Stages and processes of change among polydrug users in methadone maintenance treatment

ELSEVIER

Drug and Alcohol Dependence39 (1995)45-53

Stages and processes of change among polydrug users in methadone maintenance treatment Mark A. Belding*a, Martin Y. Iguchia, R.J. Lamba, Martin Lakinb, Robert Terryb “Division of Addiction Research and Treatment, Department of Psychiatry, Medical College of Pennsylvania and Hahnemann University, MS 984, Broad and Vine Streets, Philadelphia, PA 19102-1192. USA bDepartment of Psychology, Social and Health Sciences, Duke University, Durham, NC, USA

Received11January1995;accepted13April 1995

Abstract We applied the stages-of-changemodel of Prochaskaand DiClemente to the problem of drug useamong methadonemaintenance patients to examine correlates of different stagesof treatment readiness.The 276 subjectswere divided into stagecategoriesbased on self-reported drug use and questionnaire responsesregarding plans to discontinue unauthorized drug use in the future. Confirmatory factor analytic proceduresvalidated four processscalesderived from a 60-item questionnaire. Each stagewas characterized by a profile of change-processscoreslargely consistent with predictions, though thesescoresdid not distinguish stagesas clearly as has been reported in previous research. Analysis of subject characteristics revealed that those in the Precontemplation stage reported significantly longer treatmenttenuresthan subjectsin any other stage besidesMaintenance. Keywords:

Methadone; Stages-ofchange; Heroin; Treatment readiness

1. Iotroduction

Methadone maintenance is the most effective widely available treatment for opiate addiction; its effectiveness, however, varies greatly, and many patients continue using heroin and other drugs during treatment (Ball et al., 1988;Iguchi et al., 1993).Although counseling may be an effective adjunct to methadone treatment (McClellan et al., 1988), there is little consensusregarding which specific approaches are most effective or how interventions should be tailored to suit the needsof individual patients. Treatment failures are typically ascribed to the insufficient motivation of the addicts to change. However, such attributions do little to improve treatment, and little empirical researchhas directly addressed this issue. The stages-of-changemodel articulated by Prochaska and DiClemente provides a way of conceptualizing and * Correspondingauthor.

measuring readinessto change. Applied to this population, the model could offer a means of identifying different levels of treatment motivation in different patients and a framework for effective treatment matching. According to the model, individuals attempting to modify a particular behavior progress through a series of five stages:(1) Precontemplation, (2) Contemplation, (3) Preparation, (4) Action, and (5) Maintenance (Prochaska et al., 1992a). Each stage represents a distinct constellation of attitudes, behaviors, and intentions regarding the problem behavior (see Table 1). Precontemplation, for example, describes individuals with minimal awarenessof the problem and no intention of changing in the near future. By contrast, Maintenance describes individuals who have succeededin changing the problem behavior and who continue working to maintain the changes they have achieved. In addition to the stagesof change, twelve cognitive and behavioral processesthat individuals use to effect

03768716/95/$%X50 0 1995ElsevierScienceIreland Ltd. All rights reserved SSDZ 0376-8716(95)01135-L

46

MA.

Belding et al. /Drug and Alcohol Dependence 3945 (1995) 45-55

therapeutic changehave also been identified (Prochaska et al., 1992a).These processesrepresent activities ranging from ConsciousnessRaising (increasing awareness about the existence of a problem in need of change) to Stimulus Control (avoiding the stimuli that occasion the problem behavior). Research with smokers suggests each stageof changeis characterized by the use of a different pattern of change processes (Prochaska and DiClemente, 1983;DiClemente et al., 1991). Generally, those in the earlier stagesof change are more likely to use processesassociatedwith insight-oriented therapies - processesserving to increase understanding of the nature and implications of a particular problem. Individuals in the later stagesare more apt to use actionoriented processeswhich serve to extinguish problem behaviors while facilitating and reinforcing more adaptive ones. Measures for assessingboth stagesand processeshave demonstrated reliability and validity with populations of smokers, psychotherapy patients, alcoholics in treatment, women enrolled in a weight control program (Prochaska et al., 1992a), and employeesin a work site health promotion project (Marcus et al., 1992). In several studies, these measures have been shown to predict progress in treatment (DiClemente et al., 1991; Prochaska et al., 1992b). The present study was designed to assess the generalizability and applicability of previous findings to the complexities of the methadone maintenance treatment setting. The presenting problem of individuals seekingmethadone maintenance is opioid (usually heroin) dependence.However, treatment programs typically regard all illicit drug use as a target of behavior change. This contrasts with previous research utilizing the stages-of-changemodel, in which the target of change is a single discretebehavior. Also, by definition, all methadone patients make use of a powerful pharmacological agent to alter their drug use. It remains to be determined whether the same stagesof change can be identified in this population and to what extent methadone patients see themselves as using any change processesbesides methadone itself. If the processesof change have clinical relevance in this setting, they should also demonstrate someassociation with recent drug use. Specifically, those patients making use of processesassociatedwith the later stages of change should be engaging in less drug use. On the other hand, one might expect a positive correlation between drug useand processesassociatedwith the earliest stagesof change (e.g., ConsciousnessRaising and Environmental Re-evaluation) since patients using these processesmay be only beginning to consider changing their drug using behavior. 2. Method 2.1. Subjects The original sample consisted of 289 methadone

maintenance patients recruited from four Philadelphiaarea methadone clinics by means of bulletins posted in eachof the clinics. Recruitment and data collection took place during medicating hours when the largest number of patients could be offered the opportunity to participate. Data from 13 subjects were judged invalid and therefore omitted. Subject data were ruled invalid if: (i) the subject appeared to have answered randomly or in a clear pattern for more than two pages(e.g., all twos or alternating ones and threes); or (ii) the subject did not follow instructions in filling out the demographic data sheet (e.g., answering ‘yes’ and ‘no’ to questions which asked for numbers or dates). Nine subjectsmet the first criterion and three subjectsmet the second. The data of the thirteenth subject were discarded because his case defied stagecategorization. Although he acknowledged chronic use of a variety of drugs (including Percocet, marijuana, benzodiazepines),he stated that the program approved his drug use for the relief of AIDS-related symptoms. 2.2. Measures 2.2.1. Stagesof change. Measures of stagesand processesof ‘change’ presuppose a problem or behavior which is being changed. For the measuresin this study, the problem specified on the questionnaires was use of illicit drugs (i.e., mood-altering drugs which have not been explicitly approved by a program physician, excluding caffeine, alcohol and tobacco). This problem is specific enough to be operationally definable; at the sametime it has broader clinical relevance than use of a single drug (e.g., heroin), since many methadone patients use more than one class of drugs. Polydrug use is also the implicit treatment goal of FDA regulations which do not distinguish among different types of drug use in providing guidelines for determining program privileges. Thus most programs penalize useof any illicit substances by restricting take-home medication privileges. Moreover, in many clinics, chronic illicit drug use (regardless of substance) results in administrative detoxification and discharge from the program. Subjectswere divided into five stagecategories based on their self reported drug use and their responsesto a questionnaire asking whether or not they planned to discontinue unauthorized drug usein the next 30 days or the next six months. This measurewas basedon an algorithm developed by Prochaska and colleagues in their previous research (DiClemente et al., 1991). Table 1 (column 2) illustrates the algorithm used to classify subjects in this study. 2.2.2. Processesof changescale. Processesof change were assessedwith a 60-item questionnaire based on similar measuresdesigned for use in other areas of addiction research including smoking (Prochaska et al., 1988),weight control (Prochaska et al., 1992b),and cocaine use (Rosenbloom, 1991). For this version, items from related versions were modified and supplemented

M.A. Belding et al. /Drug and Alcohol Dependence 3945 (1995) 45-55

with severalnew items generatedby the authors to apply to the problem of unauthorized drug use among methadone maintenance patients. The measure includes five items for each of twelve hypothesized subscales: (1) ConsciousnessRaising, (2) Self Re-evaluation, (3) Self Liberation, (4) Counterconditioning, (5) Stimulus Control, (6) Contingency Management, (7) Helping Relationships, (8) Dramatic Relief, (9) Environmental Re-evaluation, (10) Social Liberation, (11) Interpersonal Control and (12) Medication. Each item describes a strategy or situation which an individual might use to discontinue drug use. For each item, subjects are asked to rate on a five-point Likert scale how frequently they make use of the particular activity to avoid using drugs. 2.2.3. Drug use. The accuracy of self-reported drug use of addicts has been relatively well established (Magma et al., 1987;Maisto et al., 1990). In this study, subjectscompleted questionnaires indicating how many days in the past 30 they had used each of five categories of drugs other than methadone: opiates, cocaine, benzodiazepines,marijuana, and an ‘other’ category which included amphetamines, barbiturates and propoxyphene. Subjects indicating use of ‘other’ drugs were asked to specify which drugs. For prescribed drugs, subjectswere asked to indicate whether their use was explicitly approved by the program. Approved use of prescribed drugs was not counted as illicit use. To create an indicator of polydrug use, the sum of the days of reported use for any drug was calculated for each subject. Thus, a subject reporting 22 days of opiate use and 11 days of benzodiazepine use in the past 30 would obtain a score of 33 on the polydrug use variable. 2.2.4. Demographic questionnaire. Subjects also recorded their age, sex, racial/ethnic background, level of education, employment status, years of drug use, years of opiate use, and amount of time in treatment.

41

appear to have been used with methadone maintenance patients and since, in this case,we were examining processesutilized to discontinue polydrug use rather than use of a single drug, it was necessaryto confirm the internal validity of the measure.EQS confirmatory factor analytic procedures (Bentler, 1991) were used to verify the factor structure of the process measure and to produce scales for subsequent analyses. To determine whether stageclassification actually did produce distinct groups of subjects, the data were analyzed for stagerelated differences in drug use, demographic characteristics, and reported use of change processes. An alpha level of 0.05 was used for all statistical tests. 3. Results 3.1. Subject characteristics

Forty-four percent of subjects were female and 53% were white, with 36% black, 9% Hispanic, 1% Native American and 1%Asian. Subjectsranged in age from 20 to 63 years (mean = 39.4 years) and reported a mean educational level of 11.6years. Nineteen percent of subjects reported full-time employment with 32% unemployed, 15% homemakers, 1% full-time students, 20% receiving disability payments and 13% reporting occasional or part-time employment. Subjects averaged 17 years of drug use and 14 years of opiate use prior to their current treatment episodeswhich ranged in duration from two days to 23 years (mean = 3.8 years, Mdn = 1.9 years). Analyses of available data indicated no significant differencesbetween the samplesand the clinic populations from which they were recruited on any demographic variables except employment status. The samples contained greater proportions of disabled patients as well as smaller proportions of full-time employed patients than the clinics from which they came.

2.3. Procedures

At clinics 2-4, the measureswere administered on a single day to groups of approximately 15 subjects at a time. Subjectsat theseclinics were assuredthat the questionnaires were completely anonymous and provided no means for connecting specific individuals to their responses.At clinic 1, where follow-up data were collected as part of a larger study (Belding, 1993), subjects were recruited over a two-month period. The measures were administered either individually or in groups of up to ten subjects.The questionnaires could not be anonymous at clinic 1 since they had to be linked to follow-up data; however, subjects were assured that their responseswould be kept confidential from clinical staff and would not affect their treatment. All subjects completed informed consent forms. All questionnaires were completed on site and in the presence of research assistants to minimize incomplete or random responding. Subjects received US$lO for their participation. Sincethe Processesof Change Questionnaire does not

3.2. Stages of change

Thirteen of the 276 subjects could not be classified into a stage on the basis of the algorithm because of missing or inconsistent information. Five of these subjects failed to provide a date for their last episode of drug use. Five subjects indicated on the drug use questionnaire that they were currently using marijuana/hashish but on the stage algorithm indicated that they had not used any illicit drugs in the past 30 days. Three subjects acknowledged use of other drugs including opiates in the last 30 days on the drug use questionnaire but then indicated that they had not used any illicit drugs in the past 30 days on the stage algorithm. The results for the remaining 263 subjectsare summarized in column 4 of Table 1. Each of the stagesof change was well represented in the sample. Over 65% of subjects acknowledged unauthorized drug use in the 30-day pretest period and were consequently classified into one of the first three

48

MA.

Belding et al. /Drug and Alcohol Dependence 3945 (1995) 45-55

Table 1 Stagesof change: operational definitions, characteristics, and percentagesof subjects in each Stage of change

Operational definition as assessedby stages of change questionnaire

Hypothesized attributes associatedwith each stage

Percentageof subjects per stage (N = 263) (%)

Precontemplation

No plan to discontinue unauthorized drug use in the next 6 months Plan to discontinue unauthorized drug use in the next 6 months but not in the next 30 days Plan to discontinue unauthorized drug use in the next 30 days No unauthorized drugs in the past 30 days but use within the past 6 months No unauthorized drugs in the past 6 months

No awarenessof problem and no desire to change behavior in the near future Aware of problem and seriously considering behavior change but have taken no action yet Intend to change behavior and have begun to take steps in that direction Actively working to change behavior or environment to overcome problem Working to maintain changes in behavior which have already been achieved

20

Contemplation Preparation Action Maintenance

stagesof change. The percentagesof subjects reporting various types of drug use did not vary greatly across stages.Opiates were the most widely used drugs with 80% of drug-using subjectsacknowledging opiate use in the past 30 days. Fifty-four percent reported cocaine use, 45% reported benzodiazepine use, 26% reported using marijuana or hashish, and 11% reported ‘other’ drug use. To determine whether there were differences across the first three stages in the amount of acknowledged drug use, a seriesof ANOVAs with Tukey comparisons were performed using the number of days of use for the various classes of drugs as the dependent variables. Theseanalysesindicated no differences across stagesin the use of any single class of drugs except the ‘other’ category, F(2,170) = 4.44, P c 0.05. Subjects in the Precontemplation stage (mean = 3.23, S.D. = 8.33) acknowledged more use of these drugs (mostly propoxyphene and barbiturates) than subjects in the Contemplation (mean = 0.43, SD. = 2.38) or Preparation stages (mean = 0.64, S.D. = 3.25). Precontemplators also reported significantly more polydrug use (mean = 32.17, S.D. = 22.42) than subjects in the

16 30 16 18

Preparation stage (mean = 22.23, S.D. = 18.84). On average,drug-using subjectsreported use of two classes of drugs including about ten days of opiate use. 3.3. Demographic data To be certain that the stage measure was not confounded by some demographic variable and to explore any relationships between stages and other subject characteristics, it was important to screen for the possibility of associations between these variables. To examine the relationship between the stagesof change and the categorical demographic variables, multiple chisquared tests were conducted, using Bonferroni adjustments and crossing stagesof change with sex, ethnic background, marital status, and employment status. The tests showed no significant relationships except between the stagesof change and employment status, x2 (16, N = 261) = 48.08, P < 0.001. However, further review of the these data revealed no clear pattern in the relationship betweenemployment and stagesof change. A MANOVA was conducted to determine whether there was a significant relationship betweenthe stagesof change and the following continuous variables: age,

Table 2 Comparisons of age and treatment tenure at current clinic across stage categories Variable

Precontemplation (N = 52), mean (S.D.)

Contemplation (N = 43), mean (SD.)

Preparation (N = 78), mean (S.D.)

Action (N = 42), mean (S.D.)

Maintenance (N = 47), mean (S.D.)

F-test

Tukey comparisons, P < 0.05

Age

39.25 (5.34)

40.16 (8.24)

37.18 (7.63)

39.57 (7.41)

41.69 (7.09)

PA < M

Months in treatment

74.08 (77.32)

35.17 (52.94)

30.63 (45.90)

31.26 (43.67)

56.79 (52.08)

F(4,258)= 3.17* F(4.253)= 6.29**

Categories: Precontemplation (PC), Contemplation (C), Preparation (PA), Action (A) and Maintenance (AI). +P < 0.05.

l *P < 0.001.

PC > PA,C,A

M. A. Belding et al. /Drug and Alcohol Dependence 3945 (1995) 45-55

x

35

5 2

3o

c”

25

49

the Maintenance stage can be found in the interval of 2-5 years in treatment. The highest proportion of subjects in the Action stagecan be found among those who have been in treatment for six months or less. A chisquared test indicated significant differences among these proportions, even after eliminating Maintenancestage subjects, whose treatment tenures are necessarily related to stageclassification, x2 (g,N = 2 15) = 2 1.63, p < 0.05.

sFL 20

3.4. Processes of change

% L

To test the validity of the twelve hypothesized processesof change, a confirmatory factor analytic model was specified in which each factor was represented by five items. Each item could load only on its designated factor, and all twelve factors could correlate with each other. This model failed to achieve an adequate fit with the data. The maximum likelihood solution yielded a significant chi-square goodness-of-fit value, x2 (1644, N = 275) = 3086.02, P < 0.001’. In confirmatory factor analysis, a non-significant chi-square suggestsa wellfitting model. However, for large samples, a nonsignificant value occurs only rarely and the chi-square goodness-of-fit test results in the rejection of virtually any model (Bentler and Bonett, 1980; Marsh et al., 1988).Alternate measuresof goodness of tit have been proposed by Bentler and Bonett (1980) and Bentler (1990); however, these measuresalso failed to support the model: Bentler-Bonett Non-normed Fit Index = 0.76, Comparative Fit Index = 0.77. Bentler (1992) suggeststhat values below 0.90 on these indices indicate a poorly specified model. Inspection of the results indicated that the failure of the model was due, at least in part, to high correlations among the factors. Correlations betweentwo setsof factors (ConsciousnessRaising and Self Re-evaluation as well as Self Re-evaluation and Dramatic Relief) attained a value of 1 while eight other correlations reached a value of 0.90 or greater. Using such results to alter a model marks a divergence from strictly confirmatory procedures. However the continued use of structural equation modeling allowed for theory-based respecilication of the model. The twelve hypothesized factors were consolidated into six factors by combining items from factors that were highly intercorrelated in the original model. All items composing Consciousness Raising, Dramatic Relief, Environmental Re-evaluation, and Self Re-evaluation (all intercorrelations greater than 0.90) were allowed to load on a single factor hereafter termed Re-evaluation. These four hypothesized processesall involve raising one’s awarenessof a problem and assessingits impact on one’s self and others. Thesefactors have been related

1.5 10 5 0 <==6mos n=58

>6mos <= 2 yrs n--El Time in Treatment

>2 yrs. <= 5 yn. n=66

25 yn. n=58

at Current Clinic

Fig. 1. Percentageof subjects in each of the stagesof change for four different treatment intervals.

level of education, years of drug use prior to current treatment episode, years of opiate use prior to current treatment episode, and months in methadone treatment at current clinic. The MANOVA was significant, Wilk’s Lambda (20, 826.79) = 0.63, P c 0.0001. Univariate F tests were significant for age and months in treatment only (Table 2). A Tukey test indicated that those in the Maintenance stagewere significantly older than those in the Preparation stage. More surprising was the strength and direction of the association between the stages of change and the amount of time spent in treatment by each subject at their current methadone clinic. Precontemplators reported an average treatment tenure of approximately six years at their current clinics significantly longer than subjects in the Contemplation, Preparation and Action stages.On the other hand, they did not differ significantly from the other groups in the amount of time they reported using drugs prior to entering treatment. To better understand the relationship between the stagesof change and time in treatment, time in treatment was divided into four intervals with roughly comparable numbers of subjects in each. Though the study is cross-sectional,Fig. 1 indicates that the proportion of those in the Precontemplation stage increasesas a function of time in treatment. Among subjectsreporting five or more years in treatment at their current clinic, approximately 35% fell into the Precontemplation stage while only 12%of those who had been in treatment for six months or less were Precontemplators. The proportion of those in the Preparation stage decreasesas a function of time. The greatest proportion of subjects in

I Due to a clerical error, the questionnaire was not administered to one subject, leaving a sample size of 275 for these analyses.

50

M.A. Belding et al. /Drug and Alcohol Dependence 3945 (199s) 45-H

to each other empirically in previous research which suggests they are all most frequently utilized in the earliest stagesof the change (Prochaska et al., 1992a). Those items composing Interpersonal Stimulus Control, Stimulus Control and Counterconditioning (all intercorrelations greater than or equal to 0.80) were allowed to load on a single factor hereafter termed Behavioral Processes.These three factors all relate to behaviorally oriented treatments and have been shown to mediate the transition from Action to Maintenance (Prochaska et al., 1992a). Finally, the items from factors 6 and 7 (Helping Relationships and Contingency Management) were allowed to load on a single factor hereafter termed Reinforcing Relationships. These two original factors exhibited an intercorrelation of greater than 0.90. Previous researchindicates that Helping Relationships and Contingency Management should be most widely utilized in the Action and Maintenance stagesof change (Prochaska et al., 1992a).Of the five items which make up the Contingency Management scale, four involve being rewarded by others for avoiding drug use. In this respect, the item content bears considerable similarity to that of the hypothesized Helping Relationships scale. The remaining three factors were left unchanged in the respecified model becausethere was no clear rationale for combining them. Factor 12 (Medication) did not correlate highly with any other factors (all intercorrelations lessthan 0.40), and neither theory nor previous researchsuggestedclose relationships with other factors. Factors 5 and 11 (Self Liberation and Social Liberation) each correlated highly with theoretically disparate factors.

The respecified model incorporated all 60 items into the six factors described above; the factors were all allowed to correlate with each other. However, the new model lit the data little better than the original model. The maximum likelihood solution yielded significant chi-square goodness-of-fit value, x2 (1695, N = 275) = 3275.64, P < 0.001, with the following alternate goodness-of-fit indices: Bentler-Bonett Non-normed Fit Index = 0.74, Comparative Fit Index = 0.75. Inspection of the results indicated that two of the factors (Social Liberation and Medication) remained weakly identified. Neither factor included more than two items with loadings above 0.50. All items comprising these factors were dropped from subsequent analyses. The model was respecilied a second time, eliminating the two weakly identified factors mentioned above and those items which failed to contribute significantly to the identification of their particular factors; this third model included only those items with factor loadings of 0.60 or above. These modifications reduced the total number of items to 25. SeeTable 3 for a summary of the respecified model and definitions of the factors. This third model fit the data considerably better than the previous two. The maximum likelihood solution again yielded a significant chi-square goodness-of-lit value, x2 (269, N = 275) = 543.25, P < 0.001. However the alternate tit indices equaled or exceeded the 0.90 value cited as desirable by Bentler: Bentler-Bonett Nonnormed Fit Index = 0.90, Comparative Fit Index = 0.91. On the basis of these scores,it was determined that this substantially respecified model fitted the data adequately.

Table 3 Summary of final scale composition including number of items contributed to each scale from hypothesized original factors Final scale

Definition

Hypothesized factors with items contributing to final scale

Re-evahtation

Increasing awarenessof the problem and its negative consequencesfor self and others

(I) ConsciousnessRaising (2) Self re-evaluation (8) Dramatic relief (9) Environmental re-evaluation

Self liberation

Making a conscious commitment to stop using illicit drugs

(3) Self liberation

3

Reinforcing relationships

Availing oneself of relationships with others who support the effort to change and who provide rewards for not using dNa

(6) Contingency management (7) Helping relationships

4 3

Behavioral processes

Avoiding people, places, and situations associatedwith drug use as well as engaging in alternative behaviors instead ofdnlguse

(4) Counterconditioning (5) Stimulus control (11) Interpersonal stimulus control

3 2 3

No. of items included on final scale

M.A. Belding et al. /Drug and Alcohol Dependence 3945 (1995) 45-55

Four processesof change scaleswere created by combining the items which made up the factors in the final model. Scores on these scales were derived by calculating the mean of the individual item scores for each scale.These calculations resulted in four processes of change scores for each subject with values ranging from 0 to 5. These,in turn, were converted into T-scores (mean = 50, S.D. = 10) for purposes of comparison. The scalesdemonstrated a high degree of internal consistency, yielding Cronbach’s alpha coefficients ranging from 0.70 (Self Liberation) to 0.86 (Reinforcing Relationships). The scales were also moderately intercorrelated with Pearson correlations ranging from 0.49 (Reinforcing Relationships and Reevaluation) to 0.68 (Self Liberation and Behavioral Processes). A seriesof analyseswas conducted to examine possible relationships between the process scalesand demographic variables. A MANOVA was conducted with process scale scores as the dependent variable and the following variables defining groups: sex, ethnic background, marital status, and employment status. Only main effects were examined since inclusion of interaction effects would have resulted in many cells with few subjects.The MANOVA indicated no significant effects for any of these variables. The scalesalso did not correlate significantly with any of the continuous demographic variables. 3.5. Processes of change and drug use

To assessthe relationship between the processesof change and recent drug use, Pearson correlations were calculated between the process scalesand the polydrug use variable. The results were generally consistent with the predictions of the model. Self Liberation, Reinforcing Relationships, and Behavioral Processes, all of which are associatedwith the later stagesof change,produced modest but significant (P < 0.01) negative correlations with polydrug use over the past 30 days (r = -0.21, -0.17 and -0.26, respectively). Thus, subjects reporting greater use of these processes also reported less drug use. On the other hand, Reevaluation, which has been associated with the two earliest stagesof change, produced no correlation with polydrug use (r = 0.00). A stepwise multiple regression (probability of F-toenter = 0.05, probability of F-to-remove = 0.10) indicated that Helping Relationships and Self Liberation did not explain any additional variance in recent drug use after the Behavioral Processesvariable entered the equation. However, a positive association did emerge between Re-evaluation and polydrug use. The final equation included only Behavioral Processes (6 = -0.33) and Re-evaluation (/3 = 0.16) and yielded an adjusted R-square of 0.08. This finding suggeststhat Reevaluation, by itself, is not associated with more drug use. Rather those subjects who report the greatest

51

n q

56

Precontemation Contarfplatim

(PC) (C)

T

42 40 0

Reevaluation no significant CllfferellCeSf

Self Liberation PC,C
Reinforcing Relationships PC
Behavioral Processes PC
Fig. 2. Mean processesof change scoresfor subjectsin each of the five stagesof change. l Tukey comparisons, P < 0.05.

amount of drug use are those who report engaging in Re-evaluation without also using Behavioral Processes to address their drug problems. However, neither variable accounted for much of the variance in the polydrug use variable. To determine whether there were stage-relateddifferences in the reported use of processesof change, a MANOVA was performed with process scores as the dependent variables and stagesof change as the grouping variable. The MANOVA was significant, Wilk’s Lambda (16, 773.57) = 0.76, P < 0.001. Univariate Ftests were significant for each of the scales except Re-evaluation. Many of the differences between adjacent stageswere not statistically significant; however all of the significant differences were in the predicted directions. Tukey comparisons indicated subjects in the Action and Maintenance stages scored significantly higher than subjects in the Precontemplation stage on the use of all the processesexcept Re-evaluation and higher than subjects in the Contemplation stage on the use of Behavioral Processes(Fig. 2). Prepared subjects distinguished themselvesfrom subjects in the first two stagesby their high scoreson the use of the Self Liberation process. Inspection of Fig. 2 reveals that although subjects in the Precontemplation and Contemplation stagesdid not score higher than other subjects on Re-evaluation, they did obtain their highest scoreson this scaleas the model would predict. Researchwith other populations has suggestedthat the transition from Preparation to Action is mediated by Self Liberation, the process of consciously making the commitment to change. The results obtained

52

MA. Belding et al. /Drug and Alcohol Dependence3945 (1995) 45-55

here are consistent with this finding in that those in the Preparation and Action stages obtained their highest mean scores on the Self Liberation scale. Those in the Maintenance stage obtained their highest scoreson the behavioral processeswhich are hypothesized to be most useful in maintaining behavior change. 4. Discussion

The results provide qualified support for the application of the stagemodel to the problem of polydrug use among methadone maintenance patients. The generalizability of the results may be limited in that the study sample over-represented disabled patients while under-representing patients with full-time jobs. However, comparison of the sample characteristics with demographic data from the clinics revealed no significant differences in other areas. These findings argue against the possibility of other selection biases being introduced by relying upon patient volunteers. Nonetheless, future studies might incorporate procedures to recruit more patients who work full-time. The results suggest that methadone maintenance patients do not discriminate among all twelve hypothesized processesof change. Instead, only four different types of change-relatedactivities could be identified: Reevaluation, Self Liberation, Reinforcing Relationships, and Behavioral Processes.Although these four scales representa significant revision of the original model, an attempt was made to keep respecifications consistent with previous research and theory, Still, it is possible that the final model capitalized on somechance relationships among items. Any future studies should include an effort to confirm these scaleson another sample. Ten processeswere representedby at least one or two items in the final four scales,though the remaining two scales (Medication and Social Liberation) were eliminated entirely. In retrospect, it was not surprising that the Medication items failed to load together. Two of the items involved behavior which is often encouraged (i.e., requesting a higher methadone dose to avoid drug use) while the other three items describe behavior which would be considered inappropriate in most clinics (e.g., using diierent kinds of street drugs to avoid using other illicit drugs). Medication use probably should have beenconceived as two separateprocesses- one involving the useof methadone and approved medications and the other involving the use of unauthorized drugs and alcohol. Social Liberation may simply not have great relevance for methadone maintenance patients. Prochaska et al. (1992a) define this process as “increasing alternatives for non-problem behaviors available in society“ (p. 37). Since polydrug use is generally considered socially unacceptable (more so, for instance, than smoking or eating) there may be little need to engagein this process.Regarding heroin use, for exam-

ple, there is no equivalent to advocating for ‘no smoking’ areas or for the rights of non-smokers. Subjects in different stages produced profiles of change-process scores largely consistent with predictions, though these scores did not distinguish stage categoriesas clearly as in previous research. In general, subjectsclassified into adjacent stagesof change did not differ significantly from each other on most of the process scales. The ambiguity of the results may be due partly to the complexity of methadone maintenance treatment itself. Patients may view methadone as an end in itself rather than as one of severalmeansfor changing behavior. They may have difficulty clearly identifying other processesof behavior change. Also, many patients in this study were polydrug users;they were asked about their intentions to quit all unauthorized drug use. Some patients might make use of different processesof change for different drugs. Thus, a patient might be actively working to avoid opiate use while still re-evaluating the need to stop using benzodiazepines. Such a patient might have difficulty articulating a clear pattern of change processesin response to questions about stopping all illicit drug use. All four of the processes of change subscales demonstrated relationships with self-reported drug use which were consistent with the predictions of the stage model. Subjects reporting less drug use in the past 30 days reported more frequent use of Self Liberation, Reinforcing Relationships, and Behavioral Processes. This suggeststhat those subjectswho were the most able to abstain from illicit drugs were those making conscious commitments to stop using drugs (Self Liberation), relying upon relationships with others who supported abstinence (Reinforcing Relationships), and finding alternatives to drug using behaviors while avoiding people and situations which were likely to trigger urges to use drugs (Behavioral Processes).Behavioral Processes accountedfor about 7% of the variance in reported drug use and washed out the associations between the other two process scales and drug use. It may be that Self Liberation and Reinforcing Relationships facilitate the use of the Behavioral Processeswhich have the most direct impact on drug use. Reevaluation, on the other hand, correlated positively with recent drug use after controlling for Behavioral Processesscores. Thus, patients involved in evaluating the negative consequences of their drug use without taking behavioral steps to addressthe problem reported more drug use. The study also identifies a group of patients, namely Precontemplators, who may require alternative interventions. Despite extended treatment tenures, most of thesepatients reported regular use of opiates and at least one other class of unauthorized drugs. They report relatively little change-related activity and no intention of quitting drugs in the near future. By contrast, many Prepared subjects reported that they had been in treat-

MA.

Belding et al. /Drug and Alcohol Dependence 3945 (1995) 45-55

ment for six months or less and almost two-thirds reported treatment tenures of less than two years. The results suggest that it may be important to facilitate behavior change early in treatment to capitalize on this apparent window of opportunity. However, it remains for longitudinal studies to determine whether differences in stageclassification and differential use of changeprocessesactually predict treatment outcome. Acknowledgment

This research was supported by grant number R18 DA0 6096 awarded by the National Institute on Drug Abuse. This support is gratefully acknowledged. References Ball, J., Lange, W.R., Myers, C.P. and Friedman, S.R. (1988) Reducing the risk of AIDS through methadone maintenance treatment. J. Health Sot. Behav. 29, 214-226. Belding, M.A. (1993) An analysis of stages and processesof change among opioid addicts in methadone maintenance treatment (Doctoral dissertation, Duke University, 1993).Dissertation Abstr. Int. 55 (2B), 9414119. Bentler, P.M. (1990) Comparative tit indexes in structural models. Psychol. Bull. 107, 238-246. Bentler, P.M. (1991) EQS/Mac (Version 3.0) (Computer software). BMDP Statistical Software, Inc., Los Angeles, CA. equations program manual. Bender, P.M. (1992) EQS StNCtUral BMDP Statistical Software, Inc., Los Angeles, CA. Bentler, P.M. and Bonett, D.G. (1980)Significance teatsand goodness of fit in the analysis of covariance structures. Psychol. Bull. 88, 588-606. DiClemente, C.C., Prochaska, J.O., Fairhurst, S.K., Velicer, W.F., Velasquez, M.M. and Rossi, J.S. (1991) The process of smoking

53

cessation: an analysis of precontemplation, contemplation, and preparation stages of change. J. Consult. Clin. Psychol. 59 (2), 295-304. Iguchi, M.Y., Handelsman, L., Bickel, W.K. and GrifEths, R.R. (1993)Benzodiazepineand sedative use/abuseby methadone maintenance clients. Drug Alcohol Depend. 32, 257-266. Magma, S., Goldsmith, D., Casriel, C., Goldstein, P.J. and Lipton, D.S. (1987) The validity of methadone clients’ self-reported drug use. Int. J. Addictions 22 (8), 727-749. Maisto, S.A., McKay, J.R. and Connors, G.J. (1990)Self-report issues in substanceabuse: state of the art and future directions. Behavioral Assess.12, 117-134. Marcus, B.H., Rossi, J.S., Selby, V.C., Niaura, R.S. and Abrams, D.B. (1992) The stages and processesof exercise adoption and maintenance in a worksite sample. Health Psychol. 11 (6), 386-395. Marsh, H.W., Balla, J.R. and McDonald, R.P. (1988)Goodness-of-tit indexes in confirmatory factor analysis: the effect of sample size. Psychol. Bull. 103 (3), 391-410. McLellan, A.T., Woody, G.E., Luborsky, L. and Goehl, L. (1988)Is the counselor an ‘active ingredient’ in substanceabuse rehabilitation? J. Nerv. Ment. Dis. 176 (7), 423-430. Prochaska,J.O. and DiClemente, CC. (1983) Stagesand processesof self-changeof smoking: toward an integrative model of change. J. Consult. Clin. Psychol. 51 (3), 390-395. Prochaska, J.O., DiClemente, CC. and Norcross, J.C. (1992a) In searchof how people change: applications to addictive behaviors. Am. Psychologist 47 (9), 1102- 1114. Prochaska, J.O., Norcross, J.C., Fowler, J.L., Follick, M.J. and Abrams, D.B. (1992b) Attendance and outcome in a work site weight control program: processesand stagesof change as process and predictor variables. Addictive Behav. 17, 35-45. Prochaska,J.O., Velicer, W.F., DiClemente, CC. and Fava, J. (1988) Measuring processesof change: applications to the cessation of smoking. J. Consult. Clin. Psychol. 56 (4), 520-528. Rosenbloom, D. (1991) A transtheoretical analysis of change among cocaine users. Unpublished doctoral dissertation, University of Rhode Island, Kingston, RI.