Treatment Retention Satisfaction, and Therapeutic Progress for Justice-Involved Individuals Referred to Community-Based Medication-Assisted Treatment

Abstract

Background:

Limited research has examined factors associated with medication-assisted treatment for justice-involved individuals.

Objectives:

The current study used a mixed-method design to examine the influence of client- and counselor-level factors on 90-day treatment retention, satisfaction, and progress for justice-involved individuals referred to medication-assisted treatment.

Methods:

The effects of co-occurring disorders (i.e., psychiatric symptoms, anxiety, depression), social functioning (i.e., social support, self-esteem), substance use severity, and treatment motivation on treatment retention, treatment satisfaction, and treatment progress while controlling for counselor-level variance were assessed through multilevel modeling.

Results:

Fewer co-occurring disorders and more social support were related to greater treatment satisfaction and progress. A higher level of treatment motivation was associated with greater treatment progress. Mediation of treatment satisfaction on the relationship between client-level factors and treatment progress also was tested. Depression was negatively associated with treatment satisfaction, which in turn led to lower ratings of treatment progress. Social support was positively correlated with treatment satisfaction, which in turn was positively correlated with treatment progress. The association of client substance use severity with treatment retention differed between counselors, so did the association of co-occurring disorders and treatment motivation with treatment satisfaction. Qualitative analyses that were derived from counselors’ perception of factors relating to recovery success underscored the importance of integrated interventions, social support, treatment motivation, and therapeutic alliance, and their associations with treatment outcomes.

Conclusions/Importance:

The current findings highlight the importance of integrated treatment services, collaborating with community corrections, and teaching clients strategies for dealing with deviant peers as to facilitating recovery.

Introduction

After release, many prisoners with addiction problems may receive community-based aftercare services designed to address risk for drug relapse and provide support for a drug-free living environment in the community (Brown et al., 2001). Medication-assisted treatment (MAT) has been shown to reduce drug use, prolong abstinence, increase treatment completion and retention, and achieve favorable legal outcomes (Bisaga, 2011; Coviello et al., 2012; Gavin, 2012). For those who achieved abstinence during incarceration and may be at risk of relapse and overdose after release, it would be appropriate to include MAT during community supervision, as an aid in building engagement in a long-term recovery process.

Client-level characteristics

A critical component for optimizing treatment is the provision of matching services to identified problem domains (Friedmann, Hendrickson, Gerstein, & Zhang, 2004; Grella & Stein, 2006; Guerrero, Garner, Cook, Kong, Vega, & Gelberg, 2017; McLellan, et al., 1997). This strategy points to the need for identifying treatment needs and examining the association of these needs with treatment process and outcomes. Building upon the literature on substance abuse treatment in general, the current study specifically examined the impact of client-level characteristics, including mental health symptoms, social functioning, substance use severity, and treatment motivation on treatment outcomes in a MAT program. In general, more mental health problems have been associated with a lower likelihood of treatment completion (Guerrero et al., 2017; Petry & Bickel, 1999), higher rates of dropout, substance-use relapse, and a greater urge of substance use in high-risk situations (Bell, Atkinson, Williams, Nelson, & Spence, 1996; Brown et al., 2001).

Among positive factors associated with favorable treatment outcomes is social support (see a meta-analysis review, Hogan, Linden, & Najarian, 2002). Interventions using social network strategies help increase treatment attendance and enhance treatment retention for people with substance use problems (Landau et al., 2000; Soyez, De Leon, Broekaert, & Rosseel, 2006). Social networks comprised of people encouraging recovery-related behaviors and showing favorable attitudes toward recovery predict a greater intention to change drug use behaviors (Matto, Miller, & Spera, 2007). While lower self-esteem has been implicated in the development of substance use problems (e.g., Khantzian, 1997; Mariani, Khantzian, & Levin, 2014; Ojo, Akintoyese, Adenibuyan, Adegbohun, & Abiri, 2013), mixed results have been reported for the relationship between self-esteem and treatment outcomes. For example, a study using alcohol-dependent participants from an inpatient program found that self-esteem was not associated with post-treatment drinking outcomes (Trucco, Connery, Griffin, & Greenfield, 2007). Among participants in prison-based therapeutic communities, high self-esteem was associated with post-treatment re-arrest for men, whereas low self-esteem predicted post-treatment re-arrest for women (Yang et al., 2015). Further, Kiecolt (1994) argued that clients with lower self-esteem may seek help and initiate the pursuit for change if they hit rock bottom and substance use no longer suppresses anxiety or supports self-identity and self-esteem. With relatively greater attention being paid to women, research is still needed to look at the impact of self-esteem on male offenders.

Greater substance use severity at treatment entry is an important factor as it is predictive of shorter stays in treatment (Guerrero et al., 2017; Lang & Belenko, 2000; Marrero et al., 2005). However, research also indicates the positive effect of longer treatment retention for those with greater substance use severity when they receive more service intensity and have greater treatment satisfaction (Hser, Evans, Huang, & Anglin, 2004). These authors suggested that counselors may tend to pay additional attention to clients with more severe substance use problems and prioritize their service needs, which in turn may increase client motivation and attendance. This highlights the importance of a positive treatment experience.

Treatment motivation has been associated with treatment participation, completion, and outcomes among clients from general substance use treatment settings and criminal justice systems (De Leon, Melnick, & Tims, 2001; Simpson & Joe, 1993). Joe, Simpson, and Broome (1999) found that pretreatment motivation significantly predicted treatment engagement and retention among individuals from 37 community-based treatment programs; treatment readiness was more robust than demographic, substance use, criminality, and other pretreatment variables. Likewise, Prendergast, Greenwell, Farabee, and Hser (2009) reported that motivation predicted a low likelihood of 12-month post-treatment arrest for substance-abusing offenders recruited from 38 treatment assessment sites.

The aforementioned factors allude to the importance of clients’ treatment experiences. Satisfaction with services, program convenience, better counseling relationships, and self-reported treatment effectiveness are related to treatment retention, completion, drug use abstinence, and less criminal activity at 1-year follow-up (Carlson & Gabriel, 2001; Hser et al., 2004; Kelly, O’Grady, Mitchell, Brown, & Schwartz, 2011). Clients who perceive the program to be less accessible and structured are more likely to drop out from methadone maintenance treatment (Joe, Simpson, & Hubbard, 1991). It is believed that positive MAT treatment experience (e.g., treatment satisfaction) may serve as a precursor to favorable treatment outcomes and contributes to the association between client-level factors and treatment outcomes, having important implications for treatment planning.

Counselors’ perceptions of client recovery

Although few studies have been done, insights into counselor perceptions could be helpful in understanding clients’ recovery processes. Long, Midgley, and Hollin (1997) found that staff perceptions predicted 1-year post-treatment drinking status and psychological distress, whereas peer perceptions did not. Walton, Blow, and Booth (2000) compared the role of client to counselor perceptions in predicting treatment outcomes and found that counselor ratings of their client’s coping skills predicted alcohol relapse but not substance use relapse, whereas client’s ratings of their coping skills, leisure activities, and social support predicted both alcohol and substance use relapse. Palmer, Murphy, Piselli, and Ball (2009) used a mixed-methods study with 22 clients and 22 counselors in outpatient programs and showed that clinicians tended to attribute treatment dropout to client-level factors (e.g., a lack of motivation and support from family, physical or mental health issues) more than did clients; focus group data from clients and clinicians indicated that a lack of working alliance with staff and a lack of motivation were major reasons for dropout. Those studies highlight the value of counselor perspectives and suggest the need for drawing on perspectives from multiple agents involved in the recovery process (Flynn, Knight, Godley, & Knudsen, 2012). With so much having been done with client samples, more research with advanced methodologies is needed to integrate counselor opinions in exploring the complexity of treatment process.

This observational study includes a convenience sample of justice-involved individuals in community-based MAT treatment. It used a naturalistic design intended to further our understanding of specific factors associated with treatment process and outcomes (i.e., treatment retention, treatment satisfaction, and therapeutic progress), shedding light on strategies for the optimum delivery of MAT treatment services. The current study was unique because (1) it used a mixed-methods design wherein a qualitative exploration of counselor perspective augmented the understanding of client experience, and (2) the sample included clients referred to MAT and their counselors. This study had three goals (see Figure 1). First was an examination of the influences of client mental health symptoms, social functioning, substance use severity, and treatment motivation on treatment retention, satisfaction and progress. It was hypothesized that, when compared to their counterparts, clients with fewer mental health symptoms, lower ratings of anxiety and depression, higher levels of social support, higher ratings of self-esteem, less drug use severity, and higher levels of treatment motivation will be more likely to remain in treatment and have greater treatment satisfaction and progress. The second goal investigated treatment satisfaction as a mediator of client characteristics on treatment progress. It addressed whether clients with fewer mental health symptoms, lower anxiety, lower depression, a higher level of social support, higher self-esteem, less substance use severity, and higher motivation would report higher ratings of treatment satisfaction which in turn would be related to an increased rating of treatment progress. Third, this study utilized qualitative research methods to investigate counselors’ perceptions of barriers and facilitators to recovery process, which encompasses the period between treatment intake and discharge. Qualitative analyses were used to help provide a more in-depth understanding of the counselor-level factors associated with client recovery.

Figure 1.

Conceptual model of the study goals.

Methods

Procedures

An embedded design was used to guide the research (Creswell, 2013) in which the qualitative component was embedded within a major quantitative design of evaluating treatment process and outcome. Quantitative client data were collected as part of an administrative supplement (Desmarais et al., 2016; Gray, Desmarais, Cohn, Doherty, & Knight, 2015) to the national Criminal Justice Drug Abuse Treatment Studies (CJDATS-II) protocol titled Medication Assisted Treatment Implementation in Community Correctional Environments (MATTICE; Friedmann et al., 2012). Participants (N = 90) were male offenders who were referred to a community-based MAT treatment program in a large American Midwest city that provided options for buprenorphine and depot naltrexone. In addition to the medication, clients visited the treatment facility three to five times per week for a combination of group and individual counseling.

A graduate research (RA) assistant at the treatment clinic approached individuals who had received MAT within the two previous weeks. Participants who agreed to participate in the study were provided information about the study and ask to provide informed consent. The informed consent procedures were conducted by the RA prior to the completion of a battery of assessments at the treatment clinic that lasted about one hour. Only one participant declined to continue study participation after a baseline interview; thus their data were removed from the dataset. A one-hour post-treatment survey was administered onsite by an RA 90 days after MAT initiation. Seventy-one percent (n = 64) completed the follow-up. The Texas Christian University (TCU) Institutional Review Board approved the participant information tracking, informed consent, pre- and post-treatment surveys, and the interview guide. Qualitative data were collected by semi-structured interviews focusing on counselor’s self-reported factors that impact the client recovery processes. All seven counselors had referred clients to MAT services during the study. Six interviews were conducted at the treatment clinic; one interview was conducted over the phone. Six interviews were conducted by a Ph.D. candidate (at that time) and the remaining one by a Ph.D. researcher. The interviews took 20 minutes on average (range =13–48).

Participants

Clients

The sample included both clients and substance use counselors from a community-based drug treatment program in a large Midwestern metropolitan area. The client participants included 90 male offenders (treated by 12 counselors; 99% were on probation or parole at the time of recruitment) attending a community-based substance abuse treatment program wherein MAT was integrated in the treatment plan. The majority of participants were African American (79%; see Table 1) and the average age was 36 years (SD = 10). Seventy percent of participants identified heroin as the most serious problem, followed by alcohol (6%) and the mixture of heroin and cocaine (5%); the remaining reported other opiates, street methadone, or hallucinogens alone as the most serious problem. The majority of the clients received outpatient or intensive-outpatient services (see Table 1), wherein clients received one-on-one sessions with a primary counselor and attended three to five group sessions per week.

Table 1.

Clients demographics and characteristics.

	Baseline (N = 90) Frequency (%)	Treatment retention (n = 64) Frequency (%)	Dropout (n = 26)a Frequency (%)
Race
African American	79	78	81
White	17	18	15
Other	4	3	4
Never married	78	80	73
High school diploma or GED	61	59	65
Heroin use frequency in the latest 6 months before treatment
None	30	37	21
1–3 times per month	22	23	18
Once or more per week	48	40	61
Current treatment service type
Intensive Outpatient	30	46	27
Outpatient	70	54	73
	Mean (SD/range)	Mean (SD/range)	Mean (SD/range)
Previous treatment episodes	3.25 (1.46)	2.97 (1.38)	3.73 (1.48)
Criminal history
Total number of arrests	18 (20.24/1–150)	18 (21.13/1–150)	20 (18.68/2–85)
Total number of drug-related arrests	14 (19.71/0–150/)	14 (20.60/0–150)	14 (18.14/0–85)
Age	36 (10.38)	36 (10.51)	35 (9.67)
Duration of any illicit drug use/heroin use (yr)	22.44(10.71)/717.00(12.28)	22.69(10.69)/18.39(12.98)	21.85(10.92)/13.73(9.92)

χ² was used for percentages and t-tests for means when making comparisons between those retained in treatment and those who dropped out.

^aAmong these participants, 14 had been re-arrested and another was a murder victim. The remainder were discharged for non-adherence, absconding, or were referred to a higher level of treatment.

^bClients who retained in treatment reported fewer previous substance treatment episodes than those who dropped out, t = −2.01, p = .025.

Counselors

Although all of the substance abuse treatment counselors for these 90 MAT clients were originally targeted for inclusion in the study, 4 of the 12 had left their employment at the facility before the qualitative data collection began. Seven counselors (average age = 47) completed the interviews, and one counselor declined to participate. Four were African American and three were white. Two counselors had Master’s degree, two had Bachelor’s degree, and the remaining held Associate degrees. All had substance abuse counseling credentials and on average, they each had 13 years of experience delivering treatment (SD = 13; range = 1–28). Cognitive-behavioral therapy and 12-step therapy (which focuses on facilitating the involvement in 12-step self-help groups conducted at the clinic and in the community) were the most frequently used counseling approaches by these staff.

Measures

Client-level baseline measures

Co-occurring mental health symptoms and social functioning was measured using self-reported psychiatric symptoms, depression, anxiety, self-esteem, and social support at baseline. Psychiatric symptoms were rated with 11 items adapted from the Brief Psychiatric Rating Scale (e.g., “During the past week, how much were you bothered by feeling low in energy or slowed down?”; Cronbach’s α = 0.78; Andersen, Larsen, Kørner, & Bjørum, 2009). The adapted Brief Psychiatric Rating Scale used a 7-point Likert scale (1 = “Disagree Strongly” to 7 = “Agree Strongly”; α = 0.75 in the current study). The TCU Psychological Functioning (PSYFORM) and Social Functioning (SOCFORM) assessments (Cronbach’s α ≥ 0.75; Simpson, Joe, Knight, Rowan-Szal, & Gray, 2012) were used to measure depression (e.g., “You feel sad or depressed,” α = 0.77), anxiety (e.g., “You have trouble sitting still for long;” α = 0.75), self-esteem (e.g., “You have much to be proud of;” α = 0.73), and social support (e.g., “You have people close to you who can always be trusted;” α = 0.72).

Treatment motivation was measured using the TCU Motivation (MOTFORM) scales (α ≥ 0.81; Simpson et al., 2012). A sample item from the MOTFORM is: “Your drug use is a problem for you.” The TCU PSYFORM, TCU SOCFORM, and TCU MOTFORM utilized a 5-point Likert scale (1 = “Disagree Strongly” to 5 = “Agree Strongly;” α = 0.82 in the current study).

Substance use severity refers to client self-reported problem severity pertaining to drug use, measured by the TCU Drug Screen II (α = 0.89; Knight, Simpson, & Morey, 2002; sample item—“Did your drug use cause emotional or psychological problems?). Substance use severity items used a dichotomous scale (0 = no, 1 = yes; α = 0.80 for the current sample). The composite score of each scale was used in data analysis.

Client-level follow-up assessment

Treatment retention was defined as whether or not clients were retained for the 90 days following MAT referral (0 = no, 1 = yes). Treatment satisfaction items were adapted from the Mental Health Statistics Improvement Project Survey (MHSIP; Center for Mental Health Services, 2004; α = 0.87 for the current sample). These client-rated items asked about general treatment satisfaction (two items; e.g., “I like the services that I have received in the past three months”), access to services (two items; e.g., “Staff where I received services were willing to see me as often as I felt it was needed”), and perception of counseling quality and appropriateness (four items; e.g., “Staff where I received services encouraged me to take responsibility for how I live my life”). Treatment progress measured how much improvement clients perceived after 90 days of treatment. It was measured by six MHSIP-adapted items for substance-use treatment (e.g., “I deal more effectively with daily problems”; α = 0.74 for the current sample). The subscales in the MHSIP have good internal reliability (α ≥ 0.81, Eisen, Shaul, Leff, Stringfellow, Clariridge, & Cleary, 2001). All items for treatment satisfaction and progress utilize a 5-point Likert scale ranging from “Disagree Strongly” to “Agree Strongly.”

Counselor-rated factors related to client recovery

The qualitative section of this study used a semistructured interview to investigate counselor perceptions of barriers and facilitators to treatment satisfaction, retention, and progress. Initial questions for the interview were developed using literature-based factors relevant to client treatment success and then revised after the authors conducted mock interviews. Appropriate probing questions were added for clarification. Interviewers had prior qualitative research experience derived from a large-scale, multisite qualitative project. All the interviews were recorded and transcribed verbatim.

Data analysis

Quantitative data analysis

This study used multilevel modeling, with clients nested under counselors, to examine the impact of client-level factors on outcome variables. Because of the concern for multicollinearity and suppressor effects, the analyses of the multilevel models were performed separately in four domains: (1) co-occurring mental health symptoms, (2) social functioning, (3) substance use severity, and (4) motivation. Because of the high correlations among the variables and concern for multicollinearity issues and suppressor effects; for the same reason, the impacts of three variables of co-occurring mental health symptoms were also examined in separate multilevel models. The results of preliminary power analyses (Faul, Erdfelder, Buchner, & Lang, 2009) indicated sample sizes ranging from 52 to 90 participants would yield a desirable power of 0.80 with a small-to-medium effect size (the effect size of the significant relationships in Tables 3 and 4, R² = 0.07–0.15) on the hypothesized relationships. In addition, multilevel modeling was used to examine the mediation of treatment satisfaction on the association of pretreatment characteristics to treatment progress with the procedures proposed by Krull and MacKinnon (1999), which have been widely used in analyzing multi-level mediation. The current study used Sobel’s test and a bootstrapping procedure including a total of 1000 bootstrap samples for testing indirect effects. The non-parametric bootstrapping also works very well with a moderately sized sample (Ong, 2014). Unequal sample size within each level 2 unit (e.g., unequal number of clients nested in each counselor) may lead to an inconsistent bootstrap sample size and the sampling distribution is associated with a specific sample size (Roberts & Fan, 2004). Thus, the current study used a resampling approach wherein a bootstrap sample of clients with replacements was drawn and the nested data structure was ignored.

Table 3.

Coefficients and variances for the impact of predictors on treatment satisfaction.

	Fixed effect		Random effect
	Intercept (SE]	Coefficient of level 1 predictor, β (SE); p value	Intercept, u0; p valueb	Variance component of level 2 slope, u1 (SD)c; p value	Level 1 residual, r1 (SD)	Effect size (R2)
Psychiatric symptoms	32.71 (0.47)	−0.23 (0.15); .11	0.41 (0.64); .35	0.09 (0.31); .03	10.48 (3.24)	14.87%
Anxiety	32.68 (0.43)	−0.14 (0.06); .02	0.07 (0.26); .42	-	11.32 (3.37)	8.04%
Depression	32.69 (0.42)	−0.14 (0.06); .03	0.003(0.06); .43	-	11.44 (3.38)	7.07%
Social supporta	32.69 (0.41)	0.25 (0.10); .02	0.004(0.07); .37	-	10.88 (3.30)	11.62%
Self-esteema		0.04 (0.07); .53
Motivation	32.58 (0.25)	0.19 (0.13); .16	0.10 (0.32); >.50	0.04 (0.21); .03	10.65 (3.26)	13.48%
Substance use severity	32.69 (0.44)	−0.30 (0.38); .44	0.004 (0.07); .32	-	12.19 (3.49)	0.97%

^aSocial support and self-esteem are in the same model and share the same intercept and effect size.

^bIn the HLM program, p values larger than .50 for level-2 variance are presented as p > .50.

^cPreliminary analyses indicated that the random effects of social support and self-esteem were not significant; therefore the slope was fixed. So was the slope of substance use severity. Effect size $R^2$ is computed by $\frac{r_0-r_1}{r_0}$, where $r_0$ is the within-group residual of the unconditional model and $r_1$ is the within-group residual of the model with predictor(s).

Table 4.

Coefficients and variances for the impact of predictors on treatment progress.

	Fixed effect		Random effect
	Intercept (SE)	Coefficient of level 1 predictor, β (SE); p value	Intercept, u0; p value	Variance component of level 2 slope, u1 (SD)b	Level 1 residual, r1 (SD)	Effect size (R2)
Psychiatric symptoms	24.06 (0.40)	−0.05 (0.09); .57	0.21 (0.46); .14	−	8.65 (2.94)	< .01%
Anxiety	24.03 (0.37)	−0.09 (0.05); .08	0.01 (0.08); .28	−	8.48 (2.91)	1.51%
Depression	24.06 (0.38)	−0.17 (0.03); < .001	0.19 (0.43); .20	−	7.54 (2.75)	12.43%
Social supporta	24.21 (0.47)	0.24 (0.08); .007	1.17 (1.08); .02	-	6.37 (2.52)	26.02%
Self-esteema		0.10 (0.06); .09
Motivation	24.09 (0.41)	0.18 (0.08); .04	0.34 (0.58); .13	−	8.01 (2.83)	6.97%
Substance use severity	24.06 (0.40)	0.05 (0.33); .88	0.21 (0.46); .14	−	8.70 (2.95)	< .01%

^aSocial support and self-esteem are in the same model and share the same intercept and effect size.

^bPreliminary analyses indicated that none of the random effects of predictors in each domain were significant; therefore the slope was fixed. Effect size $R^2$ is computed by $\frac{r_0-r_1}{r_0}$, where $r_0$ is the within-group residual of the unconditional model and $r_1$ is the within-group residual of the model with predictor(s).

Qualitative data analysis

Qualitative data from transcribed semi-structured interviews were coded manually for qualitative content analysis to help understand the recovery process (Hsieh & Shannon, 2005). The data analysis started with establishing a start-list of initial codes based on the interview guide. The first coder began to code the transcripts using the start-list of codes. Simultaneously, the coding process was open to additional codes emerging during the analysis. Coding concluded when new codes stopped emerging in the transcripts. Then, the second coder coded all the transcripts with the established codebook. These two coders debriefed, reached a consensus on the coding with extremely high agreement (intercoder agreement > 0.90), and finalized the codebook. Next, by consensus two coders classified codes into general themes. Themes that merged from early rounds of data analysis were iteratively used to review all data for negative cases (i.e., experiences or viewpoints differed from the current themes). This promoted the trustworthiness of the data (Creswell, 2013).

Results

Quantitative data analyses results

Sixty-four out of 90 baseline participants completed the follow-up and the attrition rate of the study was 29%. The descriptive statistics for the sample are presented in Table 1. The zero-order correlations between variables are provided in Table 2.

Table 2.

Correlations between client-level variables (N = 90).

Variables	Mean (SD)	1	2	3	4	5	6	7	8
1. Anxiety	26.70 (6.96)
2. Depression	24.33 (6.29)	0.55***
3. Psychiatric symptoms	16.39 (4.16)	0.56***	0.57***
4. Self-esteem	36.00 (5.76)	−051***	−0.71***	−0.42**
5. Social support	40.59 (3.95)	−0.16	−0.44***	−0.27*	0.36**
6. Substance Use Severity	8.00 (1.15)	0.16	0.13	0.25*	−0.29*	−0.13
7. Motivation	37.20 (4.25)	0.19	0.12	0.26*	−0.20	0.26*	0.26*
8. Treatment satisfactiona	32.69 (3.54)	−0.28*	−0.27*	−0.20	0.19	0.33**	−0.20	0.24*
9. Treatment progressa	24.03 (3.01)	−0.22	−0.36**	−0.07	0.31 *	0.34**	−0.07	0.25*	0.44***
10. Treatment retentionb	0.71	0.12	0.06	−0.15	−0.03	0.01	−0.03	−0.11	-

^aData of the variables were collected at follow-up, the sample size was N = 64.

^bBecause clients who did not complete 90-day treatment did not conduct follow-up assessment, the correlations between treatment satisfaction, progress, and retention were not performed.

^*p < .05,

^**p < .01,

^***p < .001.

Unconditional models

The proportion of variance in the data attributable to counselors, as estimated by the intraclass correlation (ICC; Snijders & Bosker, 2012), for treatment retention was 0.02, χ²(11) = 9.95, p > .50, indicating that the between-counselor variables (level-2) accounted for 2% of the total variance in treatment retention and 98% of the total variance was explained by the between-client differences (level-1). The ICC was 0.0003 (χ²(11) = 12.53, p = .33) for treatment satisfaction, indicating counselor-level differences and individual differences explained 0.03% and 99.9% of the total variance; however, the ICC was 0.18 (χ²(11) = 16.30, p = .13) for treatment progress, indicating counselor-level differences and individual differences explained 18% and 82% of the total variance, respectively. While only the between-counselor variance of treatment progress exceeded 10%, multilevel modeling was used to control for the counselor-level variances in all of the subsequent analyses.

Overall, the multilevel analyses indicated the client-level predictors were significantly associated with treatment satisfaction and progress, but not retention. The results of multi-level modeling indicated that generally the client-level predictors were not significantly associated with treatment retention. In these models, both fixed and random slopes were also modeled and used to examine counselor differences on treatment retention. Only the slope of substance use severity was significant signifying counselor-level differences for treatment retention: Var = 1.18, p = .01.

The results revealed that anxiety (β = −0.14, p = .02, R² = 8.04%), depression (β = −0.14, p = .03, R² = 7.07%), and social support (β = 0.25, p = .02, R² = 11.62% for both social support and self-esteem) were significantly associated with treatment satisfaction, respectively; but none of the other predictors were significant predictors (see Table 3). With regard to the counselor-level variance, the influence of psychiatric symptoms on treatment satisfaction differed between counselors (Var = 0.09, p = .03), as did motivation (Var = 0.04, p = .03).

The results of a series of modeling on treatment progress indicated that depression (β = −0.17, p < .001, R² = 12.43%), social support (β = 0.24, p = .007, R² = 26.02% for both social support and self-esteem), and motivation (β = 0.18, p = .04, R² = 6.97%) were the significant predictors of treatment progress (see Table 4), but none of the other predictors. With regard to counselor-level variance, all the slopes were fixed because preliminary analyses revealed that counselor-level variance was not significant.

Mediation

Based on the previous analyses, mediation analyses were performed to examine treatment satisfaction as a mediator on the association between depression and treatment progress. The results indicated statistical support for the indirect effect, Sobel’s z = −1.96, SE = 0.02, p = .05; bootstrapping estimate b = −0.047, p = .028, 95% confidence interval with bias correction (CI): [−0.113 to −0.005]. Likewise, analyses testing the mediational role of treatment satisfaction on the association between social support and treatment progress revealed statistical support for the indirect effect, Sobel’s z = 1.98, SE = 0.04, p = .048; bootstrapping estimate b = 0.09, SE = 0.05, p = .042, 95% CI with bias correction: [0.003 to 0.209]). Moreover, both models revealed partial mediation (see Figure 2).

Figure 2.

Mediation models. Left panel: How treatment satisfaction mediates the influence of depression on treatment progress. Right panel: How treatment satisfaction mediates the influence of social support on treatment progress (this model did not include self-esteem as a predictor, unlike results in Table 4). Unstandardized parameter estimates were reported. The number under the dotted arrow presents the indirect effect of predictor on treatment progress through treatment satisfaction (Krull & MacKinnon, 1999).

Qualitative data analyses results

The following qualitative data analyses used counselor inputs to aid in the understanding of client recovery process in regards to treatment process and outcomes. Eighteen counselor-rated barriers and facilitators, grouped into three levels: individual, program, and society, emerged from the data (see Figure 3). Discussed below are five themes built from the 18 factors.

Figure 3.

Barriers and facilitators to treatment retention and recovery success identified by counselors through qualitative analysis.

The first theme was reducing resistance and enhancing motivation. Counselors indicated that clients tended to resist referral to treatment because of frustration associated with the mandated continued care after discharge from institutional treatment. In some cases, the resistance was persistent throughout the duration of treatment, interfering with treatment retention. When asked what reasons led to treatment retention, all counselors referred to client motivation for treatment as a facilitator. While counselors strived to enhance intrinsic treatment motivation by a variety of counseling techniques including motivational interviewing, they also mentioned legal pressure as both a barrier and a motivator for offenders mandated to community treatment. Legal pressure may deter clients attending treatment, especially for those with high no-show rates.

The second theme was building strong therapeutic relationships which eventually promote treatment retention and other therapeutic benefits. One of the primary facilitators endorsed by counselors is the positive therapeutic relationship in which mutual trust, honesty, and genuineness are the core properties. They said that gaining trust and asking clients to share their treatment goals and disclose their addiction and other pertinent problems is effective for maintaining treatment retention and adherence. Counselors also explained that clients often felt resistant to treatment when they perceived that counselors and probation officers teamed up together “on the other side,” which echoes the importance of establishing mutual trust.

The third theme was cultivating a social support system which helps clients remain in treatment, work through stress and frustration, and eventually succeed. Family and other support systems (e.g., 12-step meetings, faith-based programs) could give clients hope and help them sustain recovery. Staff in the program who are in recovery themselves also served as role models and helped maintain clients’ hope.

Reducing relapse and recidivistic risk was another theme. Counselors noted that criminal lifestyle and criminal thinking, particularly related to drug dealing, was a major barrier to recovery and treatment progress. Consistent with previous findings (Bahr, Amstrong, Gibbs, Harris, & Fisher, 2005), counselors mentioned the importance of steering away from deviant peer networks (e.g., by teaching clients how to make changes in lifestyle and gain control over their environment) in order to minimize recidivism and substance use.

[Clients] may still live in the same environment, but [they] have to do new things in this environment that will manifest [they] stay clean.

The fifth theme concerned a variety of practical factors. Counselors recognized a lack of transportation as a challenge for treatment retention, and unemployment and conflicting treatment schedule as barriers to treatment success. In addition, a low level of literacy and a lack of vocational skills tend to constrain client life opportunities and further marginalize them, which makes their recovery process even harder.

Discussion

In the framework of a mixed-methods design, quantitative analyses explored the influence of client-level factors on these treatment outcomes and the role of treatment satisfaction in explaining the association between predictors and therapeutic progress, and qualitative methods revealed counselor-reported factors that were associated with treatment process and outcomes.

The impact of client-level factors

The quantitative findings revealed anxiety and depression as risk factors and social support and treatment motivation as protective factors for treatment outcomes. Clients with more anxiety tended to report less treatment satisfaction, which resonates with a recent study indicating that anxiety/depressive symptoms are associated with substance use at 12-month follow-up (Gil-Rivas, Prause, & Grella, 2009). If clients perceive that their treatment is unable to alleviate their comorbid problems, they may report less treatment satisfaction. However, anxiety was not a significant predictor of retention or treatment progress. Clients with a higher level of depression reported less treatment satisfaction and progress. Depression impacts a variety of cognitive functions (see a review by Hammer & Årdal, 2009) which in turn negatively impacts one’s ability to learn new coping skills. A negative affect could impair treatment participation and engagement, which in turn negatively impacts the perceptions of the program, leading to less satisfaction with treatment services. Consistent with the results from a meta-analysis indicating depression was one of the strongest predictors of continued illicit substance use for opiate addiction clients (Brewer, Catalano, Haggerty, Gainey, & Fleming, 1998), clients in the current study with more depressive symptoms tended to report less treatment progress including learned skills to cope with drug-related symptoms and psychosocial dysfunction. Furthermore, treatment satisfaction partially mediated the influence of depression on treatment progress. That is, clients with more depressive symptoms reported more negative experiences and less satisfaction, which could be associated with a reduced tendency to use treatment services, eventually leading to a lack of therapeutic benefits—less treatment progress in this study.

The quantitative findings indicated that social support was associated with both treatment satisfaction and progress, the importance of which was corroborated by the qualitative results. Studies have recognized the role of social support in recruiting and retaining individuals in treatment, as well as serving as a protective factor for drug relapse (Hser, Grella, Hsieh, Anglin, & Brown, 1999; Landau et al., 2000; Soyez et al., 2006). Individuals with substance use problems often come from marginalized social groups or dysfunctional families (Henderson, Boyd, & Mieczkowski, 1994), which may play a detrimental role in their development of self-esteem and evaluation of self-worth, and thus a lack of belief that they can succeed in treatment. Social support likely serves as an incentive to enhance psychological strength and encourage clients to get access to and retain in treatment and promote recovery. It is of note that one particularly important support mechanism is from staff, especially those in recovery, who provided strong feelings of hope and encouraged their clients’ retention in treatment and persistence in recovery. It is likely that all the encouragement leads to a high level of treatment satisfaction, and progress.

The findings of the current study also underscore the importance of social support in enhancing treatment satisfaction which in turn helps to promote treatment progress. In addition, the mediating effect of treatment satisfaction on both depression and social support is consistent with the literature indicating the role of treatment satisfaction in relation to favorable treatment outcomes (Carlson & Gabriel, 2001; Hser et al., 2004; Kelly et al., 2011). This suggests that treatment satisfaction could serve as a marker for evaluating treatment delivery and designing strategies to facilitate treatment access.

Both quantitative and qualitative findings supported the importance of motivation in treatment retention and recovery. Treatment motivation significantly predicted treatment progress; clients who were more motivated reported more treatment progress than their counterparts, which is consistent with the literature, as numerous studies have demonstrated the importance of treatment motivation in engaging clients in treatment and generating greater treatment achievements (Cosden et al., 2006; Joe et al., 1999; Simpson & Joe, 1993). Motivated clients tended to be more engaged in treatment-seeking behaviors (e.g., adhering to interventions), display greater commitment to treatment plans, and have fewer health-compromising behaviors (e.g., drug relapse), all leading to greater treatment progress. Additionally, the cultivation of motivation should persist throughout the whole treatment experience and appropriate counseling strategies are needed to break down treatment resistance and convert external motivation (e.g., legal pressure) to intrinsic treatment, the process of which builds therapeutic alliances (Ryan, Lynch, Vansteenkiste, & Deci, 2011). All of the aforementioned efforts point to the importance of counselor influence on client treatment process and outcomes.

Counselors’ influence

The impact of client psychiatric symptoms and motivation on treatment satisfaction differed among counselors, suggesting counselors as a key component in a client’s treatment experience through their involvement in designing treatment plans as well as engaging and retaining clients in treatment. This is consistent with previous findings on therapeutic alliance being linked to treatment engagement and favorable treatment outcomes (Meier, Barrowclough, & Donmall, 2005). Likewise, this study discovered that the influence of substance use severity on treatment retention differed among counselors, which points to the impact of therapy attributes on treatment outcomes (Project MATCH Research Group, 1998). Counselors may design individualized treatment plans, in terms of treatment type, intensity, and duration for instance, to meet with different client levels of substance use severity and thus affect the tendency of staying in treatment.

Limitations

The current study has several limitations. First, this study was based on a relatively small sample of client participants from one treatment facility, and there may have been lack of sufficient statistical power to test the hypotheses satisfactorily. Relatedly, several participants dropped out; however, the 90-day attrition rate for the study (~29%) approximates the normal range (>30%) for addiction treatment (Reyes, 2002; Simpson, Joe, Rowan-Szal, & Greener, 1997). Second, the current study sampled male justice-involved individuals mandated to community-based drug treatment and the results may differ for females, given that the prevalence of psychiatric problems and the characteristics of psychosocial functioning may differ between the sexes. Third, the majority of the participants listed heroin as the primary drugs of choice; the results may not be generalized to individuals who primarily use other types of drugs. Fourth, objective outcome measures (e.g., urinalysis results, recidivism data) or follow up with the dropouts for this convenience sample were not consistently available from the treatment provider. Future studies should anticipate this unavailability of information, which might help paint a bigger picture of justice-involved individuals mandated to community drug treatment. Fifth, the current dataset did not allow for separating out those who received just buprenorphine, just depot naltrexone, or a combination of the two. Thus, the association between type of MAT medication received and MAT outcomes was not examined. Sixth, specific counselor attributes (e.g., intervention training experience, disease model beliefs, theoretical orientation, etc.) may relate to treatment process and outcomes. Given the results suggesting counselor influence on treatment retention and satisfaction, consideration of counselor attributes seems important. The qualitative findings were based on feedback from seven counselors which might differ from the perceptions of those who did not participate. Despite the limitations of this administrative supplement, results reveal information of import to MAT treatment professionals.

Clinical implications

The current study lends evidence to best practices for the delivery of MAT treatment services and resonates with the call for studying substance abuse treatment via a dynamic systems perspective with multiple agents (e.g., counselors, clients, managers) nested in the complex content of healthcare (Flynn et al., 2012). The findings provide support for referring clients with co-occurring disorders to mental health services or embedding these services into substance abuse treatment (if this is not already the case). These efforts enhance treatment satisfaction and progress. The results also support the need for individualized treatment plans to incorporate information on specific treatment needs. Clients benefit from increased access to support groups (such as 12-step groups, NA, and AA) and provide role models of successful recovery. The current study further supports the importance of treatment motivation; counselors may consider providing clients with a clear understanding of the benefits of aftercare services and building therapeutic alliance. Moreover, considering that legal pressure can serve as a strong external motivator for some, it may be important for program administrators to work strategically with community corrections to find ways to assist in increasing intrinsic motivation, thereby engaging clients. These show potential for helping to improve treatment outcomes, which might be particularly crucial for justice-involved individuals. Counselor inputs also suggest that program may consider teaching clients self-management strategies about how to deal with deviant peers and attending to logistical factors (such as appointment schedule and service delivery methods) which would contribute to favorable client outcomes.