- Email: [email protected]
Different intake procedures
0740-5472/92 SS.00 + .OO Copyright 0 1992 Pergamon Press Ltd.
Journal of Suhs~anceAbuse Treatment, Vol. 9, pp. 53-50, 1992 Printed in the USA. All rights reserved.
INTERNATIONAL
PERSPECTIVE
Different Intake Procedures The Influence on Treatment Start and Treatment ResponseA Quasi-Experimental Study
EDLE RAVNDAL, Cand 8ocio1, AND PER VAGLUM, MD Department of Behavioural Sciences in Medicine, University of Oslo, Oslo, Norway
Abstract - Two hundred Norwegian substance abusers who consecutively applied for treatment in a hierarchical therapeutic community were divided into two different groups according to intake procedures: the intake group (IG) and the nonintake group (No-IG). Using a prospective design, we found that beginners in the program were more often infected with human immunodeficiency virus and that they used amphetamine more frequently and alcohol less frequently than nonbeginners. The type of intake procedure did not influence the percen rage of those who started in the program, but it increased the number of clients who completed Phase 1, the l-year inpatient phase of the program. Twice as many clients in the IG completed Phase 1 compared to those in the No-IG. From a clinical point of view, we conclude that the IG procedure should be offered to all applicants but that the model should be developed more as a role induction strategy, in which learning practical coping skills to adjust to the treatment program is essential. Keywords-intake procedures; substance abuse; treatment start; treatment response; hierarchical therapeutic community.
clients who really may profit from the treatment program. Different intake procedures may also influence treatment outcome in different ways. By demanding certain behaviors or participation in “intake groups” during the waiting time, one may start the process of socialization to the client role of the particular treatment program, thus providing clients with coping skills that increase their ability to adjust to the program. This may help clients to continue with their treatment. Unfortunately, there is a great lack of research in this area, and with the exception of some few studies using waiting list controls (Eriksen, 1986; Grenier, 1985), we do not know what the effects of different selection procedures are. The length of the waiting time before residential treatment seems not to be associated with treatment motivation (Brown et al., 1989). Ideally, an experimental design should be used when studying such questions. However, because of practical clinical problems, this was not feasible. Because so little research exists in the field, we chose to undertake the study but with a quasi-experimental design (Campbell & Stanley, 1%7). The study should,
INTRODUCTION AN INCREASING CONCERN for treatment administrators and planners has been the need to maintain often lengthy waiting lists in the face of a widespread demand for drug abuse treatment. In a recent survey from the United States, 75% of 42 cities reported using waiting lists of 7 to 26 weeks in duration (Brown, B.S., Hickey, J.E., Chung, A.S., Craig, R.D., & Jaffe, J.H., 1989). For some drug treatment modalities in Scandinavia, the demand for treatment is now far greater than the capacity. A possible consequence of this situation is the establishment of different intake procedures, which may influence selection of clients and outcome of treatment in different ways. Different intake procedures may be more or less unsuccessful in preventing general dropout during the intake process and in recruiting those
Requests for reprints should be addressed to Edle Ravndal, University of Oslo, Department of Behavioural Sciences in Medicine, P.O. Box 1111 Blindern, N-0317 Oslo 3, Norway.
53
54
E. Ravndal and P. Vaglum
therefore, be considered as a preliminary report. On this basis, we conducted a study of two different intake procedures at Phoenix House, Oslo, which is a hierarchical therapeutic community. Using a prospective design, we explored whether there were clinical characteristics that differentiated clients who dropped out during the intake process from those who started in treatment and whether the dropouts from the two intake procedures differed in such characteristics. Second, we explored whether the intake procedures managed to select those clients who were expected to profit the most from the treatment, namely, the clients with a high degree of acting out behaviors and a low degree of schizotypal features. Third, we investigated whether there were any differences between the two intake procedures in recruitment or treatment response.
thology, and current psychiatric symptomatology. The interview and the self-report instruments were administered by a trained interviewer and four assistants. Through a principal-component analysis of the scores on the 11 MCMI personality scales, two indices have been constructed: The “general psychopathology” index measures the degree of psychopathology. On the “dramatic-schizotypal” index, a high score indicates a high degree of dramatic (histrionic, antisocial, and narcissistic) personality features, whereas a low score indicates a high degree of schizotypal features. A combination of a high score on the dramatic-schizotypal index and a low score on the general psychopathology index was nonsignificantly related to completion of Phase 1 in the program. This variable will, therefore, be used as one criterion of high treatment suitability in this article.
MATERIALS AND METHODS The subjects were 200 Norwegian substance abusers who consecutively applied for treatment at Phoenix House, Oslo. This program is organized as in the Phoenix Houses in England and the United States (De Leon & Ziegenfuss, 1986) and is intended to treat substance abusers with long addictive careers. The total treatment program takes on average of 14 years; the inpatient period, Phase 1, takes 1 year. The mean age of the subjects was 27.5 years (SD = 4.7; range = 18 to 46); 3 1Vo were women. Drugs had been used for an average of 10.4 years (SD = 4.6; range = 1 to 22). Most of the subjects used a mixture of substances. During the last 6 months before intake, 78% had used opiates. The average reported consumption of pure alcohol per year was 26.0 L (SD = 41.6). Ninetythree percent of the sample were unemployed, and only 18% were married or living as married. Sixty-eight percent had previously received inpatient treatment at psychiatric institutions and/or at alcohol clinics. Sixtyseven percent had been in prison. Eighty-one percent had fathers who were skilled or unskilled laborers. Eighteen of the applicants (the “prisoners”) were directly transferred from prison to the treatment program to serve a sentence. Members of this group are excluded from the analysis in this article because of their special reason for entering treatment. All applicants first underwent a short intake interview with the staff. A personal structured interview followed that covered sociodemographic data, family background, education, employment, substance abuse, legal problems, social adjustment, treatment received, prostitution, and sexual assaults. In addition, they completed three self-report instruments: the Millon Clinical Multiaxial Inventory (MCMI; Millon, 1982), the Basic Character Inventory (Torgersen, 1980), and the Hopkins Symptom Checklist-90 (Derogatis, Lipman, & Covi, 1973), which assessed personality, psychopa-
THE INTAKE PROCEDURES The applicants were recruited from all over the country but predominantly from Oslo. Initially, all had to go through an intake interview with the staff. Then, based on geography, they were divided into two groups: 1. The Intake group (IG) living in Oslo or in the area near by (n = 147); 2. The nonintake group (No-IG) living at least 100 km outside of Oslo (n = 35). Because most of the applicants came from the Oslo region, unfortunately, there had to be a large size difference between the groups. The IG clients came to an information group twice to inform themselves about the program. If they decided to continue, they had to go through the intake interview and attend a group meeting 1 hr each week until space became available at the facility. All group meetings took place in a separate building away from the treatment center. The two group leaders, one staff member, and one senior resident were from Phoenix House. The official intention of the intake group was to further motivate the applicants to start treatment while they waited for entry. Therefore, the main tasks of the group meetings were to give more information about the program, to get the applicants to know each other, and to prepare the applicants both practically and mentally to enter treatment. At some meetings, but more or less arbitrarily, clients were trained in practical coping skills (i.e., confrontation techniques, vocal assertion) that should make it easier for them to adjust to the program. The intake group was not compulsory. However, absence from the group twice consecutively or more, and no change in substance abuse, which was only reported on by the clients themselves, would clearly delay entrance to treatment. Thus, a real, but not explicitly pronounced, goal of the intake group also was to select the most motivated clients.
55
Intake Procedures and Treatment
The No-IG group had to come to Oslo for information about the program and to take the intake interview at the same time. They then returned home and waited until they were offered a place to start the program. Because the treatment model presupposes no vacancy, all applicants in both intake procedures had to wait to enter treatment until some of the clients dropped out of the program or finished the inpatient phase. In the meantime, they were urged to stop taking drugs and prepare themselves both practically and mentally to enter treatment. Not any of the applicants were refused entry to treatment as long as they were able to follow the intake procedures and cut down on their substance abuse before intake.
TABLE 1 Background Variable8 Tested in the Bivarlato Analysis
RESULTS
Substance abuse history Median time daily drug abuse Median time daily alcohol abuse Liters of pure alcohol last year before intake’ Median number of overdoses Frequency of usea Alcohol Benzodiazepines Cannabis Amphetamines** Opiates LSD
Demographic Age Sex Social group Area of residence’ Marital statusa Education Employmenta Index of family stability during childhood Social functioninga Petersen’s index of social coping Social belonging Drug Taking Evaluation Scale (DTES) Legal involvement Median time in prison Index of negative life events
Differences Between Beginners and Nonbeginners Altogether, 128 (70%) of the 182 subjects started in the program. According to bivariate analysis, none of the background data except use of amphetamine (p < 0.01) and alcohol consumption the last year before intake (p < 0.05) differed significantly between beginners and nonbeginners (see Table 1). The more often a client had used amphetamine, the greater were the chances that he or she started treatment; however, the more alcohol the client had consumed, the less were the chances that he or she would start treatment. All but one of the human immunodeficiency virus (HIV)-infected applicants (n = 14) entered the program. Degree or type of psychopathology, measured by the MCMI, did not separate beginners from nonbeginners. Thus, the intake procedures did not select in a positive way those clients with a high treatment suitability. The differences between beginners and nonbeginners in each type of intake procedure was very much the same as between beginners and nonbeginners in the total sample. To investigate the predictive power of amphetamine, alcohol, HIV, and sex in relation to treatment start or not, a logistic regression analysis was performed. Alcohol and amphetamine were significant as predictors in the model, whereas the analysis produced an almost significant value for HIV and a nonsignificant beta value for sex. However, the predictive value of the model was low. To search for other possible combinations of variables that could yield more as predictors in a logistic regression model, we performed both a principal component analysis of all the substance abuse variables and a multiple regression analysis between a number of background variables and amphetamine, alcohol, and HIV. However, no other combination of variables
HIV positive Psychopathology Hopkins Symptom Checklist (SCL-90) Millon Clinical Multiaxial Inventory (MCMI) “General psychopathology” index (MCMI) “Dramatic-schizotypal” index (MCMI) Combination of the two MCMI indexes Basic Character Inventory (BCI) Median number of suicide attempts Median time inpatient treatment Median time outpatient treatment *Last 6 months before intake. ‘p < 0.05, l*p < 0.01.
separated beginners from nonbeginners in a better way than did the aforementioned model. Recruitment to the Two Intake Procedures: Differences in Background Variables A few background variables differed significantly between applicants recruited by the two different intake procedures (see Table 2). Applicants in the IG were older (p < 0.01); they had a more frequent use of opiates (p < 0.001) and a longer drug abuse history (p < 0.05) than the No-IG applicants. The No-IG applicants used more alcohol (p < 0.01). As for prior treatment attempts, there was no differences between the
56
E. Ravndal and P. Vaglum TABLE 2 Significant Group Differences on Background Variables (n = 182)
TABLE 3 Logistic Regression Model for Treatment Start
Predictor Variables Group 1
Age (years) Daily use of opiates (%) Drug abuse history (years) Alcohol consumptiona Daily use of alcohol (%)
SE
Odds Ratio8
Group 2
(n=147)
(n=35)
p
28 43
25
co.01
10.7 1.8 20
Estimate
(n = 182)
11 8.8 2.8 42
‘The numbers represent the natural logarithm of the total alcohol consumption in liters of pure alcohol during the last year before intake: 24.5 Land 32.8 L.
two groups of applicants. The median time in intake before admittance was 8.8 weeks (SD = 11.6; range = 1 to 57) for the IG and 12 weeks (SD = 19.9; range = 0 to 78) for the No-IG. To examine how strong the relationship was between individual background data and intake procedure, a logistic regression analysis was performed with the four variables that significantly separated the IG and the No-IG (age, alcohol, opiates, drug career). Except for drug career, all four predictors were significant, which implies that these three variables were good predictors for separating the intake procedures. The Influence of Intake Procedure for Treatment Start The percentages of beginners from the IG and No-IG were similar (70.1% and 71.4%). To explore the impact of the type of intake procedure for treatment start, a logistic regression analysis was completed with the three variables that best separated between beginners and nonbeginners (amphetamine, alcohol, HIV), a “type of intake procedure” variable (IG/No-IG) and the three significant predictors of the differences between the IG and the No-IG (opiates, alcohol, age). Because the last three variables produced very low beta values, they were excluded from the final model. When amphetamine, alcohol, and HIV were controlled for, type of intake procedure was not significant as a predictor in the model, showing that being HIV infected or not or having a history of heavy use of amphetamine or alcohol was more decisive for treatment start than the kind of intake procedure the applicants were offered (see Table 3). The strongest predictor of treatment start was alcohol consumption, and it was negative. When dividing that variable into categories of less than and more than 60 L alcohol per year and when controlling for the other predictors, the odds ratio for alcohol was 0.18, meaning that the odds for treatment start for appli-
Amphetamine (low-high) Alcohol (low-high) Group procedure HIV (negative-positive)
1.007 -1.730 -0.1163 1.963
0.3997 0.5002 0.4450
2.74 0.18 0.89
1.085
7.12
HIV = human immunodeficiency virus. ‘The odds ratio, a measure of association, is the ratio of two odds. Odds are themselves ratios of the number of events to the number of nonevents. An odds ratio greater than 1 .O indicates an increased likelihood of the event occurring (i.e., starting treatment), whereas an odds ratio less than 1 .O indicates a decreased likelihood of the event occurring (Morgan 8 Teachman, 1988).
cants who consumed 60 L or more alcohol per year was just 0.18 compared with those who consumed less than 60 L. However, the probability of starting treatment was highest among the HIV-infected applicants with an odds ratio of 7.12. When using a cutoff point at 0.6, the predicted value of the model was 73% and was, thus, no better than the “natural selection” with 72% beginners. The Influence of Type of Intake Procedure for Completion of Phase 1 Only 15.4% (n = 4) of the beginners (n = 26) in the No-IG completed Phase 1 compared to 32.4% (n = 33) of the beginners (n = 102) in the IG. The difference was not significant. To explore the influence of type of intake procedure for completion of Phase 1, another logistic regression analysis was performed. Because there was a positive correlation between amphetamine and completing Phase 1 (p < 0.05) and a negative correlation between alcohol and completing Phase 1 (p < 0.05) (Ravndal & Vaglum, 1990), these variables together with HIV and the type of intake procedure variable were entered in the regression equation. The probability for completing Phase 1 increased for dramatic clients with minimal schizotypal features when, in addition, their score on the general psychopathology index was below the median. Therefore, this variable also had to be controlled for in the analysis. The individual background variables that were significantly different in the two groups were still not significant as predictors in this analysis and were, thus, excluded from the final model. The same was the case with the alcohol predictor. As shown in Table 4, when controlling for amphetamine, HIV, and the psychopathology variable, the beta value for type of intake procedure was not significant, although its influence on completing Phase
57
Intake Procedures and Treatment TABLE 4 Logistic Regression Model for Completion of Phase 1 (n = 182) Estimate
Predictor Variables Amphetamine (low-high) Group procedure HIV (negative-positive) High on the dramaticschizotypal index and low on the general psychopathology index
SE
Odds Ratio
1.214 1.103
0.4279 0.6195
3.37 3.01
1.196
0.6462
3.31
0.9716
0.5519
2.64
HIV = humanimmunodeficiency virus.
1 or not was much stronger than for treatment start. The odds ratio for completing Phase 1 was approximately the same for heavy users of amphetamine (3.37), for HIV-infected clients (3.31), and for clients who had attended the intake group before entering treatment (3.01). The odds ratio for the psychopathology variable was somewhat less (2.64) (see Table 4). DISCUSSION
The more often an applicant had used amphetamine, the greater were the chances that he or she would start treatment, whereas the more alcohol the client had consumed, the less were the chances for treatment start. This means that independent of intake procedure the most heavy users of alcohol were “naturally” selected out. From earlier studies in the field, we know that the heaviest users of alcohol have bad prognoses (Hammer, Ravndal, & Vaglum, 1985; Sang, 1978; Vaglum, 1979), which implies that this selection was a positive one in relation to treatment outcome. The selection of more applicants with heavy use of amphetamine is more difficult to interpret, but on the basis of the results from an earlier study (Ravndal & Vaglum, 1991), this also seems to be a positive selection for treatment outcome because heavy users of amphetamine seem to complete Phase 1 more often than the less addicted amphetamine users. All but one of the HIV-infected applicants entered the program, which might also be considered positive in relation to the number that complete Phase 1. However, used as predictors for treatment start in a logistic regression model, these three variables were no better predictors of beginners and nonbeginners than the “natural selection,” which means that the statistical importance of the variables was limited. However, the clinical value of these correlations may still be important. As for recruitment to the two intake procedures, there were differences in background variables. However, after controlling for these variables through lo-
gistic regression analysis, we found that they were not good predictors for treatment start. The intake procedures seemed not to have positively selected those who were most suited for the treatment program based on psychopathology, and type of intake procedure had no influence on the number of beginners from each group. The only possible influence that type of intake procedure might have was to increase the number of clients who completed Phase 1 of the program. Twice as many clients who had attended the IG completed Phase 1 compared to the No-IG clients. Clinically, this difference is interesting, even though it was not statistically significant. However, this might be because the number of No-IG clients was relatively small. On the other hand, because of all the other factors that operate in the intake process, one cannot expect the intake procedures to have too great an impact, either for treatment start or for treatment outcome. The odds ratio for completing Phase 1 was approximately the same for HIV-infected clients, clients with a heavy use of amphetamine, and clients attending the IG. It was unexpected that applicants from the No-IG started treatment as often as applicants from the IG because they had no organized intake group to attend while waiting. Also, because they were younger and had histories of heavier alcohol consumption than did IG applicants, they were expected to be more exposed to “failure,” both during the waiting period and in the program. On the other hand, we know little about the local network and possible support to start treatment for this group outside of Oslo. If the primary aim of the IG was to reduce the dropout rate during the waiting time, the IG may seem to be a waste of time. However, because this is a quasi-experimental study, we cannot completely rule out the possibility that the older and more heavily addicted IG clients would have had an even higher dropout rate without this intake procedure. It is also possible that this perhaps more dysfunctional city population stayed longer in treatment because of the severity of their addiction and because of their somewhat higher scores on the dramatic psychopathology dimension. One should, therefore, study these questions in a true experiment in which only applicants from the same area took part. Although the results were not statistically significant, the odds for completing Phase 1 were three times higher in the IG than in the No-IG, and taking part in the IG was as important for completing Phase 1 as being HIV-infected or having a history of heavy amphetamine use. A possible explanation might be that the IG clients had been somewhat better socialized into the client role by attending the IG, implying that they had learned coping skills necessary for adjusting to the program. If this finding can be replicated, offering an IG to applicants would seem to be worthwhile. However, a controlled study of role induction in a commu-
58
nity outpatient drug treatment center found’almost the opposite of our results (Stark & Kane, 1985). Clients who received role induction procedures were more likely than controls to start treatment, but analysis of attendance rates after 3 months showed that role induction had no effect on continuation in treatment. Zweben and Li (1981), however, found results similar to ours. Clients who were more favorably disposed to an outpatient program were more likely to remain in care than those who were not, but only if role induction was provided. Because both of these studies are from outpatient settings, the results are not easily comparable with the results from this study. However, it does seem that clients who attended the IG were in some way more suitable for the program. This may be due to reduced cognitive inaccuracies regarding the nature of the treatment process (Heitler, 1976), but it may also be a result of an established relationship to the staff, which modifies the stress of changing to an inpatient role. This question needs further study, also with qualitative data. From a clinical point of view, however, it seems appropriate to conclude that this kind of intake group procedure should continue, but that the model should probably be developed more as a role induction strategy in which learning practical coping skills to adjust to the treatment program is essential. From a cost-benefit analysis, such an intake procedure would far outweigh the costs if the results were similar to our findings that twice as many clients who had attended the IG completed Phase 1. Certain limitations of the present study need to be emphasized. The limited number of clients in the NoIG increases the probability of Type II error. A quasiexperimental design also raises questions of both validity and replicability. Therefore, the findings should be regarded mainly as generation of a hypothesis, and the study should be replicated with an experimental design. In this way, the study may be a starting point for further research in the area. However, this is the first prospective study that has systematically studied intake procedures and their impact on treatment start and treatment outcome for inpatients after 1 year. The treatment model is a well-defined and in-
E. Ravndal and P. Vaglum
ternationally known model; therefore, these preliminary results may be of interest to similar programs. REFERENCES Brown, B.S., Hickey, J.E., Chung, A.S., Craig, R.D., & Jaffe, J.H. (1989). The functioning of individuals on a drug abuse treatment waiting list. American Journal of Drug and Alcohol Abuse, 15, 261-274. Campbell, D.T., & Stanley, J.C. (1%7). Experitnentat and quasieXperimenta designs for research. Chicago: Rand McNally. De Leon, G., & Ziegenfuss, J.T. (1986). Therapeutic communities for addictions. Springfield, IL: Charles C Thomas. Derogatis, L.R., Lipman, R.S., & Covi, L. (1973). An outpatient psychiatric rating scale-Preliminary report. Psychopharmacofogy Bulletin, 9, 13-28. Eriksen, L. (1986). The effect of waiting for inpatient alcoholism treatment after detoxification: An experimental comparison between inpatient treatment and advice only. Addictive Behaviors, 11,389-397. Heitler, J.B. (1976). Preparatory techniques in initiating expressive psychotherapy with lower-class, unsophisticated patients. Psychological Buttetins, 83, 339-352. Hammer, T., Ravndal. E., & Vaglum, P. (1985). Work is not enough: A quasi-experimental study of a vocational training programme for young drug and alcohol abusers. Journal of Drug Issues, 15, 393-403. Grenier, C. (1985). Treatment effectiveness in an adolescent chemical dependency treatment program: A quasi-experimental design. International Journal of the Addictions, 20, 381-391. Millon, T. (1982). The Milton Clinical Multiaxial Inventory manual (2nd ed.). Minneapolis, MN: National Computer Systems. Morgan, S.P., & Teachman, J.D. (1988). Logistic regression: Description, examples, and comparison. Journal of Marriage and the Family, 50, 929-936. Ravndal, E., & Vaglum, P. (1991). Psychopathology and substance abuse as predictors of program completion in a drug abuse therapeutic community. A prospective study. Acta Psychiattica Scandinavica, 83, 217-222. Stang, H.J. (1978). Ungdom pa drift. Oslo: Universitetsforlaget. Stark, M.J., &Kane, B.J. (1985). General and specific psychotherapy role induction with substance-abusing clients. International Journal of the Addictions, 28, 1135-l 141. Torgersen, S. (1980). The oral, obsessive and hysterical personality syndromes. Archives of General Psychiatry, 37, 1272-1277. Vaglum, P. (1979). Unge stoffmksbrukere i et terapeutisk samfunn. Oslo: Universitetsforlaget. Zweben, A., & Li, S. (1981). The efficacy of role induction in preventing early dropout from outpatient treatment of drug dependency. American Journal of Drug and Alcohol Abuse, 8, 171-183.
Journal of Suhs~anceAbuse Treatment, Vol. 9, pp. 53-50, 1992 Printed in the USA. All rights reserved.
INTERNATIONAL
PERSPECTIVE
Different Intake Procedures The Influence on Treatment Start and Treatment ResponseA Quasi-Experimental Study
EDLE RAVNDAL, Cand 8ocio1, AND PER VAGLUM, MD Department of Behavioural Sciences in Medicine, University of Oslo, Oslo, Norway
Abstract - Two hundred Norwegian substance abusers who consecutively applied for treatment in a hierarchical therapeutic community were divided into two different groups according to intake procedures: the intake group (IG) and the nonintake group (No-IG). Using a prospective design, we found that beginners in the program were more often infected with human immunodeficiency virus and that they used amphetamine more frequently and alcohol less frequently than nonbeginners. The type of intake procedure did not influence the percen rage of those who started in the program, but it increased the number of clients who completed Phase 1, the l-year inpatient phase of the program. Twice as many clients in the IG completed Phase 1 compared to those in the No-IG. From a clinical point of view, we conclude that the IG procedure should be offered to all applicants but that the model should be developed more as a role induction strategy, in which learning practical coping skills to adjust to the treatment program is essential. Keywords-intake procedures; substance abuse; treatment start; treatment response; hierarchical therapeutic community.
clients who really may profit from the treatment program. Different intake procedures may also influence treatment outcome in different ways. By demanding certain behaviors or participation in “intake groups” during the waiting time, one may start the process of socialization to the client role of the particular treatment program, thus providing clients with coping skills that increase their ability to adjust to the program. This may help clients to continue with their treatment. Unfortunately, there is a great lack of research in this area, and with the exception of some few studies using waiting list controls (Eriksen, 1986; Grenier, 1985), we do not know what the effects of different selection procedures are. The length of the waiting time before residential treatment seems not to be associated with treatment motivation (Brown et al., 1989). Ideally, an experimental design should be used when studying such questions. However, because of practical clinical problems, this was not feasible. Because so little research exists in the field, we chose to undertake the study but with a quasi-experimental design (Campbell & Stanley, 1%7). The study should,
INTRODUCTION AN INCREASING CONCERN for treatment administrators and planners has been the need to maintain often lengthy waiting lists in the face of a widespread demand for drug abuse treatment. In a recent survey from the United States, 75% of 42 cities reported using waiting lists of 7 to 26 weeks in duration (Brown, B.S., Hickey, J.E., Chung, A.S., Craig, R.D., & Jaffe, J.H., 1989). For some drug treatment modalities in Scandinavia, the demand for treatment is now far greater than the capacity. A possible consequence of this situation is the establishment of different intake procedures, which may influence selection of clients and outcome of treatment in different ways. Different intake procedures may be more or less unsuccessful in preventing general dropout during the intake process and in recruiting those
Requests for reprints should be addressed to Edle Ravndal, University of Oslo, Department of Behavioural Sciences in Medicine, P.O. Box 1111 Blindern, N-0317 Oslo 3, Norway.
53
54
E. Ravndal and P. Vaglum
therefore, be considered as a preliminary report. On this basis, we conducted a study of two different intake procedures at Phoenix House, Oslo, which is a hierarchical therapeutic community. Using a prospective design, we explored whether there were clinical characteristics that differentiated clients who dropped out during the intake process from those who started in treatment and whether the dropouts from the two intake procedures differed in such characteristics. Second, we explored whether the intake procedures managed to select those clients who were expected to profit the most from the treatment, namely, the clients with a high degree of acting out behaviors and a low degree of schizotypal features. Third, we investigated whether there were any differences between the two intake procedures in recruitment or treatment response.
thology, and current psychiatric symptomatology. The interview and the self-report instruments were administered by a trained interviewer and four assistants. Through a principal-component analysis of the scores on the 11 MCMI personality scales, two indices have been constructed: The “general psychopathology” index measures the degree of psychopathology. On the “dramatic-schizotypal” index, a high score indicates a high degree of dramatic (histrionic, antisocial, and narcissistic) personality features, whereas a low score indicates a high degree of schizotypal features. A combination of a high score on the dramatic-schizotypal index and a low score on the general psychopathology index was nonsignificantly related to completion of Phase 1 in the program. This variable will, therefore, be used as one criterion of high treatment suitability in this article.
MATERIALS AND METHODS The subjects were 200 Norwegian substance abusers who consecutively applied for treatment at Phoenix House, Oslo. This program is organized as in the Phoenix Houses in England and the United States (De Leon & Ziegenfuss, 1986) and is intended to treat substance abusers with long addictive careers. The total treatment program takes on average of 14 years; the inpatient period, Phase 1, takes 1 year. The mean age of the subjects was 27.5 years (SD = 4.7; range = 18 to 46); 3 1Vo were women. Drugs had been used for an average of 10.4 years (SD = 4.6; range = 1 to 22). Most of the subjects used a mixture of substances. During the last 6 months before intake, 78% had used opiates. The average reported consumption of pure alcohol per year was 26.0 L (SD = 41.6). Ninetythree percent of the sample were unemployed, and only 18% were married or living as married. Sixty-eight percent had previously received inpatient treatment at psychiatric institutions and/or at alcohol clinics. Sixtyseven percent had been in prison. Eighty-one percent had fathers who were skilled or unskilled laborers. Eighteen of the applicants (the “prisoners”) were directly transferred from prison to the treatment program to serve a sentence. Members of this group are excluded from the analysis in this article because of their special reason for entering treatment. All applicants first underwent a short intake interview with the staff. A personal structured interview followed that covered sociodemographic data, family background, education, employment, substance abuse, legal problems, social adjustment, treatment received, prostitution, and sexual assaults. In addition, they completed three self-report instruments: the Millon Clinical Multiaxial Inventory (MCMI; Millon, 1982), the Basic Character Inventory (Torgersen, 1980), and the Hopkins Symptom Checklist-90 (Derogatis, Lipman, & Covi, 1973), which assessed personality, psychopa-
THE INTAKE PROCEDURES The applicants were recruited from all over the country but predominantly from Oslo. Initially, all had to go through an intake interview with the staff. Then, based on geography, they were divided into two groups: 1. The Intake group (IG) living in Oslo or in the area near by (n = 147); 2. The nonintake group (No-IG) living at least 100 km outside of Oslo (n = 35). Because most of the applicants came from the Oslo region, unfortunately, there had to be a large size difference between the groups. The IG clients came to an information group twice to inform themselves about the program. If they decided to continue, they had to go through the intake interview and attend a group meeting 1 hr each week until space became available at the facility. All group meetings took place in a separate building away from the treatment center. The two group leaders, one staff member, and one senior resident were from Phoenix House. The official intention of the intake group was to further motivate the applicants to start treatment while they waited for entry. Therefore, the main tasks of the group meetings were to give more information about the program, to get the applicants to know each other, and to prepare the applicants both practically and mentally to enter treatment. At some meetings, but more or less arbitrarily, clients were trained in practical coping skills (i.e., confrontation techniques, vocal assertion) that should make it easier for them to adjust to the program. The intake group was not compulsory. However, absence from the group twice consecutively or more, and no change in substance abuse, which was only reported on by the clients themselves, would clearly delay entrance to treatment. Thus, a real, but not explicitly pronounced, goal of the intake group also was to select the most motivated clients.
55
Intake Procedures and Treatment
The No-IG group had to come to Oslo for information about the program and to take the intake interview at the same time. They then returned home and waited until they were offered a place to start the program. Because the treatment model presupposes no vacancy, all applicants in both intake procedures had to wait to enter treatment until some of the clients dropped out of the program or finished the inpatient phase. In the meantime, they were urged to stop taking drugs and prepare themselves both practically and mentally to enter treatment. Not any of the applicants were refused entry to treatment as long as they were able to follow the intake procedures and cut down on their substance abuse before intake.
TABLE 1 Background Variable8 Tested in the Bivarlato Analysis
RESULTS
Substance abuse history Median time daily drug abuse Median time daily alcohol abuse Liters of pure alcohol last year before intake’ Median number of overdoses Frequency of usea Alcohol Benzodiazepines Cannabis Amphetamines** Opiates LSD
Demographic Age Sex Social group Area of residence’ Marital statusa Education Employmenta Index of family stability during childhood Social functioninga Petersen’s index of social coping Social belonging Drug Taking Evaluation Scale (DTES) Legal involvement Median time in prison Index of negative life events
Differences Between Beginners and Nonbeginners Altogether, 128 (70%) of the 182 subjects started in the program. According to bivariate analysis, none of the background data except use of amphetamine (p < 0.01) and alcohol consumption the last year before intake (p < 0.05) differed significantly between beginners and nonbeginners (see Table 1). The more often a client had used amphetamine, the greater were the chances that he or she started treatment; however, the more alcohol the client had consumed, the less were the chances that he or she would start treatment. All but one of the human immunodeficiency virus (HIV)-infected applicants (n = 14) entered the program. Degree or type of psychopathology, measured by the MCMI, did not separate beginners from nonbeginners. Thus, the intake procedures did not select in a positive way those clients with a high treatment suitability. The differences between beginners and nonbeginners in each type of intake procedure was very much the same as between beginners and nonbeginners in the total sample. To investigate the predictive power of amphetamine, alcohol, HIV, and sex in relation to treatment start or not, a logistic regression analysis was performed. Alcohol and amphetamine were significant as predictors in the model, whereas the analysis produced an almost significant value for HIV and a nonsignificant beta value for sex. However, the predictive value of the model was low. To search for other possible combinations of variables that could yield more as predictors in a logistic regression model, we performed both a principal component analysis of all the substance abuse variables and a multiple regression analysis between a number of background variables and amphetamine, alcohol, and HIV. However, no other combination of variables
HIV positive Psychopathology Hopkins Symptom Checklist (SCL-90) Millon Clinical Multiaxial Inventory (MCMI) “General psychopathology” index (MCMI) “Dramatic-schizotypal” index (MCMI) Combination of the two MCMI indexes Basic Character Inventory (BCI) Median number of suicide attempts Median time inpatient treatment Median time outpatient treatment *Last 6 months before intake. ‘p < 0.05, l*p < 0.01.
separated beginners from nonbeginners in a better way than did the aforementioned model. Recruitment to the Two Intake Procedures: Differences in Background Variables A few background variables differed significantly between applicants recruited by the two different intake procedures (see Table 2). Applicants in the IG were older (p < 0.01); they had a more frequent use of opiates (p < 0.001) and a longer drug abuse history (p < 0.05) than the No-IG applicants. The No-IG applicants used more alcohol (p < 0.01). As for prior treatment attempts, there was no differences between the
56
E. Ravndal and P. Vaglum TABLE 2 Significant Group Differences on Background Variables (n = 182)
TABLE 3 Logistic Regression Model for Treatment Start
Predictor Variables Group 1
Age (years) Daily use of opiates (%) Drug abuse history (years) Alcohol consumptiona Daily use of alcohol (%)
SE
Odds Ratio8
Group 2
(n=147)
(n=35)
p
28 43
25
co.01
10.7 1.8 20
Estimate
(n = 182)
11 8.8 2.8 42
‘The numbers represent the natural logarithm of the total alcohol consumption in liters of pure alcohol during the last year before intake: 24.5 Land 32.8 L.
two groups of applicants. The median time in intake before admittance was 8.8 weeks (SD = 11.6; range = 1 to 57) for the IG and 12 weeks (SD = 19.9; range = 0 to 78) for the No-IG. To examine how strong the relationship was between individual background data and intake procedure, a logistic regression analysis was performed with the four variables that significantly separated the IG and the No-IG (age, alcohol, opiates, drug career). Except for drug career, all four predictors were significant, which implies that these three variables were good predictors for separating the intake procedures. The Influence of Intake Procedure for Treatment Start The percentages of beginners from the IG and No-IG were similar (70.1% and 71.4%). To explore the impact of the type of intake procedure for treatment start, a logistic regression analysis was completed with the three variables that best separated between beginners and nonbeginners (amphetamine, alcohol, HIV), a “type of intake procedure” variable (IG/No-IG) and the three significant predictors of the differences between the IG and the No-IG (opiates, alcohol, age). Because the last three variables produced very low beta values, they were excluded from the final model. When amphetamine, alcohol, and HIV were controlled for, type of intake procedure was not significant as a predictor in the model, showing that being HIV infected or not or having a history of heavy use of amphetamine or alcohol was more decisive for treatment start than the kind of intake procedure the applicants were offered (see Table 3). The strongest predictor of treatment start was alcohol consumption, and it was negative. When dividing that variable into categories of less than and more than 60 L alcohol per year and when controlling for the other predictors, the odds ratio for alcohol was 0.18, meaning that the odds for treatment start for appli-
Amphetamine (low-high) Alcohol (low-high) Group procedure HIV (negative-positive)
1.007 -1.730 -0.1163 1.963
0.3997 0.5002 0.4450
2.74 0.18 0.89
1.085
7.12
HIV = human immunodeficiency virus. ‘The odds ratio, a measure of association, is the ratio of two odds. Odds are themselves ratios of the number of events to the number of nonevents. An odds ratio greater than 1 .O indicates an increased likelihood of the event occurring (i.e., starting treatment), whereas an odds ratio less than 1 .O indicates a decreased likelihood of the event occurring (Morgan 8 Teachman, 1988).
cants who consumed 60 L or more alcohol per year was just 0.18 compared with those who consumed less than 60 L. However, the probability of starting treatment was highest among the HIV-infected applicants with an odds ratio of 7.12. When using a cutoff point at 0.6, the predicted value of the model was 73% and was, thus, no better than the “natural selection” with 72% beginners. The Influence of Type of Intake Procedure for Completion of Phase 1 Only 15.4% (n = 4) of the beginners (n = 26) in the No-IG completed Phase 1 compared to 32.4% (n = 33) of the beginners (n = 102) in the IG. The difference was not significant. To explore the influence of type of intake procedure for completion of Phase 1, another logistic regression analysis was performed. Because there was a positive correlation between amphetamine and completing Phase 1 (p < 0.05) and a negative correlation between alcohol and completing Phase 1 (p < 0.05) (Ravndal & Vaglum, 1990), these variables together with HIV and the type of intake procedure variable were entered in the regression equation. The probability for completing Phase 1 increased for dramatic clients with minimal schizotypal features when, in addition, their score on the general psychopathology index was below the median. Therefore, this variable also had to be controlled for in the analysis. The individual background variables that were significantly different in the two groups were still not significant as predictors in this analysis and were, thus, excluded from the final model. The same was the case with the alcohol predictor. As shown in Table 4, when controlling for amphetamine, HIV, and the psychopathology variable, the beta value for type of intake procedure was not significant, although its influence on completing Phase
57
Intake Procedures and Treatment TABLE 4 Logistic Regression Model for Completion of Phase 1 (n = 182) Estimate
Predictor Variables Amphetamine (low-high) Group procedure HIV (negative-positive) High on the dramaticschizotypal index and low on the general psychopathology index
SE
Odds Ratio
1.214 1.103
0.4279 0.6195
3.37 3.01
1.196
0.6462
3.31
0.9716
0.5519
2.64
HIV = humanimmunodeficiency virus.
1 or not was much stronger than for treatment start. The odds ratio for completing Phase 1 was approximately the same for heavy users of amphetamine (3.37), for HIV-infected clients (3.31), and for clients who had attended the intake group before entering treatment (3.01). The odds ratio for the psychopathology variable was somewhat less (2.64) (see Table 4). DISCUSSION
The more often an applicant had used amphetamine, the greater were the chances that he or she would start treatment, whereas the more alcohol the client had consumed, the less were the chances for treatment start. This means that independent of intake procedure the most heavy users of alcohol were “naturally” selected out. From earlier studies in the field, we know that the heaviest users of alcohol have bad prognoses (Hammer, Ravndal, & Vaglum, 1985; Sang, 1978; Vaglum, 1979), which implies that this selection was a positive one in relation to treatment outcome. The selection of more applicants with heavy use of amphetamine is more difficult to interpret, but on the basis of the results from an earlier study (Ravndal & Vaglum, 1991), this also seems to be a positive selection for treatment outcome because heavy users of amphetamine seem to complete Phase 1 more often than the less addicted amphetamine users. All but one of the HIV-infected applicants entered the program, which might also be considered positive in relation to the number that complete Phase 1. However, used as predictors for treatment start in a logistic regression model, these three variables were no better predictors of beginners and nonbeginners than the “natural selection,” which means that the statistical importance of the variables was limited. However, the clinical value of these correlations may still be important. As for recruitment to the two intake procedures, there were differences in background variables. However, after controlling for these variables through lo-
gistic regression analysis, we found that they were not good predictors for treatment start. The intake procedures seemed not to have positively selected those who were most suited for the treatment program based on psychopathology, and type of intake procedure had no influence on the number of beginners from each group. The only possible influence that type of intake procedure might have was to increase the number of clients who completed Phase 1 of the program. Twice as many clients who had attended the IG completed Phase 1 compared to the No-IG clients. Clinically, this difference is interesting, even though it was not statistically significant. However, this might be because the number of No-IG clients was relatively small. On the other hand, because of all the other factors that operate in the intake process, one cannot expect the intake procedures to have too great an impact, either for treatment start or for treatment outcome. The odds ratio for completing Phase 1 was approximately the same for HIV-infected clients, clients with a heavy use of amphetamine, and clients attending the IG. It was unexpected that applicants from the No-IG started treatment as often as applicants from the IG because they had no organized intake group to attend while waiting. Also, because they were younger and had histories of heavier alcohol consumption than did IG applicants, they were expected to be more exposed to “failure,” both during the waiting period and in the program. On the other hand, we know little about the local network and possible support to start treatment for this group outside of Oslo. If the primary aim of the IG was to reduce the dropout rate during the waiting time, the IG may seem to be a waste of time. However, because this is a quasi-experimental study, we cannot completely rule out the possibility that the older and more heavily addicted IG clients would have had an even higher dropout rate without this intake procedure. It is also possible that this perhaps more dysfunctional city population stayed longer in treatment because of the severity of their addiction and because of their somewhat higher scores on the dramatic psychopathology dimension. One should, therefore, study these questions in a true experiment in which only applicants from the same area took part. Although the results were not statistically significant, the odds for completing Phase 1 were three times higher in the IG than in the No-IG, and taking part in the IG was as important for completing Phase 1 as being HIV-infected or having a history of heavy amphetamine use. A possible explanation might be that the IG clients had been somewhat better socialized into the client role by attending the IG, implying that they had learned coping skills necessary for adjusting to the program. If this finding can be replicated, offering an IG to applicants would seem to be worthwhile. However, a controlled study of role induction in a commu-
58
nity outpatient drug treatment center found’almost the opposite of our results (Stark & Kane, 1985). Clients who received role induction procedures were more likely than controls to start treatment, but analysis of attendance rates after 3 months showed that role induction had no effect on continuation in treatment. Zweben and Li (1981), however, found results similar to ours. Clients who were more favorably disposed to an outpatient program were more likely to remain in care than those who were not, but only if role induction was provided. Because both of these studies are from outpatient settings, the results are not easily comparable with the results from this study. However, it does seem that clients who attended the IG were in some way more suitable for the program. This may be due to reduced cognitive inaccuracies regarding the nature of the treatment process (Heitler, 1976), but it may also be a result of an established relationship to the staff, which modifies the stress of changing to an inpatient role. This question needs further study, also with qualitative data. From a clinical point of view, however, it seems appropriate to conclude that this kind of intake group procedure should continue, but that the model should probably be developed more as a role induction strategy in which learning practical coping skills to adjust to the treatment program is essential. From a cost-benefit analysis, such an intake procedure would far outweigh the costs if the results were similar to our findings that twice as many clients who had attended the IG completed Phase 1. Certain limitations of the present study need to be emphasized. The limited number of clients in the NoIG increases the probability of Type II error. A quasiexperimental design also raises questions of both validity and replicability. Therefore, the findings should be regarded mainly as generation of a hypothesis, and the study should be replicated with an experimental design. In this way, the study may be a starting point for further research in the area. However, this is the first prospective study that has systematically studied intake procedures and their impact on treatment start and treatment outcome for inpatients after 1 year. The treatment model is a well-defined and in-
E. Ravndal and P. Vaglum
ternationally known model; therefore, these preliminary results may be of interest to similar programs. REFERENCES Brown, B.S., Hickey, J.E., Chung, A.S., Craig, R.D., & Jaffe, J.H. (1989). The functioning of individuals on a drug abuse treatment waiting list. American Journal of Drug and Alcohol Abuse, 15, 261-274. Campbell, D.T., & Stanley, J.C. (1%7). Experitnentat and quasieXperimenta designs for research. Chicago: Rand McNally. De Leon, G., & Ziegenfuss, J.T. (1986). Therapeutic communities for addictions. Springfield, IL: Charles C Thomas. Derogatis, L.R., Lipman, R.S., & Covi, L. (1973). An outpatient psychiatric rating scale-Preliminary report. Psychopharmacofogy Bulletin, 9, 13-28. Eriksen, L. (1986). The effect of waiting for inpatient alcoholism treatment after detoxification: An experimental comparison between inpatient treatment and advice only. Addictive Behaviors, 11,389-397. Heitler, J.B. (1976). Preparatory techniques in initiating expressive psychotherapy with lower-class, unsophisticated patients. Psychological Buttetins, 83, 339-352. Hammer, T., Ravndal. E., & Vaglum, P. (1985). Work is not enough: A quasi-experimental study of a vocational training programme for young drug and alcohol abusers. Journal of Drug Issues, 15, 393-403. Grenier, C. (1985). Treatment effectiveness in an adolescent chemical dependency treatment program: A quasi-experimental design. International Journal of the Addictions, 20, 381-391. Millon, T. (1982). The Milton Clinical Multiaxial Inventory manual (2nd ed.). Minneapolis, MN: National Computer Systems. Morgan, S.P., & Teachman, J.D. (1988). Logistic regression: Description, examples, and comparison. Journal of Marriage and the Family, 50, 929-936. Ravndal, E., & Vaglum, P. (1991). Psychopathology and substance abuse as predictors of program completion in a drug abuse therapeutic community. A prospective study. Acta Psychiattica Scandinavica, 83, 217-222. Stang, H.J. (1978). Ungdom pa drift. Oslo: Universitetsforlaget. Stark, M.J., &Kane, B.J. (1985). General and specific psychotherapy role induction with substance-abusing clients. International Journal of the Addictions, 28, 1135-l 141. Torgersen, S. (1980). The oral, obsessive and hysterical personality syndromes. Archives of General Psychiatry, 37, 1272-1277. Vaglum, P. (1979). Unge stoffmksbrukere i et terapeutisk samfunn. Oslo: Universitetsforlaget. Zweben, A., & Li, S. (1981). The efficacy of role induction in preventing early dropout from outpatient treatment of drug dependency. American Journal of Drug and Alcohol Abuse, 8, 171-183.