{Received 24 June 1996; in revised form 7 November 1996; accepted 11 November 1996)

Transcription

1 Alcohol & Alcoholism Vol. 32, No. 3, pp , 1997 THE RELATIONSHIP BETWEEN HELPING ALLIANCE AND OUTCOME IN OUTPATIENT TREATMENT OF ALCOHOLICS: A COMPARATIVE STUDY OF PSYCHIATRIC TREATMENT AND MULTIMODAL BEHAVIOURAL THERAPY AGNETA OJEHAGEN*, MATS BERGLUND 1 and LARS HANSSON Department of Neuroscience, Division of Psychiatry, University Hospital of Lund, S Lund, Sweden and 'Department of Alcohol Diseases, MalmS General Hospital, Malmo, Sweden {Received 24 June 1996; in revised form 7 November 1996; accepted 11 November 1996) Abstract During the last decades, a positive relation between a good alliance and a successful therapy outcome has been demonstrated across a variety of different therapeutic modalities. The relationship between alliance and drinking outcome in long-term treatment of alcoholics (12 months or more) has not, as far as we know, been presented. In the present study, alcoholics were randomized to two outpatient treatment programmes; multimodal behavioural therapy (MBT) and psychiatric treatment based on a psychodynamic approach (PT). As part of the study, analyses were performed concerning differences in alliance between the two treatment models (MBT, n = 17; PT, n = 18), and concerning the relationship between alliance and treatment outcome. A Swedish version of the Helping Alliance Questionnaire was used to measure alliance. All therapy sessions were tape-recorded. An independent researcher rated the patient's and therapist's contribution to the alliance at the third session (early alliance). Early patient and therapist alliances were not related to sociodemographic data or initial measures of alcohol severity, psychiatric symptoms, or personality structure in either therapy. Early therapist alliance was better in MBT in comparison with PT. For MBT patients, a significant positive correlation was found between early patient alliance and mood dimensions at 6 months. There were no significant positive correlations between early alliance and drinking outcome during the course of treatment and in the third year after start of treatment. Mood and drinking outcome also showed low correlations. In conclusion, an initial positive alliance seems insufficient to reduce alcohol misuse. The relationship between early alliance and improvement in alcohol misuse needs to be further investigated. INTRODUCTION In recent decades the role of alliance in psychotherapy has received special attention. The construct of therapeutic alliance has been described in several ways, but most constructs include some of the following components: (1) the patient's affective relationship to the therapist; (2) the patient's capacity to work purposefully in therapy; (3) the therapist's empathic understanding and involvement; (4) patient-therapist agree- Author to whom correspondence should be addressed. ment about the goals and tasks of therapy (Gaston, 1990). The positive relation between a good alliance and a successful therapy outcome is reasonably well documented across a variety of different therapies (Horvath and Luborsky, 1993), although the average overall effect size (ES = 0.26) is not very large (Horvath and Symonds, 1991). The helping alliance in treatment of substance use disorders has been addressed by Luborsky and coworkers in a methadone maintenance sample (Luborsky et al., 1985). They found that early alliance (third session) was positively related to outcome measures of ASI (Addiction Severity Index), i.e. drug use, employ Medical Council on Alcoholism

2 242 A. OJEHAGEN et al. ment status and psychological functioning (mean r s = 0.65) after 7 months. The duration of the intervention in that study was weeks. However, the relationship between alliance and drinking outcome in alcohol use disorders in longterm treatment, i.e. 12 months or more, has not been presented as far as we know. In comparison with general psychotherapy, the objectives in alcoholism treatment are twofold: to reduce the alcohol abuse and to achieve an improvement in other areas, i.e. mental health and social adjustment. Although the concept of therapeutic alliance was originally grounded in psychoanalytic theory, there is little reason to believe that the quality of alliance, as it is currently defined, is more important in analytically oriented interventions than in other modalities of therapy. On the contrary, a better alliance has recently been found in cognitive and behavioural therapy, in comparison with psychodynamic-interpersonal therapy (Raue et al., 1993). In the present study, alcoholics were randomized to two outpatient treatment programmes; multimodal behavioural therapy (MBT) (Lazarus, 1981), and psychiatric treatment based on a psychodynamic approach (PT) (Luborsky, 1984). Both treatment programmes have been described in a previous paper (Ojehagen et al., 1992). The main results of the randomized part of the study and the influence of psychopathology on clinical outcome have been reported in previous articles. There were no significant outcome differences between the two treatment models during the course of treatment or in the third year after start of treatment (Ojehagen et al., 1992). A favourable drinking outcome was predicted by a better initial psychiatric status and fewer psychiatric symptoms at intake (Ojehagen et al, 1991). In this part of the study three issues in patients who completed therapy will be addressed: (1) is there a difference in early alliance between MBT and PT?; (2) is the level of early alliance related to mood outcome during treatment?; (3) is the level of early alliance related to drinking outcome during treatment and in the third year after start of treatment? Two other issues are also addressed: (1) are initial patient characteristics related to the level of early alliance?; (2) are mood dimensions and drinking outcome related during the first year? PATIENTS AND METHODS Patients The series consisted of all subjects who attended an outpatient alcoholism treatment study at the Department of Psychiatry, University Hospital, Lund from July 1983 to June The Medical Ethics Committee of the University of Lund approved of the study, and the subjects had given their consent to participate in writing. Inclusion in the programme was preceded by two sessions together with a senior psychiatrist with at least one week between the sessions. The patient could be either an in- or outpatient at the first session, but had to be an outpatient at the second session, when he/she made the decision to enter the programme. The patient had to be sober at both sessions. No other exclusion criteria were used. Seventy-two patients, 60 men and 12 women, accepted participation in the study after two information sessions with a senior psychiatrist, whereas 57 did not join the programme (Ojehagen et al., 1992). The acceptors had a mean age of 37 ± 9 years, 74% were married or cohabiting, 68% were regularly employed and 57% had 10 years of education or more. Half of the sample had previously been treated for alcoholism (Ojehagen et al., 1992). The subgroup of patients included in this investigation of early alliance (n = 35) did not differ significantly from the others in measurements used in this study. A factorial design was used in which subjects were randomized to psychiatric treatment based on a psychodynamic approach (n = 36) or multimodal behaviour therapy (n = 36), and to 1 or 2 years in treatment. The treatment programmes were not based on manuals. Twenty-five patients (35%) did not complete therapy; three of them had died, four moved to another area and 18 (seven in PT) left therapy prematurely. Forty-seven patients completed treatment, and 43 of them were followed up 36 months after start of treatment (Ojehagen et al., 1992). Of these 43, only patients with complete sets of mood assessments (before treatment and three, six and 12 months after start of treatment) were included in the analysis (MBT, n = 17; PT, n = 18). Nineteen of these patients had 1 year of treatment, and 16 patients had 2 years of treatment. Three patients in MBT and one in PT were women. The number (mean ± SD) of treatment sessions

3 HELPING ALLIANCE AND OUTCOME IN ALCOHOLISM TREATMENT 243 (MBT: 24.7 ± 1.7; PT: 24.6 ± 1.7) did not differ between the therapies, and was not related to drinking outcome. Nor did number of treatment sessions differ between 1 and 2 years of treatment (Qjehagen et al., 1992). Therapists In MBT, there was one therapist, while there were three therapists in PT. A change of therapist occurred for one-third of the patients during the first half of the study period, because of pregnancy. The therapists did not differ significantly concerning education and training. They were all trained in the specific type of treatment given (Qjehagen et al., 1992). Methods Initial ratings were performed by the senior psychiatrist at the second information session. The self-rating instruments were all presented by independent research assistants before the randomization procedure. The intake- and outcome measures except for the measure of helping alliance and mood dimensions below have been described in detail in a previous paper (Qjehagen et al., 1992). Helping alliance. A Swedish version of the Helping Alliance Questionnaire (Morgan et al., 1982) translated by B.-A Armelius and M. Phelan was used to measure alliance. The concept 'helping alliance' is operationalized into 20 items. It contains two parts, one measuring the therapist's contribution to the alliance, and the other the patient's contribution. Each item has a negative and a positive pole, 1-10, where 10 is the most positive value. All therapy sessions were taperecorded. From these tape-recordings, an independent rater with experience as a therapist rated the patient's and therapist's contribution to the alliance at the third session (early alliance) and at the third to last session of the therapy (late alliance) for patients who completed treatment (n = 43). In this paper, only the assessments of the early alliance are presented. The rater did not know when the session took place during therapy. This rater and a second rater made independent assessments of 10 sessions with an interrater reliability of r s = 0.58 (P < 0.05). Intake measures The Comprehensive Psychopathological Rating Scale (CPRS). The CPRS, an interview-based symptom scale, was used to evaluate psychopathological symptoms (Asberg et al., 1978). Drug Taking Evaluation Scale (DTES). By interview, the severity of abuse, social functioning, social belonging, and psychic status were assessed in the four sub-scales of DTES (Holsten and Waal, 1980). These subscales consist of nine operationally defined steps, where scores 1-3 denote normality and higher scores signify increasing severity of symptoms or dysfunction. DSM-III. The patients were asked about the presence or absence of all items included in the definition of alcohol dependence (American Psychiatric Association, 1980). Problem Drinking Scale (PDS). The PDS is an interview-based measure of the severity of alcoholism consisting of 16 items (Vaillant, 1983). The Swedish version of the Alcohol Use Inventory (AVI). This is a self-rating questionnaire, which is based on the Alcohol Use Inventory (Wanberg et al., 1977), and standardized on a sample of more than 600 Swedish alcoholics (Berglund et al, 1988). Outcome measures Interviews and self-ratings were administered by independent research assistants 3, 6, 12, 24 and 36 months after the start of treatment. In this study, self-ratings of mood at 3, 6 and 12 months and measurement of drinking outcome at 6,12 and 36 months will be used. Mood dimensions. A Swedish Mood Adjective Check List (MACL), a self-rating questionnaire (Sjtiberg et al., 1979; BokstrSm et al., 1991), comprising 71 mood-related adjectives grouped into six bipolar dimensions, was used. These dimensions are: pleasantness/unpleasantness (happy, sad, etc.); activation/deactivation (active, drowsy, etc.); extroversion/introversion (talkative, silent, etc.); calmness/tension (calm, nervous, etc.); positive/negative social orientation (cooperative, unreasonable, etc.); and control/lack of control (self-confident, insecure, etc.), with the dimensions before the slanting strokes representing positive tones. The dimensions pleasantness/ unpleasantness, activation/deactivation and calmness/tension are regarded as basic mood dimen-

4 244 A. OJEHAGEN et al. sions, whereas the other three dimensions are considered to be more closely associated with social relations and situations. The presentation of the adjectives is randomized. The possible range for each dimension is 1 4. A high value denotes a positive tone. The MACL questionnaire was administered by the research assistant before treatment and 3, 6 and 12 months after the start of treatment. Drinking outcome. The number of abuse days was registered. Abuse days were defined as days with a consumption of >4 drinks (1 drink 3.8 cl of 40% liquor) during continuous drinking, or >6 drinks on occasional drinking days (Sobell and Sobell, 1978). During the periods of follow-up, the number of abuse days was used as a measure of drinking outcome. The same cut-off points for favourable outcome as in our previous studies were used (Nordstrom and Berglund, 1987; Ojehagen et al., 1988). A maximum of 7 abuse days at 6 month follow-up periods and 14 abuse days during 1 year follow-up periods was considered a favourable outcome. In this sample, corroboration of positive drinking outcome during the third year was made by relatives and/or through liver enzyme measurements (Ojehagen et al., 1992). Statistics The Mann Whitney {/-test was used as a nonparametric test for differences between subgroups, and when appropriate Student's Mest and ANOVA were used in the analyses of differences between means. Spearman rank correlations were also performed. A stepwise logistic regression analysis was used in analyses of the relationship between alliance and drinking outcome. The software used for statistical analyses was SPSS for Windows, Version 6.0 (Norusis, 1993). RESULTS Comparison between the two therapies regarding early alliance Multimodal therapy (MBT, n = 17) had significantly better early therapist alliance in comparison with psychiatric treatment (PT, n = 18) according to Mann-Whitney 17-test (MBT: mean ± SEM = 67.2 ± 1.0; PT: 61.3 ± 1.0; P < 0.01). There were no significant differences between the two therapies regarding early patient alliance (MBT: 62.5 ± 1.4; PT: 60.1 ± 1.2). An ANOVA showed no differences in early therapist or patient alliance with regard to length of therapy, nor were there any significant interactions between mode of therapy and length of therapy. Within the respective therapies, there were significant correlations (P < 0.001) between early patient and therapist alliance (r$ = 0.81 in MBT and r s = 0.81 in PT). Early alliance (patient or therapist) was in neither of the two therapies related to sociodemographic data (sex, marital and occupational status), nor initial measures of severity of alcohol misuse and psychopathology (DTES, PDS, AVI or CPRS) (n = 35). Early alliance in relation to mood dimensions during treatment There were no significant differences with regard to the six mood dimensions between patients in the two therapies before, or at 3, 6 and 12 months after the start of treatment. In Fig. 1, correlations are presented between early patient alliance and mood dimensions before treatment, and after 3, 6 and 12 months in MBT and PT; Fig. la concerns three basic mood dimensions (pleasantness, activation and calmness), and Fig. lb concerns three mood dimensions associated with social relations and situations (extroversion, positive social orientation and control). In MBT, there were significant positive correlations between early patient alliance and three mood dimensions after 6 months (in Fig. la: pleasantness +0.53; in Fig. lb: extroversion +0.48, and control +0.56; P < 0.05). The correlations between early therapist alliance and mood dimensions were similar to the correlations between mood and early patient alliances, but did not reach a significant level [in MBT the therapist correlations to basic mood dimensions were: pleasantness +0.27, activation +0.09, calmness +0.16; and to mood dimensions associated with social relations and situations: extroversion +0.31, positive social orientation +0.14, control (not significant)].

5 HELPING ALLIANCE AND OUTCOME IN ALCOHOLISM TREATMENT 245 0,6 i A - o- (a) (b) r i -O.20- A" 6 Months O Pleasantness - PT D Activation - PT A Calmness - PT Pleasantness - MBT Activation - MBT A Calmness - MBT O Extroversion - PT Extroversion - MBT 6 Months Positive social orientation - PT Positive social orientation - MBT A Control - PT A Control-MBT Fig. 1. (a) Correlations between early patient alliances and basic mood dimensions. Dimensions (pleasantness, activation, calmness) were assessed at the start of treatment and after 3, 6 and 12 months in multimodal behavioural therapy (MBT) and psychiatric treatment (PT). There was a significant positive correlation after 6 months in MBT between alliance and pleasantness (P < 0.05). (b) Correlations between early patient alliances and mood dimensions associated with social relations and situations. Dimensions (extroversion, positive social orientation and control) were assessed at the start of treatment and after 3, 6 and 12 months in multimodal behavioural therapy (MBT) and psychiatric treatment (PT). There were significant positive correlations after 6 months in MBT between alliance and extroversion and control (P < 0.05). 12 Early alliance in relation to drinking outcome In Table 1, correlations between early patient and therapist alliances and drinking outcome in the first and second half-year, and in the third year after start of treatment are presented separately for MBT and PT. There were no significant positive correlations between early alliances and drinking outcome. Some correlations were, however, strongly negative. A stepwise logistic regression analysis with alliance and psychiatric status as

6 246 A. OJEHAGEN et al. Drinking outcome Table 1. Spearman rank correlations between eariy patient and therapist alliances in drinking outcome Patient MBT (n = 17) Eariy alliance Therapist Patient PT (n = 18) Early alliance Therapist First half-year Second half-year Third year » Correlations were applied for the above measures in the first and second half-year, and in the third year after start of treatment in multimodal behavioural therapy (MBT) and psychiatric treatment (PT). *P < independent variables was also performed, which showed no significant associations with drinking outcome. Correlations between mood dimensions and drinking outcome during the first year Correlations between drinking outcome in the first half-year and mood dimensions at 6 months, and between drinking outcome in the second halfyear and mood dimensions at 12 months, in MBT and PT respectively, in all 24 correlations, were performed. In PT, there was a positive correlation between drinking outcome in the first half-year and activation at 6 months (+0.48, P < 0.05), whereas other correlations were positive but not significant. The correlations in MBT were generally low. DISCUSSION When interpreting the results of the present study, some shortcomings should be kept in mind. The sample groups were too small for drawing safe conclusions. The attrition from treatment is analysed in a previous article (Ojehagen et al., 1992). In the present paper, we focused on alliance in relation to outcome among treatment completers. An intention-to-treat analysis of all subjects randomized would give additional aspects on the importance of alliance. This will be analysed in a separate paper. In randomized treatment studies, some patients probably will be mismatched, and therefore it would be especially interesting to investigate alliance in relation to attrition. Furthermore, the therapist factor is not standardized; there were no treatment manuals and the number of therapists differed between the two treatment types. A greater number of therapists should perhaps have been used to prevent confounding of therapist effects with the comparison of MBT with PT. In particular, the single therapist in MBT confounds therapist effects with the treatment effect. However, our findings of a better alliance in MBT are in accordance with another recent study, thus pointing to a positive relationship between alliance and MBT beyond the therapist factor (Raue et al., 1993). The interrater reliability of the helping alliance was only performed in a few patients, and correlation coefficients were rather weak. Finally, this study recruited socially stable alcoholics, with 74% being married or cohabiting and 68% regularly employed, and it is reasonable to assume that our findings are representative only of socially stable alcoholics. In this randomized study, comparisons of helping alliance between two long-term alcoholism treatment models were performed that have not been presented previously. Both mood and drinking outcome were assessed several times, which further strengthens the design. Our main results were that early therapist alliance was better in MBT in comparison with PT. A significant positive correlation was shown between early patient alliance and mood dimensions at 6 months in MBT. There were no significant positive correlations between early alliance and drinking outcome. A better alliance in cognitive and behavioural therapy in comparison with psychodynamic-interpersonal therapy has recently been reported (Raue et al., 1993), and might be explained by

7 HELPING ALLIANCE AND OUTCOME IN ALCOHOLISM TREATMENT 247 differences in therapist style and treatment structure. These differences, also present in our study, were however not related to drinking outcome in the whole sample (Qjehagen et al., 1992). One difference between MBT and PT concerns the timing of treatment, i.e. in MBT most treatment sessions took place during the first half of the treatment period with booster sessions thereafter, whereas sessions were individually distributed in PT. There were to be 30 sessions in both treatment programmes, and the final number of sessions did not differ between the therapies (Qjehagen et al., 1992). The metaanalysis including 20 alliance studies performed by Horvath and Symonds (1991) found no correlation between alliance and treatment length (6 52 sessions). The early alliance differences concerned therapist alliance only, which may mirror the more active approach of MBT, and which might have influenced the level rating of the MBT therapist alliance. Behavioural cognitive therapy techniques have been shown to be more successful than dynamically-orientated therapies in subjects with a more severe psychopathology/sociopathy and vice versa (Cooney et al., 1991; Litt et al., 1992). Similar results have been found in this study, i.e. in PT patients a favourable drinking outcome in the third year was more strongly related to a better personality structure (i.e. DTES scores <3, which denotes normality) than in MBT patients (Qjehagen et al., 1992). In comparison with PT, which was based on a psychodynamic approach, the MBT programme focused on finding coping strategies, which may apply more to patients with a less good personality structure. These patients are often in need of strong support and structure, which might partly explain the positive correlation between early patient alliance and mood at 6 months in MBT patients. Perhaps also techniques in building a good alliance are more apparent in behavioural-orientated therapies. Gaston (1990) reported that, within treatment conditions, the alliance uniquely contributed to outcome, with an increasing variance accounted for as the therapy progressed, especially in behaviour and cognitive therapy as compared to brief dynamic therapy. In MBT treatment, alliances and mood outcome during treatment had no relation to favourable drinking outcome. In the present study, no positive relations of early alliance with favourable drinking outcome during the first and second half-year, or in the third year in any of the treatment programmes were found. On the contrary the correlation to drinking outcome in the third year was significantly negative in MBT. Thus, the predictive value of early alliance to outcome reported from other studies could not be corroborated. These studies mostly concerned psychotherapy and were mostly not evaluated in a long-term perspective. In modem alcoholism treatment studies, a follow-up period of 2 years after start of treatment is common, since stability of improvement is of importance. The negative correlation ought to be replicated in further studies. Maybe alliance early in therapy is not predictive of drinking outcome in comparison with other determinants of outcome. Factors influencing alliance, i.e. patient characteristics and therapy technique remain to be further investigated (Henry et al., 1994). Luborsky et al. (1985) analysed the alliance in methadone-maintained drug dependent patients and found significant positive correlations between alliance and a number of outcome measures at 7 months, including drug use, employment status, and psychological functioning according to ASI (Addiction Severity Index). Methadone-maintenance therapy is a highly structured treatment with daily administration of methadone, and is a considerably less intense form of alcoholism treatment (1-2 sessions/ month); this could be one explanation for the divergent results. Another difference between our study and that by Luborsky et al. (1985) concerns the method of rating the helping alliance. In Luborsky's study the therapists and patients rated the alliance after the third session, whereas in our study the ratings were made by an independent researcher based on tape-recordings. The drinking outcome measure in our study also differs, i.e. our measure of drinking outcome is categorical, favourable versus unfavourable, unlike the ASI measure used by Luborsky. Horvath and Symonds (1991) found that clients' and observers' ratings of alliance appeared to be more correlated with all types of outcomes reported than therapists' ratings. In general, the alliance measures appear to better predict outcomes tailored to the client (i.e. target complaints) than broad-range symptomatic change (Horvath

8 248 A. OJEHAGEN et al. and Symonds, 1991). Furthermore, the process of improvement concerning alcohol misuse and improvement in other areas is not clearly known (Ojehagen et al., 1986). For example, in this study, mood during treatment was only to some degree related to drinking outcome during treatment, but not to outcome in the third year after start of treatment. Apparently helping alliance is positively correlated to mood after 6 months, especially in certain types of therapy. This is of importance in understanding different treatment processes of change. Therefore the relationship of alliance to patients' mood during alcohol treatment, and the relation of mood to drinking outcome (Bokstrom et al., 1991) should be addressed in further studies. In conclusion, the relationship between early alliance and improvement in alcohol misuse needs to be further investigated. An initial good alliance seems insufficient to reduce alcohol misuse. The absence of a positive relationship between alliance and drinking outcome stresses the need for further investigations of the process of change in alcoholism treatment. Acknowledgements This study was supported by The Swedish Council for Planning and Coordination of Research, The Swedish Council for Social Research, The Swedish Medical Research Council and funds from Lund University. REFERENCES American Psychiatric Association (1980) Diagnostic and Statistical Manual of Mental Disorders, 3rd edn. American Psychiatric Association, Washington,. DC. Asberg, M., Montgomery, S. A., Penis, C, Schalling, D. and Sedwall, G. (1978) A Comprehensive Psychopathological Rating Scale. Ada Psychiatrica Scandinavica, Supplementum. Berglund, M., Bergman, H. and Swenelius, T. (1988) The Swedish Alcohol Use Inventory (AVI), a selfreport inventory for differentiated diagnosis in alcoholism. Alcohol and Alcoholism 23, Bokstrom, K., Balldin, J. and LSngstrdm, G. (1991) Individual mood profiles in alcohol withdrawal. Alcoholism: Clinical and Experimental Research 15, Cooney, N. L., Kadden, R. M., Litt, M. D. and Getter, H. (1991) Matching alcoholics to coping skills or interactional therapies: Two-year follow-up results. Journal of Consulting Clinical Psychology 59, Gaston, L. (1990) The concept of the alliance and its role in psychotherapy: theoretical and empirical considerations. Psychotherapy 27, Henry, W. P., Strupp, H. H., Schacht, T. E. and Gaston, L. (1994) Psychodynamic approaches. In Handbook of Psychotherapy and Behavior Change, 4th edn, Bergin, A. E. and Garfield, S. L. eds, pp Wiley, New York. Holsten, F. and Waal, H. (1980) The Drug Taking Evaluation Scale. Acta Psychiatrica Scandinavica 61, Horvath, A. O. and Luborsky, L. (1993) The role of therapeutic alliance in psychotherapy. Journal of Consulting Psychology 61, Horvath, A. O. and Symonds, B. D. (1991) Relation between working alliance and outcome in psychotherapy: A meta-analysis. Journal of Consulting Psychology 38, Lazarus, A. A. (1981) The Practice of Multi-modal Therapy. McGraw-Hill, New York. Litt, M. D., Babor, T. F., DelBoca, F. K., Kadden, R. M. and Cooney, N. L. (1992) Types of alcoholics, II. Application of an empirically derived typology of treatment matching. Archives of General Psychiatry 49, Luborsky, L. (1984) Principles of Psychoanalytic Psychotherapy. A Manual for Supportive Expressive Treatment. Basic Books, New York. Luborsky, L., McLellan, T., Woody, G. E., O'Brien, C. P. and Auerbach, A. (1985) Therapist success and its determinants. Archives of General Psychiatry 42, Morgan, R. W., Luborsky, L., Crits-Christoph, P. et al. (1982) Predicting the outcomes of psychotherapy by the Penn Helping Alliance rating method. Archives of General Psychiatry 39, Nordstrom, G. and Berglund, M. (1987) Type 1 and type 2 alcohoh'cs (Cloninger & Bohman) have different patterns of successful long-term adjustment. British Journal of Addiction 82, Norusis, M. J. (1993) SPSS for Windows, 6.0. SPSS Inc., Chicago. Ojehagen, A., Skjaerris, A. and Berglund, M. (1988) Prediction of posttreatment drinking outcome in a 2- year out-patient alcoholic treatment program. A follow-up study. Alcoholism: Clinical and Experimental Research 12, Ojehagen, A., Berglund, M., Appel, C.-P., Nilsson, B. and Skjaerris, A. (1991) Psychiatric symptoms in alcoholics attending outpatient treatment. Alcoholism: Clinical and Experimental Research 15, Ojehagen, A., Berglund, M., Appel, C.-P., Andersson, K., Nilsson, B., Skjaerris, A. and Toftenow-Wedlin, A.-M. (1992) A randomized study of long-term outpatient treatment in alcoholics. Psychiatric treatment versus multimodal behavioral therapy, and 1 versus 2 years of treatment. Alcohol and Alcoholism 27, Raue, P. J., Castonguay, L. G. and Goldfried, M. R. (1993) The Working Alliance: a comparison of two therapies. Psychotherapy Research 3, SjSberg, L., Svensson, E. and Persson, L.-O. (1979) The measurement of mood. Scandinavian Journal of

9 HELPING ALLIANCE AND OUTCOME IN ALCOHOLISM TREATMENT 249 Psychology 20, Harvard University Press, Cambridge, MA and Sobell, M. B. and Sobell, L. C. (1978) Behavioral London. Treatment of Alcohol Problems. Plenum Press, New Wanberg, K. W., Horn, J. L. and Foster, M. F. (1977) A York. differential assessment model for alcoholism. The Vaillant, G. E. (1983) The Natural History of Alcohol- scales of the alcohol use inventory. Journal of ism. Causes, Patterns and Paths to Recovery. Studies on Alcohol 38,