al., 2010; Guerrero et al., 2013b; Jacobson et al., 2007). However, when focusing on individual MSAs, we found that these disparities were driven largely by cases in the NYC MSA, as well as in two other smaller MSAs. 4.1. Geographic variability in racial/ethnic disparities in treatment completion The results of this study and those of Arndt et al. (2013) suggest that racial and ethnic disparities in treatment completion vary geographically – that they are not evenly distributed across cities, regions, and states. Location may be a proxy for variability in policy and regional program practice, and may intersect with such other factors as historical trajectories of race and ethnicity, cultures of drug use, variations in drug supply and drug control efforts, differences in insurance coverage, and characteristics of SUD treatment programs that may be unique to individual metropolitan areas (Arndt et al., 2013; Cummings et al., 2014; Guerrero et al., 2013a, 2017; Velez et al., 2008). Outcomes may also be influenced by the different neighborhoods within which clients live (Karriker-Jaffe et al., 2012; Mennis et al., 2012; Stahler et al., 2007, 2009). We speculate that the difference between the NYC MSA and the other metropolitan areas may in part be due to the historical legacy of heroin use in New York City (compared to other cities), particularly

4.2. The need to address racial and ethnic disparities in treatment program completion Our results emphasize the importance of identifying and addressing racial and ethnic disparities in treatment outcomes in areas where they do exist. Prior research in the US and elsewhere has documented that 174

Drug and Alcohol Dependence 190 (2018) 170–178

G.J. Stahler, J. Mennis

Table 4 Frequencies and percentages of variables by race/ethnicity for NYC MSA vs. Other MSAs. Variable

Treatment Completion Education

Age

Gender Employ-ment

Housing Status

Arrests Prior 30 days Treatment Setting Primary Drug

MAT Referral Source

Primary Route of Admin

# Substanc. at Admiss.

Total Sample

YES NO 8 Years or Less 9-11 Years 12 Years 13-15 Years 16 Or More Years 18 -24 25-34 35-44 45-54 > 55 Female Male Full Time Part Time Unemployed Not in Labor Force Homeless Dependent Living Independent Living None Once 2 or More Times Residential Outpatient Heroin Non-Prescription Methadone Other Opiates/ Synthetic YES NO Individual (Self) Drug/Alc Provider Health Provider School Employer Other Community Criminal Justice Oral Smoking Inhalation Injection One Two Three

Whites

Blacks

Hispanics

All Non-NYC cities (n = 26,527)

NYC (n = 7853)

All Non- NYC cities (n = 19,267)

NYC (n = 4951)

All Non-NYC cities (n = 3575)

NYC (n = 1167)

All Non-NYC cities (n = 2474)

NYC (n = 1521)

26% (6757) 75% (19,770) 4% (1145) 25% (6421) 45% (11,784) 22% (5699) 5% (1201) 27% (7122) 39% (10,424) 16% (4267) 12% (3218) 6% (1496) 44% (11,746) 56% (14,776) 13% (3364) 8% (2089) 48% (12,514) 32% (8375) 9% (2342) 24% (6360) 67% (17,618) 93% (24,173) 7% (1733) 1% (197) 24% (6435) 76% (20,092) 62% (15,730) 1% (258)

36% (2784) 65% (5069) 4% (319) 22% (1668) 44% (3401) 23% (1763) 8% (579) 27% (2103) 33% (2573) 18% (1441) 16% (1282) 6% (454) 31% (2448) 69% (5405) 17% (1298) 7% (493) 27% (2031) 50% (3780) 10% (757) 14% (1084) 77% (5992) 92% (7183) 7% (551) 1% (60) 35% (2729) 65% (5124) 69% (5266) 0% (23)

27% (5119) 73% (14,148) 4% (709) 21% (4044) 46% (8771) 24% (4558) 5% (1001) 30% (5829) 43% (8366) 15% (2833) 8% (1618) 3% (621) 47% (9030) 53% (10,233) 15% (2799) 9% (1676) 48% (9107) 29% (5546) 9% (1622) 22% (4194) 70% (13,293) 93% (17,529) 7% (1253) 1% (146) 25% (4760) 75% (14,507) 58% (11,072) 1% (230)

41% (2010) 59% (2941) 2% (97) 14% (698) 46% (2213) 28% (1352) 10% (485) 36% (1787) 39% (1954) 14% (666) 8% (387) 3% (157) 35% (1706) 66% (3245) 21% (1019) 8% (387) 27% (1268) 44% (2080) 6% (285) 13% (657) 81% (3993) 91% (4595) 85 (397) 1% (41) 38% (1866) 62% (3085) 60% (2960) 0% (13)

19% (662) 82% (2913) 5% (172) 36% (1267) 42% (1486) 15% (529) 2% (78) 9% (331) 18% (649) 22% (801) 32% (1133) 19% (661) 36% (1299) 64% (2276) 5% (182) 5% (166) 53% (1330) 37% (1330) 9% (305) 28% (998) 63% (2249) 94% (3322) 5% (186) 1% (21) 21% (744) 79% (2831) 80% (2843) 0% (12)

28% (839) 72% (839) 5% (57) 34% (395) 44% (505) 15% (168) 3% (36) 5% (56) 11% (126) 24% (276) 45% (530) 15% (179) 30% (348) 70% (819) 7% (82) 3% (39) 27% (302) 63% (718) 17% (202) 17% (196) 66% (768) 94% (1084) 5% (60) 0% (4) 30% (348) 70% (819) 88% (1028) 0% (2)

28% (688) 72% (1786) 8% (195) 33% (812) 42% (1031) 14% (352) 2% (53) 25% (605) 36% (879) 19% (469) 15% (370) 6% (151) 35% (877) 65% (1597) 11% (274) 7% (169) 41% (996) 42% (1020) 12% (289) 34% (834) 54% (1339) 91% (2218) 8% (204) 1% (23) 26% (650) 74% (1824) 73% (1815) 1% (16)

25% (383) 75% (1138) 10% (156) 35% (529) 40% (600) 13% (190) 2% (36) 13% (202) 27% (412) 30% (458) 23% (342) 7% (107) 22% (339) 78% (1182) 10% (154) 4% (62) 27% (396) 59% (885) 16% (243) 14% (205) 71% (1070) 93% (1402) 6% (84) 1% (15) 30% (455) 70% (1066) 84% (1278) 1% (8)

37% (9328)

31% (2350)

41% (7965)

40% (1978)

20% (720)

12% (137)

26% (643)

16% (235)

27% (7179) 73% (19,011) 48% (12,664) 10% (2571) 6% (1623) 0% (18) 0% (57) 11% (2878) 24% (6361) 28% (7455) 9% (2307) 19% (5061) 42%(11,204) 37% (9779) 36% (9560) 27% (7188)

27% (1923) 73% (5102) 50% (3898) 10% (737) 5% (367) 0% (1) 1% (64) 11% (881) 24% (1827) 22% (1713) 2% (115) 41% (3192) 36% (2798) 38% (2971) 36% (2799) 27% (2083)

25% (4699) 75% (14,287) 47% (8943) 11% (2064) 7% (1282) 0% (12) 0% (43) 12% (2196) 24% (4464) 31% (5890) 9% (1627) 14% (2758) 45% (8619) 35% (6773) 35% (6728) 30% (5766)

21% (874) 80% (3399) 51% (2513) 9% (446) 5% (238) 0% (0) 1% (54) 10% (491) 24% (1171) 27% (1344) 1% (69) 32% (1573) 39% (1948) 37% (1848) 35% (1717) 28% (1386)

36% (1289) 64% (2270) 54% (1923) 8% (273) 5% (186) 0% (2) 0% (6) 8% (265) 25% (877) 19% (656) 4% (136) 51% (1815) 25% (903) 42% (1510) 38% (1344) 20% (721)

39% (423) 61% (666) 45% (518) 11% (123) 5% (51) 0% (1) 1% (7) 15% (171) 24% (273) 11% (125) 2% (19) 68% (794) 19% (219) 41% (483) 37% (431) 22% (253)

33% (813) 67% (1634) 50% (1215) 6% (151) 4% (98) 0% (3) 0% (1) 10% (233) 31% (756) 23% (560) 13% (310) 13% (317) 51% (1254) 43% (1071) 42% (1027) 15% (376)

39% (566) 61% (899) 51% (768) 1% (146) 4% (63) 0% (0) 0% (1) 12% (183) 23% (344) 13% (199) 2% (24) 49% (739) 36% (551) 38% (573) 37% (564) 25% (384)

current purity of heroin and presence of fentanyl in so many areas of the US (Drug Enforcement Agency, 2017; Frank and Pollack, 2017). A variety of strategies to address disparities in treatment outcomes for people of color have been proposed in the literature, including the following: expanding services to address housing instability; increasing sensitivity to differences in levels of acculturation, English proficiency, and national origin; expanding health care coverage to improve SUD treatment access and length of stay; reducing waiting times for treatment entry; expanding drug court provision and diversion programs given the high levels of criminal justice involvement for minority clients; addressing the social isolation resulting from entering programs where the majority of clients are from different social and cultural backgrounds; and improving trauma-informed care and the needs of those with co-occurring mental health problems (Alegría et al., 2006; Amaro et al., 2006; Cooper et al., 2010; European Monitoring Centre for Drugs and Drug Addiction (EMCDDA, 2017b; Guerrero, 2013; Guerrero et al., 2013a; Vega and Lopez, 2001). Many of these programmatic recommendations may be relevant to all clients, but people of color

disadvantaged minorities, whether black, Hispanic, and Native American populations in the US; indigenous populations in Canada and Oceania; or migrants, refugees, or other minority groups in Europe, share many of the same barriers to accessing treatment, have greater unmet ancillary service needs such as unstable housing and health care problems, experience higher levels of psychosocial stressors and trauma, and may face greater cultural barriers such as lack of culturally sensitive treatment and language proficiency (Alegría et al., 2006; European Monitoring Centre for Drugs and Drug Addiction (EMCDDA, 2017b; Firestone et al., 2015; Gisev et al., 2014; Greenfield and Venner, 2012; Guerrero et al., 2013a; Jacobson et al., 2007; Wells et al., 2001; Wood et al., 2005). In the US, although whites have the highest incidence of overdose mortality, people of color have been adversely affected, and the rate of increase in opioid overdose deaths among blacks and Hispanics has recently become higher than whites (Seth et al., 2018). If minority opioid users face increased levels of wait times for accessing treatment and higher levels of leaving treatment prematurely, then this may place them at greater risk for fatal overdose given the 175

Drug and Alcohol Dependence 190 (2018) 170–178

G.J. Stahler, J. Mennis

service models involving harm reduction have long been a central tenet of European national drug response strategies (European Monitoring Centre for Drugs and Drug Addiction (EMCDDA, 2017b), and given the high rates of relapse, treatment non-completion, and mortality for opioid users in the US, creative and non-traditional solutions may be needed that would not necessarily have been considered previously.

may particularly benefit from these enhanced services given the differential vulnerabilities and barriers they more frequently experience. 4.3. The importance, complexity, and limitations of using treatment completion as an outcome in an era of continuing care Treatment completion may be a more conceptually complex outcome indicator for OUDs not only for those receiving long term MAT involving methadone and buprenorphine, but also because of the importance of viewing treatment for any SUD as a process of long term recovery and continuing care (Dennis and Scott, 2007; Flynn et al., 2016; McKay, 2009; White and Sanders, 2008; White, 2008). Although treatment completion has been shown to be a robust predictor of positive post-treatment outcomes (Brorson et al., 2013), our focus on the initial treatment episode has to be viewed as representing a narrow slice of the initial recovery process since successful long-term recovery may require multiple treatment episodes over time, particularly for OUDs (Guerrero, 2013; Vogel et al., 2017). For MAT, the primary clinical goal may not be “treatment completion” as in completing a single episode of treatment, but rather continuing retention and long-term maintenance (Eastwood et al., 2017; Korner and Wall, 2005; Magura and Rosenblum, 2001). While there have been recent increases in the availability of MAT (Pouget et al., 2017), one recent study found only about a quarter of treatment admissions for OUD in the US receive it (Krawczyk et al., 2017b). In comparison, within the European Union countries, about 50% of opioid users in treatment receive some form of opioid substitution treatment, though national estimates vary between 10% and 80% (European Monitoring Centre for Drugs and Drug Addiction (EMCDDA, 2017b). Given the research showing the benefits of MAT for OUD, this represents a challenge for policymakers. While there should be a greater availability for access to MAT in outpatient settings, we should also not lose sight of the need for greater access and funding for long term residential treatment for those with OUD. In the present study, treatment completion for residential settings was considerably higher compared to outpatient programs (41% vs. 21%). To assess whether the ethnic/racial disparities in treatment completion were different for outpatient and residential settings, we reran the fixed effects logistic regression displayed in Table 2 separately for these two settings. The results were similar for both settings (residential: black OR = 0.83; Hispanic OR = 0.76; outpatient: black OR = 0.79; Hispanic OR = 0.84), but the interaction of treatment setting with race and ethnicity in treatment outcomes for opioid users remains a promising area of future research, particularly since black and Hispanic users are less likely to receive residential treatment (Bluthenthal et al., 2007). Residential treatment may provide greater protection from environmental and social triggers that may lead to relapse and early departure from treatment (Stahler et al., 2016), especially when combined with MAT where appropriate and strong posttreatment linkages to ongoing continuing care (European Monitoring Centre for Drugs and Drug Addiction (EMCDDA, 2017b; Schuman-Oliver et al., 2014).

4.5. Limitations This study has a number of limitations. The analyses were based on a national administrative dataset, the TEDS-D, which is compiled by states from individual programs, and therefore is subject to inconsistent survey response reporting. It is possible that variations across MSAs could be an artifact of inconsistent reporting. Since the analyses were completed, a more recent version of the dataset (2014) is now available, and the generalizability of our results to more recent data is unknown. Definitions of “treatment completion” are made at the local program level and may thus vary across programs, and it was not possible to incorporate program level data into the analyses. Consistent with previous research utilizing the TEDS-D dataset (Arndt et al., 2013; Krawczyk et al., 2017b; Sahker et al., 2015), our sample was limited to clients with no prior treatment admissions to prevent multiple cases being associated with a single individual. This limits the generalizability of the study to clients entering their first treatment program and may bias the representativeness of the sample in the direction of a younger, more treatment naïve, and potentially less chronic population. This bias may result in a reduced likelihood of treatment completion for this sample. Additional limitations are that we were not able to assess other important post-treatment outcomes such as relapse because these data are not in the dataset, and we were not able to utilize variables in the dataset relating to co-occurring mental health problems and insurance coverage that prior research suggests are related to treatment completion (Guerrero et al., 2017; Krawczyk el al., 2017a) due to a high occurrence of missing values. Our analyses were also limited to large metropolitan areas and therefore may not generalize to smaller urban or rural areas. Finally, clients who are successfully engaged in long term MAT, even though this is generally considered a positive proximal treatment outcome, are not included in the dataset since they were not “discharged” from a program. 5. Conclusion As the opioid overdose crisis continues to impact people across the ethnic and racial spectrum in the US, it is important to better understand disparities in treatment outcomes. While large scale national data sets serve an important function in providing public health policymakers with information about national trends, our results suggest that there are important geographic variations in patterns of SUD treatment outcomes. In treatment systems with high levels of ethnic/racial outcome disparities in the US and globally, it is important to develop culturally relevant and tailored interventions for minority clients to best meet their needs within that particular area (Alegría et al., 2006; Cooper et al., 2010; Guerrero et al., 2013a; Velez et al., 2008). Examining the interaction of client and program factors, and using multiple methodologies, may help to better understand why clients drop out or complete treatment. Given the risk of overdose for those who leave treatment prematurely, new models for collaboration between treatment programs and low-demand harm reduction programs in the US based on a continuing care model may represent life-saving opportunities for future treatment re-entry and re-engagement.

4.4. The urgent need to improve treatment engagement and retention for opioid users One of our major findings is that only 28% of the total sample from large metropolitan areas completed treatment, which is consistent with prior research (Stahler et al., 2016) but also concerning. For opioid users, regardless of modality, more attention needs to be devoted toward creative treatment retention and re-engagement strategies as well as increasing the distribution of Naloxone and the use of Vivitrol for clients leaving treatment. For those who leave treatment prematurely, it may also be advantageous for treatment programs to develop new models of collaboration with low-demand harm reduction programs which may represent life-saving opportunities for future treatment reentry and engagement. Evidence-based extended continuum of care

Role of the funding source Nothing declared. 176

Drug and Alcohol Dependence 190 (2018) 170–178

G.J. Stahler, J. Mennis

Contributors

substitution therapy treatment utilisation among opioid-dependent people in contact with the criminal justice system: comparison of indigenous and non-indigenous australians. BMC Public Health 14, 920. Greenfield, B.L., Venner, K.L., 2012. Review of substance use disorder treatment research in Indian country: future directions to strive toward health equity. Am. J. Drug Alcohol Abuse 38, 483–492. Guerrero, E.G., 2013. Examination of treatment episodes among women and racial and ethnic minorities in addiction treatment. J. Soc. Work Pract. Addict. 13, 227–243. Guerrero, E.G., Marsh, J.C., Khachikian, T., Amaro, H., Vega, W., 2013a. Disparities in Latino substance use, service use, and treatment: implications for culturally and evidence-based interventions under health care reform. Drug Alcohol Depend. 133, 805–813. Guerrero, E.G., Marsh, J.C., Duan, L., Oh, C., Perron, B., Lee, B., 2013b. Disparities in completion of substance abuse treatment between and within racial and ethnic groups. Health Serv. Res. 48, 1450–1467. Guerrero, E.G., Garner, B.R., Cook, B., Kong, Y., Vega, W.A., Gelberg, L., 2017. Identifying and reducing disparities in successful addiction treatment completion: testing the role of medicaid payment acceptance. Subst. Abuse Treat. Prev. Policy 12, 27. Hand, D.J., Short, V.L., Abatemarco, D.J., 2017. Substance use, treatment, and demographic characteristics of pregnant women entering treatment for opioid use disorder differ by United States Census region. J. Subst. Abuse Treat. 76, 58–63. Jacobson, J.O., Robinson, P., Bluthenthal, R.N., 2007. Racial disparities in completion rates from publicly funded alcohol treatment: economic resources explain more than demographics and addiction severity. Health Serv. Res. 42, 773–794. Kornør, H., Waal, H., 2005. From opioid maintenance to abstinence: a literature review. Drug Alcohol Rev. 24, 267–274. Krawczyk, N., Feder, K.A., Saloner, B., Crum, R.M., Kealhofer, M., Mojtabai, R., 2017a. The association of psychiatric comorbidity with treatment completion among clients admitted to substance use treatment programs in a U.S. national sample. Drug Alcohol Depend. 175, 157–163. Krawczyk, N., Feder, K.A., Fingerhood, M.I., Saloner, B., 2017b. Racial and ethnic differences in opioid agonist treatment for opioid use disorder in a U.S. national sample. Drug Alcohol Depend. 178, 512–518. https://doi.org/10.1016/j.drugalcdep.2017.06. 009. Magura, S., Rosenblum, A., 2001. Leaving methadone treatment: lessons learned, lessons forgotten, lessons ignored. Mt. Sinai J. Med. 68, 62–74. Martins, S.S., Sarvet, A., Santaella-Tenorio, J., Saha, T., Grant, B.F., Hasin, D.S., 2017. Changes in US lifetime heroin use and heroin use disorder prevalence from the 20012002 to 2012-2013 National Epidemiologic Survey on alcohol and related conditions. JAMA Psychiatry 74, 445–455. McKay, J.R., 2009. Continuing care research: what we have learned and where we are going. J. Subst. Abuse Treat. 36, 131–145. McKay, J.R., Weiss, R.V., 2001. A review of temporal effects and outcome predictors in substance abuse treatment studies with long-term follow-ups: preliminary results and methodological issues. Eval. Rev. 25, 113–161. Mennis, J., Stahler, G.J., 2016. Racial and ethnic disparities in outpatient substance use disorder treatment episode completion for different substances. J. Subst. Abuse Treat. 63, 25–33. https://doi.org/10.1016/j.jsat.2015.12.007. Mennis, J., Stahler, G.J., Baron, D., 2012. Geographic barriers to community-based psychiatric treatment for drug dependent patients. Ann. Assoc. Am. Geogr. 102, 1093–1103. Morley, K., Logge, W., Pearson, S., Baillie, A., Haber, P., 2017. Socioeconomic and geographic disparities in access to pharmacotherapy for alcohol dependence. J. Subst. Abuse Treat. 74, 23–25. Murphy, K., Becker, M., Locke, J., Kelleher, C., McLeod, J., Isasi, F., 2016. Finding Solutions to the Prescription Opioid and Heroin Crisis: A Road Map for States. [WWW Document]. https://www.nga.org/files/live/sites/NGA/files/pdf/2016/ 1607NGAOpioidRoadMap.pdf. National Institute on Drug Abuse, 2018. Principles of Drug Addiction Treatment: A Research-Based Guide. [WWW Document]. third ed. . https://d14rmgtrwzf5a. cloudfront.net/sites/default/files/675-principles-of-drug-addiction-treatment-aresearch-based-guide-third-edition.pdf. National Research Council, 2003. In: Smedley, B.D., Stith, A.Y., Nelson, A.R. (Eds.), Unequal Treatment: Confronting Racial and Ethnic Disparities in Health Care. The National Academies Press, Washington, DC. Pouget, E.R., Fong, C., Rosenblum, A., 2017. Racial/ethnic differences in prevalence trends for heroin use and non-medical use of prescription opioids among entrants to opioid treatment programs, 2005–2016. Subst. Use Misuse 53, 290–300. Rigg, K.K., Monnat, S.M., 2015. Urban vs. rural differences in prescription opioid misuse among adults in the United States: informing region specific drug policies and interventions. Int. J. Drug Policy 26, 484–491. Roberts, E.T., Friedman, S.R., Brady, J.E., Pouget, E.R., Tempalski, B., Galea, S., 2010. Environmental conditions, political economy, and rates of injection drug use in large US metropolitan areas 1992–2002. Drug Alcohol Depend. 106, 142–153. Rossen, L.M., Khan, D., Warner, M., 2014. Hot spots in mortality from drug poisoning in the United States, 2007–2009. Health Place 26, 14–20. Sahker, E., Toussaint, M.N., Ramirez, M., Ali, S.R., Arndt, S., 2015. Evaluating racial disparity in referral source and successful completion of substance abuse treatment. Addict. Behav. 48, 25–29. Saloner, B., Le Cook, B.L., 2013. Blacks and Hispanics are less likely than whites to complete addiction treatment, largely due to socioeconomic factors. Health Aff. 32, 135–145. https://doi.org/10.1377/hlthaff.2011.0983. Saloner, B., Carson, N., Cook, B.L., 2014. Explaining racial/ethnic differences in adolescent substance abuse treatment completion in the United States: a decomposition analysis. J. Adolesc. Health 54, 646–653. https://doi.org/10.1016/j.jadohealth. 2014.01.002.

Both authors contributed to the conception and design of the study. Stahler drafted the Introduction and Discussion sections, and parts of the Methods and Results sections. Mennis developed the analytic plan, was responsible for analyzing the data, and drafted parts of the Methods and Results sections. Both authors read and approved the final manuscript. Conflict of interest No conflict declared. Acknowledgements The authors wish to thank Andrew Rosenblum, Noa Krawczyk, and Stephen Magura for sharing pre-publication drafts of their own papers and helpful discussion related to the findings of this study. References Alegría, M., Page, J.B., Hansen, H., Cauce, A.M., Robles, R., Blanco, C., Cortes, D.E., Amaro, H., Morales, A., Berry, P., 2006. Improving drug treatment services for Hispanics: research gaps and scientific opportunities. Drug Alcohol Depend. 84, 76–84. Alegría, M., Carson, N.J., Goncalves, M., Keefe, K., 2011. Disparities in treatment for substance use disorders and co-occurring disorders for ethnic/racial minority youth. J. Am. Acad. Child Adolesc. Psychiatry 50, 22–31. Amaro, H., Arévalo, S., Gonzalez, G., Szapocznik, J., Iguchi, M.Y., 2006. Needs and scientific opportunities for research on substance abuse treatment among Hispanic adults. Drug Alcohol Depend. 84S, S64–S75. Arndt, S., Acion, L., White, K., 2013. How the states stack up: Disparities in substance abuse outpatient treatment completion rates for minorities. Drug Alcohol Depend. 132, 547–554. Brorson, H.H., Arnevik, E.A., Rand-Hendriksen, K., Duckert, F., 2013. Drop-out from addiction treatment: a systematic review of risk factors. Clin. Psychol. Rev. 33, 1010–1024. Cooper, R.L., MacMaster, S., Rasch, R., 2010. Racial differences in retention in residential substance abuse treatment: the impact on African American men. Res. Soc. Work Pract. 20, 183–190. Cummings, J.R., Wen, H., Druss, B.G., 2011. Racial/ethnic differences in treatment for substance use disorders among U.S. adolescents. J. Am. Acad. Child Adolesc. Psychiatry 50, 1265–1274. Cummings, J.R., Wen, H., Ko, M., Druss, B.G., 2014. Race/ethnicity and geographic access to medicaid substance use disorder treatment facilities in the United States. JAMA Psychiatry 71, 190–196. Dennis, M.L., Scott, C., 2007. Managing addiction as a chronic condition. Addict. Sci. Clin. Pract. 4, 45–55. Drug Enforcement Administration (DEA), 2017. 2017 National Drug Threat Assessment Summary. DEA-DCT-DIR-040-17. [WWW Document]. (accessed on 05.27.18). https://www.dea.gov/docs/DIR-040-17_2017-NDTA.pdf. Eastwood, B., Strang, J., Marsden, J., 2017. Effectiveness of treatment for opioid use disorder: a national, five-year, prospective, observational study in England. Drug Alcohol Depend. 176, 139–147. European Monitoring Centre for Drugs and Drug Addiction (EMCDDA), 2015. Drugs Policy and the City in Europe. [WWW Document. (accessed on 5.31.18). http:// www.emcdda.europa.eu/system/files/publications/976/TDAU15001ENN1.pdf. European Monitoring Centre for Drugs and Drug Addiction (EMCDDA), 2017a. European Drug Report: Trends and Developments. [WWW Document]. (accessed 5.31.18). http://www.emcdda.europa.eu/system/files/publications/976/TDAU15001ENN1. pdf. European Monitoring Centre for Drugs and Drug Addiction (EMCDDA), 2017b. Health and Social Responses to Drug Problems: A European Guide. [WWW Document]. (accessed on 5.31.18). http://www.emcdda.europa.eu/publications/manuals/ health-and-social-responses-to-drug-problems-a-european-guide_en. Evans, A., Farley, T., 2017. The Mayor’s Task Force to Combat the Opioid Epidemic in Philadelphia: Final Report and Recommendations. [WWW Document]. . http:// dbhids.org/wp-content/uploads/2017/05/OTF_Report.pdf. Firestone, M., Tyndall, M., Fischer, B., 2015. Substance use and related harms among Aboriginal people in Canada: a comprehensive review. J. Health Care Poor Underserved 26, 1110–1131. Frank, R.G., Pollack, H.A., 2017. Addressing the fentanyl threat to public health. N. Engl. J. Med. 376, 605–607. Franklin, G., Sabel, J., Jones, C.M., Mai, J., Baumgartner, C., Banta-Green, C.J., Neven, D., Tauben, D.J., 2015. A comprehensive approach to address the prescription opioid epidemic in Washington State: milestones and lessons learned. Am. J. Public Health 105, 463–469. Gisev, N., Gibson, A., Larney, S., Kimber, J., Williams, M., Clifford, A., Doyle, M., Burns, L., Butler, T., Weatherburn, D.J., Degenhardt, L., 2014. Offending, custody and opioid

177

Drug and Alcohol Dependence 190 (2018) 170–178

G.J. Stahler, J. Mennis

NSDUH Series H-52). Center for Behavioral Health Statistics and Quality, Substance Abuse and Mental Health Services Administration, Rockville, MD. United Nations Office on Drugs and Crime (UNODC), 2017. World Drug Report 2017, Booklet 2: Global Overview of Drug Demand and Supply, Latest Trends, Cross-Cutting Issues. [WWW Document]. . http://www.unodc.org/wdr2017/field/Booklet_2_ HEALTH.pdf. Vega, W.A., Lopez, S.R., 2001. Priority issues in Latino mental health services research. Ment. Health Serv. Res. 3, 189–200. Veléz, M.B., Campos-Holland, A.L., Arndt, S., 2008. City’s racial composition shapes treatment center characteristics and services. J. Ethn. Subst. Abuse 7, 188–199. Vogel, M., Dürsteler, K., Walter, M., Herdener, M., Nordt, C., 2017. Rethinking retention in treatment of opioid dependence—The eye of the beholder. Int. J. Drug Policy 39, 109–111. Wang, K.H., Becker, W.C., Fiellin, D.A., 2013. Prevalence and correlates for nonmedical use of prescription opioids among urban and rural residents. Drug Alcohol Depend. 127, 156–162. Wells, K., Klap, R., Koike, A., Sherbourne, C., 2001. Ethnic disparities in unmet needs for alcoholism, drug abuse, and mental health care. Am. J. Psychiatry 158, 2027–2032. Wood, E., Li, K., Palepu, A., Marsh, D.C., Schechter, M.T., Hogg, R.S., Montaner, J.G., Kerr, T., 2005. Sociodemographic disparities in access to addiction treatment among a cohort of Vancouver injection drug users. Subst. Use Misuse 40, 1153–1167.

Schneider, E., 2008. Smack: Heroin and the American City (Politics and Culture in Modern America). University of Pennsylvania Press, Philadelphia. Seth, P., Scholl, L., Rudd, R.A., Bacon, S., 2018. Overdose deaths involving opioids, cocaine, and psychostimulants — United States, 2015–2016. MMWR Morb. Mortal. Wkly. Rep. 67, 349–358. Stahler, G., Mazzella, S., Mennis, J., Chakravorty, S., Rengert, G., Spiga, R., 2007. The effect of individual, program, and neighborhood variables on continuity of treatment among dually diagnosed individuals. Drug Alcohol Depend. 87, 54–62. Stahler, G., Mennis, J., Cotlar, R., Baron, D., 2009. The influence of the neighborhood environment on treatment continuity and rehospitalization for dually diagnosed patients discharged from acute inpatient care. Am. J. Psychiatry 166, 1258–1268. Stahler, G.J., Mennis, J., DuCette, J.P., 2016. Residential and outpatient treatment completion for substance use disorders in the U.S.: Moderation analysis by demographics and drug of choice. Addict. Behav. 58, 129–135. Substance Abuse and Mental Health Services Administration (SAMHSA), 2015. Center for Behavioral Health Statistics and Quality. Treatment Episode Data Set (TEDS): 2014. National Admissions to Substance Abuse Treatment Services. Rockville, MD: Substance Abuse and Mental Health Services Administration (BHSIS Series S-65, HHS Publication No. (SMA) 13-4772). Substance Abuse and Mental Health Services Administration (SAMHSA), 2017. Key Substance Use and Mental Health Indicators in the United States: Results from the 2016 National Survey on Drug Use and Health (HHS Publication No. SMA 17-5044,

178