Client engagement and change during drug abuse treatment

Client engagement and change during drug abuse treatment

journal of Substance Abuse, 7, 117- 134 (1995) Client Engagement and Change During Drug Abuse Treatment D. Dwayne Simpson George W. Joe Grace Rowan...

1MB Sizes 0 Downloads 54 Views

journal of Substance Abuse, 7, 117- 134 (1995)

Client Engagement and Change During Drug Abuse Treatment D. Dwayne

Simpson

George W. Joe Grace Rowan-Szal Jack Greener Texas Christian

University

Previous work has shown length of time in drug abuse treatment is associated with better outcomes, but the role of therapeutic engagement and process needs further examination. In this study, the total number of counseling sessions attended by 557 clients in their first 90 days of community-based outpatient treatment was examined in relation to indicators of treatment delivery and progress. Significant client improvements were found on behavioral criteria and psychosocial functioning during the first 3 months of treatment, and session attendance was positively related to favorable behavioral changes as well as to positive perceptions by clients and counselors of their therapeutic interactions. Client background, treatment motivation, and therapeutic focus of counseling in Month 1 were significant predictors of session attendance in the first 3 months following admission to methadone treatment.

The length of time clients spend in drug abuse treatment has been the most consistent predictor of follow-up outcomes, a finding replicated across major therapeutic settings and approaches (Ball & Ross, 1991; De Leon, Wexler, & Jainchill, 1982; Gerstein & Harwood, 1990; Hubbard et al., 1989; Simpson & Sells, 1982). National treatment evaluations and surveys indicate, however, that there are wide variations in program and clientele characteristics across agencies (Hubbard et al., 1989; Price et al., 1991; Sells 8c Colleagues, 1975). Early impact studies recognized the need for more systematic evaluations of drug abuse treatment structure and process, client attributes, and environmental context (De Leon, 1984; Hubbard, Rachal, Craddock, & Cavanaugh, 1984; Simpson, 1984). Indeed, additional research on these domains of influence is essential for understanding treatment dynamics and how to improve its effectiveness, as discussed in Pickens, Leukefeld, and Schuster (1991) and illustrated in recent field-based

This work was supported by the National Institute on Drug Abuse (Grant No. DA06162). The interpretations and conclusions, however, do not necessarily represent the position of NIDA or the Department of Health and Human Services. Data were collected following approved informed consent guidelines. Correspondence and requests for reprints should be sent to D. Dwayne Simpson, Institute of Behavioral Research, Texas Christian University, P.O. Box 32880, Fort Worth, TX 76129. 117

118

D.D.

Simpson,C.W. Joe, C. Rowan-Szal,and J. Greener

treatment enhancement demonstration projects (see Tims, Fletcher, Inciardi, Delaney, & Horton, in press). Recent multisite studies examining treatment process and structure suggest that some attributes-reflecting program accessibility, policy, organization, and counseling skills and practices-are related to client outcomes and perceptions of quality of care (Ball & Ross, 1991; Joe, Simpson, & Hubbard, 1991). Compared to treatment outcome studies in which treatment has been measured primarily by length of stay, client-level evaluations involving treatment process and related contextual influences become much more complicated (Joe, Simpson, & Sells, 1994; Moos, Finney, & Cronkite, 1990). Especially important is the development of a conceptual framework in which theoretical constructs are operationalized and incorporated into broader analytic models (see Moos & Finney, 1988; Simpson, 1993). Relevant dimensions include community environmental and contextual factors, treatment agency outreach and client referral sources, program organization and structural characteristics, service delivery resources and implementation, therapeutic intervention and counseling strategies, client characteristics and attributes, and dynamics of client and counselor interactions. Due to the complexity of these dimensions and their interrelationships in the aggregate treatment system, a series of integrated research studies is necessary in which each of these measurement domains can be examined in detail. This study is part of a comprehensive treatment research project for developing and evaluating therapeutic enhancement strategies in community-based treatment settings (Chatham & Simpson, 1994; Simpson, Chatham, 8c Joe, 1993; Simpson, Dansereau, &Joe, in press). For instance, we have prepared counseling manuals to address special needs (e.g., HIV/AIDS education, women’s assertiveness training, and transition to aftercare) and also applied a cognitive-based visual representation technique, called node-link mapping, to help improve communications between counselors and clients that bear on problem deftnitions and ways to solve them in brief therapy applications (see Dansereau, Dees, & Simpson, 1994; Dansereau, Joe, & Simpson, 1993). In addition, we have placed major emphasis on the study of treatment process; more specifically, we have focused on client interactions with counselors and other change agents in the treatment setting (e.g., other clients, staff, rules, etc.) in relation to changes in cognitive attributions, psychological adjustment, and behavioral functioning of clients. This study is the first of a series relying on longitudinal assessments of these measures and examining their relationships both over time and with duringtreatment and posttreatment outcome measures. In particular, it focuses on the relation of therapeutic engagement (defined behaviorally in terms of session attendance) with client changes and the nature of treatment interactions during the critical first 3 months after admission. Past research has generally included overall treatment tenure as the principal indicator of “treatment dose,” but a more precise measure represented by session attendance is used here for defining level of engagement. The hypothesis is that significant cognitive and behavioral improvements occur during treatment and the magnitude of improvement

119

Client Engagement and Change

is related to level of engagement. In addition, counselor perceptions and ratings of client progress should be congruent with these changes, and the focus of counseling sessions should be related to client engagement level and functioning. We selected a multimethod analytic strategy, including the use of multivariate prediction models to examine the relationships among pretreatment characteristics of clients, treatment engagement, and outcomes. Later studies will expand the set of prediction variables to address the influence of family and friends as well as our therapeutic enhancement strategies, and they will include more advanced (path analytic) statistical approaches. In addition to our measures of duringtreatment changes, new data are also near completion from posttreatment followup interviews. With respect to measures of client performance and recovery, treatment outcome studies have typically emphasized behavioral criteria such as drug use, criminal activity, and employment. Focusing exclusively on behavioral indicators, however, obscures more subtle stages of change such as those conceptualized by Prochaska, DiClemente, and Norcross (1992) involving cognitive initiation and readiness phases that occur prior to overt behavioral actions. In this study, we therefore rely on a battery of longitudinal measurements representing (a) cognitive attributions about treatment and readiness for change, (b) self-perceptions of psychological and social adjustment, and (c) behavioral indicators.

METHOD Participants Clients A total of 557 clients who completed at least 3 months of treatment and complied with program requirements of attending at least one session per month served as the sample for this study. All were daily opioid users admitted to methadone maintenance (MM) programs at three clinics participating in the Drug Abuse Treatment for AIDS-Risk Reduction (DATAR) project during 1990 to 1993 (Simpson, Chatham, & Joe, 1993). These programs were part of larger multimodality agencies located in Corpus Christi, Dallas, and Houston, Texas. It is noted that approximately 14% of MM admissions in the DATAR project terminated treatment within 30 days, 25% within 60 days, and 35% within 90 days, and a prediction study of early dropout rates is reported elsewhere (Simpson & Joe, 1993). This study was necessarily limited to clients with minimum retention of 3 months in order to assess early treatment engagement and performance indicators. At intake, the average age of these 557 clients in this study was 37. Two thirds were male, and 20% were African American, 37% were Mexican American, and 40% were white. Forty-four percent were married or living as married, and 57% had worked in the past 6 months (including 34% full-time, 13% part-time, and 10% at periodic odd jobs). Approximately 27% depended on illegal activities (including prostitution) for their major source of financial support prior to in-

120

D.D.

Simpson, G.W. Joe, G. Rowan-Szal,

and J. Greener

take, and 30% had been arrested in the previous 6 months; 43% were legally involved (awaiting trial, probation, parole), but less than 5% were referred to treatment by legal authorities or under explicit legal pressure to be in MM treatment. All were daily opioid drug users, and cocaine was the second major drug of choice (crack was seldom used by the opioid addicts included in this study); 30% used heroin only and no cocaine, and the remaining 70% reported preadmission cocaine use (31% used cocaine on a daily basis with heroin, 16% used it weekly, and 23% used it less than weekly). Counselors Of the 34 counselors

involved in the research project at the three participating agencies, 19 were males, 18 were white, 11 were African American, 4 were Mexican American, and 1 was American Indian. Their average age was 35. Twenty had bachelor’s degrees, 7 had graduate (master’s) degrees, and 2 had junior college degrees. Of the remaining 5 counselors, 2 had completed 2 years of college, 1 had attended 1 year, and 2 reported finishing high school only. Five were recovering substance abusers, and all operated on a brief therapy model emphasizing problem solving and case management. Measures

The data collection instruments and schedule for measuring client background and functioning, treatment process, and outcomes are presented in detail elsewhere (Simpson, 1992). As part of the admission process, each client completed a psychosocial self-rating form, an 88-item self-administered instrument for assessing psychological functioning (including scales for self-esteem, depression, and anxiety, with coefficient alpha reliabilities of .75 to .82), social functioning (i.e., hostility, risk taking, and social conformity, with coefficient alpha reliabilities of .64 to .83), and treatment motivation (i.e., drug use problems, desire for help, and treatment readiness, with coefficient alpha reliabilities of .72 to .89; see Knight, Holcom, & Simpson, 1994; Simpson &Joe, 1993). A structured intake interview also was conducted by a treatment counselor. This face-to-face assessment lasted approximately 1 hour and addressed the client’s sociodemographic background, family background, peer relations, criminal history, health and psychological status, drug use history, and behavioral risks for HIV/AIDS. A series of during-treatment assessments was scheduled at monthly intervals for each of the first 3 months after admission. This included repeated administrations of the psychosocial self-rating scale as well as a brief interview that assessed behavioral functioning (using a core subset of items from the intake form). In addition, each client and counselor completed monthly forms for rating cognitive appraisals and perceptions of treatment activities, progress, and therapeutic interventions (described later). Finally, counselors completed brief reports after each individual and group session to document client attendance. Research staff at each agency maintained tracking records on client services and contacts over time.

Client Engagement and Change

121

Treatment Sessions Clients were generally encouraged to attend two individual counseling sessions and four group counseling sessions per month, although program policy required no more than one per month. The number of counseling sessions actually attended by clients in this study during the first 90 days of treatment ranged from 3 to 19. Sessions were defined as face-to-face therapeutic interactions of at least 15 min; they averaged about 45 min for individual and 80 min for group sessions. Brief contacts in a hall or at the methadone dosing station were not counted as sessions. Attendance was influenced by scheduling considerations (availability and conflicts), counselor judgments of therapeutic need (specific problems), and client preferences. Level of engagement in treatment was measured by the total number of sessions attended because it was considered to be most analogous to the concept of treatment dose. Limiting sessions to those attended in the first 90 days permitted level of engagement to be measured on a common time base for all and permitted it to be differentiated from time in treatment. To simplify some of the analyses and results presented later, the sample was divided into three categories-3 to 5 sessions (n = 188,34% of the sample), 6 to 8 sessions (n = 180,32%), and 9 or more sessions (n = 189,34%)-representing low, middle, and high engagement groups. Thus, the low group averaged about one session per month, the middle group averaged about two sessions per month, and the high group attended three or more per month. The actual number of sessions attended by the low engagement group averaged 4.1 (3.30 in individual and .85 in group sessions), compared to 6.9 for middle engagement (4.96 individual and 1.94 group) and 11 .O for high engagement (6.04 individual and 4.96 group). Overall, session attendance for the total sample averaged 35% in group settings versus 65% in individual, but this was not constant across engagement groups; that is, group sessions accounted for 20% of total attendance in the low engagement group, 28% in the middle group, and 45% in the high engagement group. Urinalyses Random urine samples were collected from clients and analyzed monthly throughout treatment using the enzyme multiplication immunoassay technique (EMIT). There were two measures of “dirty urines” used in this study, based on the presence of opiate and cocaine metabolites, respectively. Although other drugs were also included in the full-screen urinalyses, their rate of usage and detection was so infrequent that they could not be meaningfully used as outcomes. Client Evaluation of the Counselor and the Treatment Program Clients used 10 items (5-point scales) to rate their counselors at the end of each month. Our previous research with these data showed that a single factor score was satisfactory for representing this evaluative dimension (see Dansereau et al., 1993). The scale, defined as the average score on the 10 items representing

122

D.D. Simpson, G.W. Joe, C. Rowan-Szal, and J. Greener

rapport, trust, and expertness, was labeled counselor respect and had a coefficient alpha reliability of .9 1. Monthly assessments of client perceptions of treatment were also obtained based on a 12-item instrument (5-point scales). Two reliable factors were used, each with coefficient alphas above .75. One was interpreted to be personal progress and was defined by overall satisfaction, progress in making changes in life, help for drug use, and help for nondrug problems. The second represented therapeutic helpfulness and was defined by satisfaction with program characteristics and sessions, including friendliness of the program staff, meeting times of sessions, helpfulness of individual counseling sessions, and helpfulness of group counseling sessions. Counselor

Evaluation

of the Client

Counselors provided monthly ratings of each client assigned to their caseload. Ratings (on a 5-point scale) were completed on 19 characteristics that addressed client functioning and progress in treatment. There were three reliable factors, each with coefficient alphas above .75. One was interpreted to be rapport; the defining items included easy to talk to, warm and caring, honest and sincere, understanding, not suspicious, and not in denial about problems. A second factor was labeled motivation and was defined by being dependable, well-organized, cooperative, motivated, and attending individual counseling as scheduled. Selfconfidence was the interpretation given to the third; it was defined by being selfconfident, persuasive, motivated, and assertive. Therapeutic strategies and areas that counselors addressed with each client during the month were described using 5-point ratings on 2 1 items. These represented five dimensions, including communication skills, goal setting, confronting self, empathy building, and problem solving (with 3 to 6 items in each scale). Coefficient alphas for these scales ranged from .71 to .84. Data Analysis

The analyses focus on client functioning in the second and third months of treatment. Except for urines, scores from these 2 months for each variable were averaged; this procedure increased reliability of measurement and limited reductions in sample size due to missing data (which ranged up to about 25% for some of the monthly client self-report variables). These measures will be referred to as Month 3 in the analyses. “Dirty urine” measures were based only on agency records for Month 3, for which there was virtually no missing data. Analyses of variance (ANOVAs) were performed to address treatment engagement effects (i.e., three levels of session attendance) and changes from intake to Month 3 (when applicable) for each measurement domain. A series of regression analyses (SAS Institute, 1990) were also carried out to examine the effect of engagement on outcome criteria after controlling for the effects of client background and motivational predictors (Simpson &Joe, 1993). Multiple regression analyses used variables in raw-score form, although distributions of continuous measures were often skewed. Logistic regressions were also per-

Client

Engagement

and Change

123

formed using dichotomous outcome scores defined for each criterion (usually representing median-split values). The results using both regression models were similar; therefore, only the conventional multiple regression analyses are presented. Percentage of criterion variance accounted for by predictor measures (i.e., R*) is shown for each of these analyses, but this value probably represents a conservative or biased estimate for criteria due to their skewed distributions. RESULTS Client Progress

and Engagement

Effects

Means, standard deviations, and the ANOVAs for measures of behavioral and psychosocial functioning are presented in Table 1. With respect to illegal activity and drug use indicators for alcohol, heroin, cocaine, and speedballs (recorded without duplication of heroin and cocaine use alone), repeated-measure ANOVAs showed there were highly significant decreases from intake to Month 3 for each measure. For instance, mean scores for days of illegal activities in the past month dropped from 11 .O to 1.5, and average frequency of needle injections was reduced from 108.2 to 9.5 per month. Although not reported in the table, persons with 1 or more days of illegal activity in the past month dropped from 57% at intake to 17% at Month 3; those with 15 or more drug injections per month dropped from 98% at intake to 12% in Month 3. There were also significant self-reported decreases in frequency for heroin, cocaine, and speedball use. Separate tabulations (not included in the table) showed that percentages admitting any use in the past month at intake and Month 3 dropped from 94% to 41% for heroin, 46% to 26% for cocaine, and 5 1% to 24% for speedballs; combining reports of cocaine and speedball usage into a single indicator reflected a drop from 61% to 37%. The significant decreases in self-reported drug use were validated by urinalysis results. That is, dirty urines for opiate metabolites dropped from 86% at intake to 38% in Month 3, and from 47% to 35% for cocaine metabolites. In addition to these main effects for time, analysis of cocaine-positive urines revealed significant main effects for engagement levels, indicating higher session attendance was related to less frequent use. Significant interactions between time and engagement level for both heroin use measures (i.e., self-report and urine data) showed that reductions over time were different among engagement groups-addicts attending more sessions had less reported use and lower percentages of dirty urines at Month 3. Mean scores for most of the criterion measures, however, dropped to very low levels during treatment and left little opportunity to detect between-group variations. Except for hostility, scores on all of the psychological and social functioning scales also improved significantly over time during treatment. Self-esteem, depression, and social conformity had significant interactions, indicating that changes were differentially associated with treatment engagement levels. These measures, as well as others that failed to reach statistical significance, were all consistent in showing that increased session attendance was related to more favorable psychosocial progress at the end of the first 3 months of treatment.

!ii

Summary

M W)

Intake

Time

M

(SD)

2.55

Social conformity

2.68

1.12 1.37

2.53 1.45 1.64

= monthly,

(.50)

(.65) (.62)

(.69) (.63) (.72)

AFrequency of use codes: 0 = none, l-3 often, and 4 = almost always. ***p < .OOl. *I < .05. **p < .Ol.

1.17 1.58

Hostility Risk taking

2.21 1.71 1.98

4-5

(.61) (.54)

(.62) (.59) (.66)

(.62) (.61) (.74)

M

9+

(4.8) (9.4) (1.4) (46) (1.3)

(SD)

271.6*** 62.8*** 3115.0*** 418.1*** 107.6*** 27.6*** 274.7*** 997.5***

Time

(.58) (.72)

codes: 0 = never,

I = rarely,

2.80 59.75*** 28.89***

94.04*** 132.43***

3 =

1.81 2.35 9.47***

3.51* 2.99* 1.62

2.57 1.42 4.33* 5.12* 1.87 1.32 1.10 1.44

TxE

2 = sometimes,

.07 1.04 4.76**

.42 .85 .48

.61 2.39 .84 1.21 1.37 4.00* 1.79 .54

Engage

ANOVA Results (F)

C.62) 112.52***

1.10 (.62) 1.30 (.60) 2.73 (.52) (n = 173)

2.59 1.44 1.64

.3 (44) .43 (1.0) 4.3 (12.0) (n = 184)

1.5 5.9 .79 31 .56

‘IItem scoring

1.10 (.69) 1.37 (.59) 2.71 (.50) (n = 156)

2.53 1.42 1.61

.78

(4.0) (10.8) (1.5)

(SD)

(49) (1.5) 42 (49) .59 (1.2) 12.6 (76.7) (n = 174)

39

1.0 7.7 1.04

M

= daily.

2.57 (.46) (n = 145)

1.18 1.46

2.44 1.51 1.67

= weekly, and 6-8

(.49)

(.64) (.58)

(.62) (.59) (.71)

Self-Ratings of Psychological and Social Functioning (N = 474) PsychologicalFunctioningh

Self-esteem Depression Anxiety Social Functioningb

(SD)

6-8

No. of Sessions Attended

Functioning

2.0 (6.5) 9.2 (11.3) 1.29 (1.8) 45 (50) .85 (1.6) 38 (49) .73 (1.4) 12.2 (27.3) (n = 163)

M

3-5

and Psychosocial

Month 3

of Results for Behavioral

Illegal Activity and Drug Use in Last 30 Days (N = 521) Days of illegal activities 11.0 1.5 (13.2) (5.2) Days of alcohol use 10.9 7.5 (10.5) (12.5) Freq. of heroin use” 6.65 1.03 (2.0) (1.6) % with dirty urine 86 (35) 38 (48) Freq. of cocaine use= 1.75 .72 (2.4) (1.5) % with dirty urine 47 35 (50) (47) .58 Freq. of speedball use” 2.62 (3.1) (1.2) No. of drug injections (57.6) 9.5 (47.4) 108.2

Table 1.

Client

125

Engagement and Change

Analysis of cognitive attributions and perceptions of clients and counselors in Month 3 showed there were consistent linear increases in scores across engagement groups. The only measure falling short of statistical significance was counselor respect (reflecting client ratings of the counselor rapport, trust, and expertness), F(2, 505) = 2.05, p > .05. In contrast, client satisfaction with personal progress, F(2,504) = 6.15, p < .O1, and with therapeutic helpfulness, F(2,505) = 6.58, p < .O1, was significantly greater among the more engaged clients. Consistent with expectation, counselor evaluations of client rapport, F(2, 522) = 21.86, p < .OOl, motivation, F(2, 522) = 68.98, p < .OOl, and self-confidence, F(2, 522) = 9.51, p < .OOl, likewise increased in relation to the frequency of session attendance. The analyses of therapeutic focus and strategy as viewed by the counselor also indicated that all five dimensions of counseling efforts were significantly higher for clients who attended more sessions. That is, clients who were more engaged were more likely to have worked on improving communication skills, F(2,522) = 9.85, p < .OOl, goal setting and planning ahead, F(2, 521) = 11.80, p < .OOl, encouraging confrontive self-appraisal and assuming personal responsibility, F(2, 521) = 3.13, p < .05, building empathy and understanding for others, F(2, 521) = 17.57, p < .OOl, and solving problems with cause-and-effect thinking, F(2, 521) = 10.31,p < .OOl. Correlational analysis was also conducted to examine relationships between counselor appraisals of clients and the areas of therapeutic focus (n = 525). Results based on the total sample showed that counselor ratings of client rapport and motivation each were positively and significantly correlated (p < .OOl) with a therapeutic emphasis on improving communication skills (.35 and .36, respectively), goal setting (.42 and .45), empathy building (.39 and .48), and problem solving (.31 and .30). Higher client self-confidence was associated with more counseling efforts devoted to goal setting (.29) and empathy building (.32). However, these client attributes-rapport, motivation, or self-confidence-appeared to have limited relationship with counseling efforts to encourage clients to confront self and take personal responsibility (.OO to .13), presumably something all clients needed. Furthermore, it was found that scores on all five dimensions of therapeutic focus were more strongly associated with frequency of attendance at individual than at group counseling sessions. Prediction

of During-Treatment

Outcomes

Based in part on results from our previous work (Simpson & Joe, 1993), six background measures were selected as pretreatment predictors of outcome criteria; these included (a) being divorced, separated, or widowed; (b) being unemployed in the 6 months before entering treatment; (c) frequency of alcohol use in the 6 months before treatment; (d) frequency of cocaine use in the 6 months before treatment; (e) number of arrests in the 6 months before treatment; and (f) total number of months in previous treatments. In addition, the Desire for Help scale was selected to represent motivational status at the time of entering treatment. Using multiple regression models, these background and motivation

126

D.D.

Simpson, C.W.

Joe, C. Rowan-Szal,

and J. Greener

measures were entered along with level of treatment engagement as predictors of Month 3 outcomes. Results are presented in Tables 2 and 3, which summarize overall statistical tests and the significance of outcome predictors. Each regression weight represents the predictor’s contribution after statistical adjustments that take into account all other predictors, thereby indicating its unique contribution. In general, the findings indicate that the level of engagement was significant even after controlling statistically for pretreatment background variables and motivation at intake, extending the findings reported earlier using ANOVAs. These results demonstrate that after adjusting for selected pretreatment differences in this manner, the relationship of session attendance with during-treatment outcomes was usually significant. In most cases, the total proportion of variance accounted for statistically in these outcome variables is small (i.e., under lo%), but the consistency of findings across measures as well as their apparent clinical implications (based on scores in Table 1) are encouraging nonetheless. As shown in Table 2, treatment engagement was a significant predictor of each major behavioral outcome except illegal activities. These results indicate that alcohol use and drug injection frequency during treatment were also associated with session attendance, even though this relationship was not revealed in the ANOVAs discussed earlier. (Heroin and cocaine use results are based on selfreport data, but findings for urine results were essentially the same and therefore are not reported in the table.) In addition to engagement, pretreatment cocaine use was predictive of cocaine use and drug injections in Month 3. During-treatment involvement in illegal activities was predicted from number of arrests in the 6 months prior to treatment, unemployment, and prior treatmentbut not session attendance. Among the psychosocial functioning measures, more favorable scores on selfesteem and risk taking were related to increasing levels of engagement. For selfesteem, engagement was only one of four moderate-level predictors, but engagement was the single significant predictor for the measure of risk taking. Table 3 shows significant predictors of ratings given by clients and by counselors. For client ratings of counselor respect, personal progress, and therapeutic helpfulness, pretreatment motivation was the strongest predictor, suggesting that cognitive readiness for treatment at the time of intake had an important influence on client perceptions and attributions over the next 3 months. The frequency of session attendance during those months, however, further promoted favorable client evaluations of self and treatment. Counselor ratings of rapport, motivation, and self-confidence reflected by clients, on the other hand, had virtually no relationship to motivational claims each client made before treatment. Instead, weights of the predictor variables suggest counselors tended to assess clients from a behavioral perspective, based primarily on how often treatment sessions were attended. To a lesser degree, counselor ratings were also associated with background information on client drug use, treatment involvement, and social functioning before treatment admission. Therapeutic focus reported by counselors likewise was predicted by engagement behaviors of the client-that is, frequency of session attendance was directly related to the amount of therapeutic attention devoted to each problem area.

Multiple

*f

*p < .05.

Separated or Divorced Unemployed Alcohol Use Cocaine Use Arrests Prior Treatment Desire for Help No. of Sessions R’ F

**p < .Ol.

-.04 .15*** .06 .Ol .15*** .13** .oo .oo .07 4.58*** 8,472

Illegal

Activity

and

ground

***p < ,001.

8,472

-.03 - .02 .62*** -.Ol -.02 .oo -.05 -.09* .41 40.5 1***

Use

Behavioral

.02 .02 .Ol -.06 .OG .02 -.04 -.12** .03 1.57 8,472

Use

-.02 .05 .07 .29*** -.03 .07 -.08 -.10* .13 8.75*** 8,472

Use

Cocaine

Functioning

Heroin

Behavioral

Predicting

Alcohol

Regressions

Engagement

Client Back-

Table 2.

-.03 -.07 .04 .19*** -.Ol .03 -.08 -.16*** .08 4.81*** 8,472

Injections

Drug

-.03 -.11* -.08 .03 -.02 -.16*** -.16*** .10* .07 4.50*** 8,463

Esteem

.03 .11* .06 -.05 .oo .09* .12** -.04 .04 2.25* 8,463

-.02 .07 .10* -.08 .05 .06 .19*** -.Ol .05 3.08** 8,463

Anxiety

Psychosocial

-.oo -.oo .09 .Ol .05 .03 .10* -.02 .02 1.23 8.463

Hostility

.04 .Ol .04 .08 .05 .07 -.08 -.11* .04 2.49* 8,463

Taking

Risk-

Treatment

Functioning

by Clients During

Depression

Functioning

Self-

and Psychosocial

Social

-.Ol -.09* -.12* -.03 -.03 -.10* .oo .07 .04 2.63** 8.463

Conform

df

*p < .05.

Separated or Divorced Unemployed Alcohol Use Cocaine Use Arrests Prior Treatment Desire for Help No. of Sessions R2 F

Predicting

**p < .Ol.

-.Ol -.06 -.02 .03 .10* -.Ol 15*** . .09* .05 2.78** 8,462

***p < ,001.

p.03 -.09 -.06 .02 .10* -.03 .24*** .16*** .ll 7.04*** 8,46 1

Personal Progress

8,462

-.02 -.02 -.07 .02 .05 .oo .21*** .17*** .09 5,37***

Perceptions

-.Ol p.04 -.10* -.03 -.05 -.12** .05 .27*** .12 8.00*** 8,474

Rapport

.03 -.08 -.11** -.09* -.05 -.11** -.02 .4s*** .29 24.52*** 8,474

Motivation

.02 -.11* -.08 -.08 -.12** -.02 .04 .17*** .09 5.73*** 8,474

SelfConfidence

.02 .05 -.08 -.05 .03 .02 .03 .15*** .03 2.43* 8,474

.02 -.03 -.Ol -.09 .08 -.Ol .03 .19*** .05 3.28** 8,473

Goal Setting

.05 -.Ol .04 .Ol .05 .06 .12** .10* .04 2.17* 8,473

Confront. Self

.Ol .02 -.03 -.05 .Ol -.02 .06 .23*** .06 3.89*** 8,473

Empathy Building

Therapeutic Focus and Strategy

Treatment

Commun. Skills

During

Counselor Evaluation of Client

Client and Counselor

Therapeutic Helpfulness

Client Evaluation of Self and Treatment

Regressions

Counselor Respect

Multiple

Client Background and Engagement

Table 3.

-.04 .03 .05 .04 .09* .17*** .05 3.33*** 8,473

.08

.oo

Problem Solving

Client Engagement and Change

Prediction

129

of Treatment Engagement

The findings presented earlier demonstrated that counseling session attendance throughout the first 3 months of treatment was associated with better outcomes and therapeutic involvement by Month 3. There were also indications that some client background and baseline variables were related to session attendance. In order to examine these relationships more directly, we evaluated the extent to which client preadmission characteristics as well as initial treatment experiences (e.g., in the first month) influenced engagement level over the S-month period of early treatment. A two-stage model was tested in which sets of predictors representing pretreatment and early treatment variables were entered sequentially into a multiple regression equation. The first stage included sociodemographic measures (age, gender, and race-ethnicity), baseline criteria (employment, illegal involvement, alcohol use, drug use, and previous treatment), and motivational assessments at intake. In the second stage, measures taken from the first month of treatment-including methadone dosage, client ratings of self and treatment, and counselor ratings of client attributes and therapeutic focus-were added to the prediction equation. Correlations of these individual predictor measures with session attendance were initially examined. Stage 1 measures that were significantly correlated (p < .05) with engagement included being white (.14) and employed (.lO). Significant correlations for Stage 2 measures were as follows: Month 1 client ratings of therapeutic helpfulness (. 15), Month 1 counselor ratings of client rapport (.16), treatment motivation (.37), self-confidence (.20), therapeutic focus on empathy building (.20), and problem solving (.ll) in Month 1. Using a sample of 359 clients (which included clients who had no missing data on any of the predictor or criterion measures), the results from the combined two-stage regression model accounted for a total of 22% of variance in session attendance (i.e., R’ = .22). Stage 1 predictors explained 6% of the variance, F( 16, 342) = 1.31, p > .05, and the variables from Stage 2 explained an additional IS%, F(27, 331) = 3.37, p < .OOOl. In the overall model, being white (vs. minority group), being perceived by counselors as having higher treatment motivation and better rapport in Month 1, and receiving counseling that emphasized problem-solving applications in Month 1 were the predictors with significant regression weights. Because the previous correlational analysis included other predictors that were also related to engagement, these results reflect the presence of intercorrelations and related measurement redundancies among the measures that influenced statistical determinations of the reduced set of significant predictors.

DISCUSSION

Findings of this study help identify some of the treatment factors and processes that account for its effectiveness. Historically, retention has been one of the most reliable predictors in outcome studies, and session attendance was

D.D. Simpson, C.W. Joe, C. Rowan-Szal, and J. Greener

130

presumed

to be a major

part

of the

treatment

engagement

and

intensity

on

which retention effects depend. As hypothesized, this study showed that engagement was positively associated with recovery-oriented perceptions as well as behavioral changes by clients. Engagement level also made unique contributions beyond those associated with favorable client background and baseline characteristics.

Analysis

of particular

client

and treatment

gagement indicated that whites had more minority groups; in addition, higher session ment

employment.

peutic

The

strongest

alliance-particularly

therapeutic focus on empathy weeks of treatment. The positive relationship consistent with results from client-level of services

influences

higher

client

building

were related motivation

and problem

groups Clients

predict

en-

to elements

and rapport solving

of thera-

ratings

during

and

the first few

between treatment participation and outcomes is other research based on program-level as well as

examined in publicly funded methadone maintenance 1991; Joe et al., 1991; Joe et al., 1994) private chemical

cy programs (McLellan, Grissom, et al., 1993), (Moos et al., 1990). In a systematic evaluation made

that

comparisons. At the program level, indicators of quality and quantity delivered to clients were associated with more favorable treatment

results when (Ball & Ross,

client-level

attributes

frequent session attendance than attendance was related to pretreat-

outcomes,

random

McLellan,

assignments

Arndt,

of opioid

and alcohol of treatment

Metzger, addicts

Wood,

to three

treatment intensity and

programs dependenprograms effects on

O’Brien

methadone

(1993)

treatment

whose services ranged from low to high in a hospital-based program. receiving the most services had the best outcomes during and at the end

of treatment. Although

previous

treatment

tion and related perceptions ments have proven difficult

evaluation

of treatment to conduct

efforts

often

are important, and incorporate

assume

client

motiva-

these kinds of assessinto the analysis of

outcomes. Results of this study give empirical support to their importance. More specifically, client motivation at intake (i.e., desire for help) was related to favorable client perceptions of treatment process features (i.e., counselor respect and therapeutic helpfulness) and psychological functioning during treatment. Motivation scores at intake did not predict long-range engagement levels, but this, in part

may have been

the result

of variance

restrictions

due to the uniformly

high scores and related exclusion of early treatment dropouts who did not qualify for this study. Cognitive processes that influence motivation and readiness for change, on the other hand, are complex, leading to dynamic short-term variations in personal attitudes. Nevertheless, counselors appeared to make reliable observations because their rating of client motivation in Month 1 was a highly significant predictor of the total number of counseling sessions attended. The inclusion of a broad range of assessment criteria in this study represents an effort to examine multidimensional changes that occur in client recovery, similar to those suggested by the work of Prochaska et al. (1992). Conceptual expansion of measurement domains has been especially important in our evaluations of therapeutic enhancements that focus on cognitive strategies and developmental changes

in client

and counselor

relationships

over

time

(see Dansereau

et al.,

Client

Engagement

and Change

131

1993). More work is needed, however, to explore the complex stages of change presumably involved in recovery from addiction. An example is the three-phase model recently proposed by Howard, Lueger, Maling, and Martinovich (1993) to explain progressive improvement of psychotherapy outcomes, using a set of simple indicators reflecting cognitive appraisal of well-being, clinical symptomatology, and behavioral functioning. They suggest that there is a sequential process of change that spans these three areas, and our findings are consistent with the model. Although the data collection instruments we used for this study yielded theoretically driven composite criteria that had good measurement properties, some refinements are needed. For example, client ratings of the therapeutic relationship with their counselors each month (based on a set of 10 items) were uniformly high, yielding a single composite measure we called “counselor respect.” Due to client response styles-probably influenced by situational demand characteristics associated in part with concerns about the confidentiality of their ratings-these items did not discriminate along the separate dimensions we originally intended. Conceptually, they represented empathy, trust, expertness, and dependability and were founded theoretically on a social-influence therapeutic model (see Corrigan & Schmidt, 1983; Heppner, Rosenberg, & Hedgespeth, 1992; Valle, 1981; Whiston & Sexton, 1993). Based on this model (also see Orlinsky SC Howard, 1986) some of our exploratory analyses of comparable items from client and counselor rating forms representing mutual rapport (i.e., “easy to talk to” and “warm and caring”) have been encouraging because they suggest that matched high scores are related to better client engagement and retention. In order to examine further the notion of therapeutic alliance, items that have highly skewed response distributions and restrictions in variance need continued work and adaptation to community treatment in field settings. Measurement of client engagement in treatment also needs to be re-examined for ways to represent other cogent sources of therapeutic contacts, including those outside formal counseling sessions. In contrast to our chosen method of relying on counselor records as the basis for measuring outpatient service delivery, McLellan and associates used a brief client interview each week to obtain information on in-program as well as out-of-program services received in major psychosocial and medical areas (see McLellan, Alterman, Cacciola, Metzger, & O’Brien, 1992). Data collected for this study represent a provider-rather than a consumer-viewpoint and contain more details on within-session counseling strategies and activities. Treatment settings and related circumstances may determine whether client or counselor reports are the most feasible and reliable data source for documenting services. Each one involves potential distortion resulting from selective recall and interpretation. Treatment process research from both approaches, however, is consistent in showing engagement effects. The predictive value of more detailed information on the type and setting of therapeutic contacts should be a subject of future work. These results reinforce the need to continue efforts for maximizing motivation levels at intake and sustaining them during treatment (e.g., Miller, 1989; Miller & Rollnick, 1991). Recovery from addiction represents a complex process of transi-

132

D.D.

tions involving in psychological

Simpson, C.W. Joe, C. Rowan-Szal, and J.

cognitive readiness for change, perceptions functioning and emotional symptomatology,

Greener

of personal progress cognitive awareness

and acquisition of better problem-solving skills, therapeutic and social interactions, and behavioral commitment. This dynamic and transitional process, however, is only partially influenced by client interactions within the treatment environment. factors from

There

that exert

are

direct

many

other

cultural,

as well as indirect

contextual,

effects

and

on persons

environmental

trying

to recover

addiction.

Additional and formulate

studies of treatment process are needed to identify parameters effective analytic frameworks that improve prediction of out-

comes during defined here

and after was based

treatment. The preliminary model of engagement as solely on the number of formal counseling sessions

attended; expanded models are needed that consider a wider array of therapeutic contacts, counseling enhancement strategies, differential applications of individual and group counseling, and improved measures of counseling interactions therapeutic alliance). On the other hand, it cannot be as(e.g., representing sumed that the formulation of treatment process and extra therapy factors that influence client change are uniform across time or settings. These and many other

issues with practical

implications

for treatment

enhancement

remain

to be

explored. REFERENCES Ball, J.C., & Ross, A. (1991). The eff ec tiveness ofmethadone maintenance treatment. New York: SpringerVerlag. Chatham, L.R., & Simpson, D.D. (1994). Delivering and evaluating outpatient treatment: Finding a practical balance. In F. Tims, B. Fletcher, J. Inciardi, & A. Horton (Eds.), Drug abuse treatment: Implementation of innovative approaches. Westport, CT: Greenwood Press. Corrigan, J.D., & Schmidt, L.D. (1983). Development and validation of revisions in the counselor rating form. Journal ofCounseling Psychology, 30, 64-75. Dansereau, D.F., Dees, S.M., & Simpson, D.D. (1994). Cognitive modularity: Implications for counseling and the representation of personal issues. Journal of Counseling Psychology, 41, 5 13-523. Dansereau, D.F., Joe, G.W., & Simpson, D.D. (1993). Node-link mapping: A visual representation strategy for enhancing drug abuse counseling. Journal of Counseling Psychology, 40(4), l-l 1. De Leon, G. (1984). Program-based evaluation research in therapeutic communities. In F.M. Tims 8c J.P. Ludford (Eds.), Drug abuse treatment evaluation: Strategies, progress, and prospects (NIDA Research Monograph No. 51, DHHS Publication No. ADM 84-1329). Washington, DC: US. Government Printing Office. De Leon, G., Wexler, H.K., & Jainchill, N. (1982). The therapeutic community: Success and improvement rates 5 years after treatment. International Journal ff the Addictioq 17, 703-747. Gerstein, D., & Harwood, H. (Eds.). (1990). Treatingdrugproblemr (Vol. 1). Washington, DC: National Academy Press. Heppner, P.P., Rosenberg, J.I., & Hedgespeth, J. (1992). Three methods in measuring the therapeutic process: Clients’ and counselors’ constructions of the therapeutic process versus actual therapeutic events. Journal ofCounseling Psychology, 39, 20-31. Howard, K.I., Lueger, R.J., Maling, MS., & Martinovich, Z. (1993). A phase model of psychotherapy outcome: Causal mediation of change. ,Journal ofConsulting and Clinical Psychology, 61, 678685.

Hubbard, R.L., Marsden, M.E., Rachal, J.V., Harwood, H.J., Cavanaugh, E.R., & Ginzburg, H.M. (1989). Drug abwe treatment: A national study of effectiveness. Chapel Hill: University of North Carolina Press.

133

Client Engagement and Change

Hubbard, R.L., Rachal, J.V., Craddock, S.G., & Cavanaugh, E.R. (1984). Treatment outcome prospective study (TOPS): Client characteristics and behaviors before, during, and after treatment. In F.M. Tims & J.P. Ludford (Eds.), Drug abuse treatment evaluation; Strategies, progress, and prospects (NIDA Research Monograph No. 51, DHHS Publication No. ADM 84-1329). Washington, DC: U.S. Government Printing Office. Joe, G.W., Simpson, D.D., & Hubbard, R.L. (1991). Treatment predictors of tenure in methadone maintenance. Journal of Substance Abme, 3, 73-84. Joe, G.W., Simpson, D.D., & Sells, S.B. (1994). Treatment process and relapse to opioid use during methadone maintenance. American Journal of Drug and Alcohol Abuse, 20, 173-197. Knight, K., Holcom, M., & Simpson, D.D. (1994). TCU psyc h asocial functioning and motivation scales: Manual on psychometric properties. Fort Worth: Texas Christian University, Institute of Behavioral Research. McLellan, A.T., Alterman, A.I., Cacciola, J., Metzger, D.S., & O’Brien, C.P. (1992). A new measure of substance abuse treatment: Initial studies of the Treatment Services Review. Journal of Nervous and Mental Disease, 180, 101-110. McLellan, A.T., Arndt, I.O., Metzger, D.S., Wood, G.E., & O’Brien, C.P. (1993). The effects of psychosocial services in substance abuse treatment. Journal of the American Medical Association, 269, 1953-1996. McLellan, A.T., Grissom, G.R., Brill, P., Durell, J., Metzger, D.S., & O’Brien, C.P. (1993). Private substance abuse treatments: Are some programs more effective than others? Journal of Substance Abuse Treatment, 10, 243-254.

Miller, W.R. (1989). Increasing

motivation

for change. In R.K. Hester & W.R. Miller (Eds.), Handbook alternatives. New York: Pergamon.

of alcoholism treatment approaches: Effective

Miller, W.R., & Rollnick, S. (199 1). Motivational interviewing: Preparing people to change addictive behavior. New York: Guilford. Moos, R.H., & Finney, J.W. (1988). Alcoholism program evaluations: The treatment domain. Drug.s and Society, Z(2), 3 l-5 1. Moos, R.H., Finney, J.W., & Cronkite, R.C. (1990). Alcoholism treatment: Context, process, and outcome. New York: Oxford University Press. Orlinsky, D.E., & Howard, K.I. (1986). Process and outcome in psychotherapy. In S.L. Garfield & A.E. Bergin (Eds.), Handbook of psychotherapy and behavior change. New York: Wiley. Pickens, R.W., Leukefeld, C.G., & Schuster, C.R. (1991). Improving drug abuse treatment (NIDA Research Monograph No. 106). Rockville, MD: National Institute on Drug Abuse. Price, R.H., Burke, A.C., D’Aunno, T.A., Klingel, D.M., McCaughrin, W.C., Rafferty, J.A., & Vaughn, T.E. (1991). Outpatient drug abuse treatment services, 1988: Results of a national survey. In R.W. Pickens, C.G. Leukefeld, & CR. Schuster (Eds.), Improving drug abuse treatment (NIDA Research Monograph No. 106, DHHS Publication No. ADM 91-1754). Rockville, MD: National Institute on Drug Abuse. Prochaska, J.O., DiClemente, C.C., & Norcross, J.C. (1992). In search of how people change: Applications to addictive behaviors. American Psychologist, 47, 1102-l 114. SAS Institute, Inc. (1990). SAS technical report P-200: Release 6.04. Cary, NC: Author. Sells, S.B., & Colleagues. (1975). The DARP research program and data system [Special issue]. American Journal of Drug and Alcohol Abuse, 1- 136. Simpson, D.D. (1984). National treatment system evaluation based on the Drug Abuse Reporting Program (DARP) follow-up research. In F.M. Tims & J.P. Ludford (Eds.), Drug abuse treatment evaluation: Strategies, progress, and prospects (NIDA Research Monograph No. 51, DHHS Publication No. ADM 84-1329). Washington, DC: U.S. Government Printing Office. Simpson, D.D. (1992). DATAR forms manual: Drug abuse treatmentfor AIDS-risks reduction (DATAR). Fort Worth: Texas Christian University, Institute of Behavioral Research. Simpson, D.D. (1993). Drug treatment evaluation research in the United States. Psychology of Addictive Behaviors, 7, 120-128. Simpson, D.D., Chatham, L.R., &Joe, G.W. (1993). Cognitive enhancements to treatment in DATAR: Drug abuse treatment for AIDS risks reduction. In J. Inciardi, F. Tims, & B. Fletcher (Eds.), Innovative approaches to the treatment of drug abuse: Program models and strategies. Westport, CT: Greenwood Press.

134

D.D. Simpson, C.W.

Joe, C. Rowan-Szal,

and J. Greener

Simpson, D.D., Dansereau, D.F., &Joe, G.W. (in press). The DATAR project: Cognitive and behavioral enhancements to community-based treatments. In F.M. Tims, B.W. Fletcher, J.A. Inciardi, P. Delaney, & A.M. Horton (Eds.), The effectiveness of innovative strategies in the treatment of drug abuse. Westport, CT: Greenwood Press. Simpson, D.D., &Joe, G.W. (1993). Motivation as a predictor of early dropout from drug abuse treatment. Psychotherapy, 30, 357-368. Simpson, D.D., & Sells, S.B. (1982). Effectiveness of treatment for drug abuse: An overview of the DARP research program. Advances in Alcohol and Substance Abuse, 2, 7-29. Tims, F.M., Fletcher, B.W., Inciardi, J.A., Delaney, P., & Horton, A.M. (Eds.). (in press). The effectiveness of innovative strategies in the treatment of drug abuse. Westport, CT: Greenwood Press. Valle, S.K. (1981). Interpersonal functioning of alcoholism counselors and treatment outcome.Journal of Studies on Alcohol, 42, 783-790. Whiston, S.C., & Sexton, T.L. (1993). An overview of psychotherapy outcome research: Implications for practice. Profesional Psychology: Research and Practice, 24, 43-5 1.