Introduction
The therapeutic relationship has long been held to be an important factor contributing to both process and outcome in psychotherapy. Previous meta-analytic reviews have demonstrated that the alliance is consistently related to treatment outcome in individual psychotherapy, across treatment modalities and patient populations (Horvath and Symonds, Reference Horvath and Symonds1991; Martin, Garske and Davis, Reference Martin, Garske and Davis2000). Although different theoretical conceptualizations as well as empirical instruments of the alliance have been developed, they all seem to capture a similar global quality of the alliance and predict outcome equally well (Martin et al., Reference Martin, Garske and Davis2000).
The construct of the alliance is usually described in dyadic terms, tapping into the quality of the relationship between a therapist and a patient. Since group therapy unavoidably involves a larger set of people, as well as different treatment structures compared to individual therapy, constructs such as climate (a sense of constructive interpersonal investigation), cohesion (a sense of belonging) and empathy (a sense of being understood) have all been suggested to tap into important relationship qualities in this treatment modality (Johnson, Burlingame, Olsen, Davies and Gleave, Reference Johnson, Burlingame, Olsen, Davies and Gleave2005). The construct of the alliance has also been adapted to be used in group treatment settings. Theoretically, they have all been suggested to be important indicators of the quality of the relationships between group members, to be a “substrate” for work in the group and to foster the development of other positive therapeutic processes (MacKenzie, Reference MacKenzie, Safran and Muran2000; Tschuschke and Dies, Reference Tschuschke and Dies1994). For example, a high sense of cohesion or engagement may represent patients' greater involvement in work-related tasks, thus increasing benefits from the group (Ogrodniczuk and Piper, Reference Ogrodniczuk and Piper2003). Moreover, others (e.g. Yalom and Leszcz, Reference Yalom and Leszcz2005) have also suggested that it is precisely the therapeutic relationships and interpersonal exchanges between group members that are the curative ingredients in effective group therapy, causing a positive outcome.
The therapeutic relationship in group therapy can thus be conceptualized as necessary prerequisites for method-specific therapeutic techniques or tasks to be implemented (that is, not curative in and of itself), or as having a direct causal effect upon outcome (curative in and of itself). These two theoretical positions largely parallel the distinction between process-groups, usually placing a high value on the interpersonal and interactional climate of the group (the group processes being the vehicle of change), and more highly-structured approaches such as cognitive-behavioural group therapy. In the latter treatment model, more attention is usually given to pre-planned, highly structured in-session activities that focus on specific change strategies, for example cognitive restructuring. Relatively little attention is given to the unique properties of the group format or atmosphere, such as cohesion or engagement, as important vehicles for treatment process and outcome (Burlingame, MacKenzie and Strauss, Reference Burlingame, MacKenzie, Strauss and Lambert2004).
In contrast to the construct of the alliance in individual therapy, no overall consensus has yet been reached as to the most appropriate definition of the helpful relationship qualities in group therapy (Johnson et al., Reference Johnson, Burlingame, Olsen, Davies and Gleave2005; Joyce, Piper and Ogrodniczuk, Reference Joyce, Piper and Ogrodniczuk2007). The use of divergent constructs (cohesion, climate, empathy, alliance) is one source of this conundrum. Varying definitions of the same construct have been given, or constructs have been used interchangeably (e.g. MacKenzie, Reference MacKenzie, Safran and Muran2000). The proliferation of different empirical instruments has also been a problem, with many of them used in only a few studies (Burlingame et al., Reference Burlingame, MacKenzie, Strauss and Lambert2004).
Overall, the concept of cohesion seems to have been the most preferred term, but reviewers (Bednar and Kaul, Reference Bednar, Kaul, Bergin and Garfield1994; Dion, Reference Dion2000) have been led to conclude that there is little cohesion in the cohesion literature. Some have suggested that the cohesion-construct is too vague and amorphous to be useful as a unitary construct (Hornsey, Dwyer and Oei, Reference Hornsey, Dwyer and Oei2007), but others caution against the development of new instruments since this may not resolve the underlying difficulties (Johnson et al., Reference Johnson, Burlingame, Olsen, Davies and Gleave2005). Interestingly, a recent multilevel structural equation model analysis suggests that the constructs of group climate, cohesion, alliance and empathy may all reflect one or more higher order constructs, which may be more related than prior evidence or clinical theory suggests (Johnson et al., Reference Johnson, Burlingame, Olsen, Davies and Gleave2005).
Empirical research on the association between the various therapeutic relationship qualities and outcome in group therapy has produced more mixed results compared to research on the association between the alliance and outcome in individual therapy. The cumulative empirical evidence for a positive association between group cohesion and treatment outcome is overall not particularly strong, which is puzzling in light of the widely held belief that group cohesion is the quintessential process variable in group therapy (Taft, Murphy, King, Musser and DeDeyn, Reference Taft, Murphy, King, Musser and DeDeyn2003). This probably reflects some of the difficulties in reaching an overall agreement on a definition, the use of different instruments as well as differences in measurement approaches, which makes comparisons between studies difficult. Some studies on analytic- and dynamic treatment models have suggested a positive relationship between group cohesion and treatment outcome with varying groups and patient problems, such as anxiety and depression (Budman et al., Reference Budman, Soldz, Demby, Feldstein, Springer and Davis1989), complicated grief (Joyce et al., Reference Joyce, Piper and Ogrodniczuk2007) and neurotic and personality disorders (e.g. Tschuschke and Dies, Reference Tschuschke and Dies1994; MacKenzie and Tschuschke, Reference MacKenzie and Tschuschke1993). Other studies have reported negative findings (e.g. Gillaspy, Wright, Campbell, Stokes and Adinoff, Reference Gillaspy, Wright, Campbell, Stokes and Adinoff2002; Lorentzen, Sexton and Høglend, Reference Lorentzen, Sexton and Høglend2004; Marziali, Munroe-Blum and McCleary, Reference Marziali, Munroe-Blum and McCleary1997).
Cognitive-behavioural therapy is often characterized as placing less emphasis on the therapeutic relationship and more focus on specific techniques and tasks as the central ingredients of effective therapy, compared to dynamic- or experientially oriented therapies. Interestingly though, studies on this usually highly structured treatment model have also produced some positive results. For example, groups with higher levels of cohesion were found to have greater improvement up to 6 months after treatment of agoraphobia compared to groups with lower cohesion (Hand, Lamontagne and Marks, Reference Hand, Lamontagne and Marks1974). Moreover, higher levels of cohesion have been found to be predictive of lower physical and psychological abuse at follow-up in abusive men (Taft et al., Reference Taft, Murphy, King, Musser and DeDeyn2003), to be related to decreased post-treatment systolic and diastolic blood pressure, as well as improved quality of life in patients with cardiac disease (van Andel, Erdman, Karsdorp, Appels and Trijsburg, Reference van Andel, Erdman, Karsdorp, Appels and Trijsburg2003) and early group cohesion has been found to predict better outcome in binge eating disorder (Castonguay, Pincus, Agras and Hines, Reference Castonguay, Pincus, Agras and Hines1998). Increase in group cohesion has also been related to improvements in social anxiety in a naturalistic study (Taube-Schiff, Suvak, Antony, Bieling and McCabe, Reference Taube-Schiff, Suvak, Antony, Bieling and McCabe2007), although another study failed to report any effect with a similar diagnostic sample (Woody and Adessky, Reference Woody and Adessky2002). Also, a transdiagnostic CBT-treatment model recently reported increase in group cohesion to be related to a better treatment outcome (Norton, Hayes and Springer, Reference Norton, Hayes and Springer2008).
Taken together, the empirical evidence mostly indicates a positive association between relationship qualities such as cohesion and outcome in CBT-based group treatments, which highlights the importance of such factors even in highly structured treatment approaches. This is also in accordance with findings from comparative trials in individual psychotherapy, where the quality of the alliance has been found to be both equally strong in CBT-oriented therapies, and to predict treatment outcome equally well, compared to dynamic- and experiential therapies (Marmar, Gaston, Gallagher and Thompson, Reference Marmar, Gaston, Gallagher and Thompson1989; Salvio, Beutler, Wood and Engle, Reference Salvio, Beutler, Wood and Engle1992; Spinhoven, Gisen-Bloo, Dyck, Kooiman and Arntz, Reference Spinhoven, Giesen-Bloo, van Dyck, Kooiman and Arntz2007). This indicates that an emphasis on pre-planned, highly structured in-session activities, and a directive therapist stance, does not necessarily undermine the importance of relationship factors in therapy. On the contrary, studies even suggest that more structure, especially in the early phase of group treatment, may improve cohesion (Stockton, Rhode and Haughey, Reference Stockton, Rhode and Haughey1992).
One way to solve some of the problems related to research on the association between the helpful relationship qualities and outcome in group therapy would be to use a generic, trans-theoretical measure that is applicable to most treatment conditions. This would obviously facilitate comparisons between studies, opening up for broader generalizations in this field of research (Burlingame et al., Reference Burlingame, MacKenzie, Strauss and Lambert2004). The Group Climate Questionnaire-Short Form (GCQ-S; Mackenzie, Reference MacKenzie, Dies and MacKenzie1983) is one such instrument, which in addition to a measure of cohesion or (a) engagement includes two other subscales called (b) avoidance (to what extent group members avoid responsibility for the change process in the group) and (c) conflict (taps into a sense of tension and conflict in the group). Higher ratings of engagement and lower ratings of both avoidance and conflict is usually thought to be associated with a positive treatment outcome.
Although extensively used across a variety of treatment settings and patient populations, the instrument has only been used in one randomized and controlled trial (Ogrodnizcuk and Piper, Reference Ogrodniczuk and Piper2003). One of the main findings from this study was that higher ratings of engagement after session four, and averaged over the course of therapy, were directly associated with in-treatment improvement in two forms of short-term dynamic psychotherapy with complicated grief patients. Increase in engagement throughout treatment did not predict outcome. Generalizations from this one study to other treatment modalities as well as patient populations are somewhat limited. Also, it is uncertain whether dimensions of group climate may predict long-term follow-up.
The primary objective of the present study was therefore to examine the predictive validity of the GCQ-S for long-term follow-up in a different and more structured group treatment modality (CBT) using a more heterogeneous patient sample and other outcome measures. We hypothesized that (1) higher ratings of engagement, (2) lower ratings of avoidance, and (3) lower ratings of conflict would predict an overall better one-year follow-up.
Method
Procedure
Data used in the present study were taken from a randomized controlled trial investigating the effectiveness of cognitive-behavioural group therapy with comorbid psychiatric disorders. A detailed description of the original design and methodology is presented by Hagen, Nordahl, Kristiansen and Morken (Reference Hagen, Nordahl, Kristiansen and Morken2005). The treatment was found to be effective compared to waiting list controls, with effects upheld at 6 and 12 months follow-up (Hagen et al., Reference Hagen, Nordahl, Kristiansen and Morken2005). Data from both treatment arms were combined in the present study.
Patients
All patients were diagnosed according to the DSM-IV system using the Structured Clinical Interview for the DSM-IV axis I (First, Spitzer, Gibbon and Williams, Reference First, Spitzer, Gibbon and Williams1995) and axis II (First, Spitzer, Gibbon, Williams and Benjamin, Reference First, Spitzer, Gibbon, Williams and Benjamin1994). Patients with active substance abuse, psychosis, suicidal behaviour and cluster A or B personality disorders were excluded. Patients were randomized to either a cognitive behavioural group therapy program or a waiting list condition, with six groups comprising 5–8 patients. Thirty-two patients completed 8 weeks of therapy. In the present study, two patients were excluded due to missing group climate data, and three more patients were excluded due to missing outcome data leaving a sample of n = 27 for further analysis. Of these, one patient did not complete the conflict scale on the process measure. A detailed description of demographic and diagnostic variables of the sample is presented in Table 1.
Table 1. Demographic and diagnostic variables for the sample
Note: Each patient could receive more than one diagnosis.
Treatment
The treatment manual (Cognitive-Behavioural Group Therapy for comorbid psychiatric disorders) was based on a modified version of the manual developed by Free (Reference Free1999). Work with the ABC-model, role-play, in vivo exposure and developing coping strategies to prevent relapse were central ingredients. The program thus provided a mix consisting of psycho-education related to depression and anxiety, group exercises and homework tasks. Overall, the treatment manual highlights the importance of structured therapeutic tasks as the central curative components of treatment. Qualities related to the therapeutic relationships are seen as important, but not hypothesized to be a central curative pathway. Treatment duration was 8 weeks, with two weekly held sessions of about 90 minutes duration.
Therapists
Two experienced female cognitive therapists led the treatment groups, and they had weekly supervision by one of the authors (HMN). The competence of the therapists was evaluated according to the Cognitive Therapy Scale (Young and Beck, Reference Young and Beck1980), using video-recordings of the third and tenth treatment-sessions. Competence is rated on a 7-point Likert scale, ranging from zero (low competence) to six (high competence). The two therapists in the original study received an overall mean score of 4.18 (SD = .32) and 4.05 (SD = .29), respectively, which is considered acceptable levels of therapist competence in cognitive therapy (Vallis, Shaw and Dobson, Reference Vallis, Shaw and Dobson1986).
Outcome measures
The outcome measures included in the original trial were also used in the present study (except for the sociotropy-avoidance scale; Beck, Epstein, Harrison and Emery, Reference Beck, Epstein, Harrison and Emery1983), using pre, post and 12-month follow-up scores. The outcome measures were as follows:
The Inventory of Interpersonal Problems (IIP-64; Horowitz, Rosenberg, Baer, Ureño, and Villaseñor, Reference Horowitz, Rosenberg, Baer, Ureño and Villaseñor1988), based upon the work of Alden, Wiggins and Pincus (Reference Alden, Wiggins and Pincus1990). This self-report instrument consists of 64-items in which 8 subscales are conceptually organized in a circumplex manner along two, main dimensions (dominance; love). Patients are asked to rate interpersonal behaviour that is “hard for you to do” or “you do too much” on a 5-point Likert scale ranging from 0 = Not at all to 4 = Extremely. The global index of interpersonal problems was used. Reliability and validity of the instrument is reported as acceptable (Horowitz et al., Reference Horowitz, Rosenberg, Baer, Ureño and Villaseñor1988).
The Symptom Checklist 90-Revised (SCL-90-R; Derogatis, Reference Derogatis1983) was used to measure general psychiatric complaints. This is a 90-item self-report instrument where patients are asked to rate symptoms on a 5-point Likert scale ranging from 0 = None to 4 = Extreme. The General Symptom Index was used, and the SCL-90-R has been shown to have good psychometric properties (Bech et al., Reference Bech, Allerup, Maier, Albus, Lavori and Ayuso1992).
The Beck Depression Inventory (BDI; Beck, Rush, Shaw and Emery, Reference Beck, Rush, Shaw and Emery1979) is a 21-item self-report instrument that measures depression during the last week. It has been shown to be both a reliable and valid measure of depression severity in both clinical and non-clinical populations (Beck, Steer and Garbin, Reference Beck, Steer and Garbin1988).
The Beck Anxiety Inventory (BAI; Beck, Epstein, Brown and Steer, Reference Beck, Epstein, Brown and Steer1988) is a 21-item self-report instrument that measures anxiety severity for the past week. The instrument is established as a reliable and valid measure of anxiety, and often recommended as a companion instrument to the BDI (Beck and Steer, Reference Beck and Steer1993).
The Young Schema Questionnaire (YSQ; Young, Reference Young1994) is a 205-item self-report questionnaire that measures 16 early maladaptive schemas. Items are answered on a 6-point Likert scale, ranging from 0 = Almost never true on me to 5 = Almost always true on me. For the current study, the total score was used. YSQ has demonstrated good levels of psychometric properties and clinical utility (Schmidt, Joiner, Young and Telch, Reference Schmidt, Joiner, Young and Telch1995).
Process measure
The Group Climate Questionnaire – Short Form (GCQ-S; MacKenzie, Reference MacKenzie, Dies and MacKenzie1983) is a 12-item self-report measure that assesses individual group members' perceptions of the group's therapeutic environment. Each statement is rated on a 7-point Likert scale, ranging from 0 = Not at all to 6 = Extremely. The GCQ-S consists of three factor-analytically derived subscales: (a) Engagement (5 items that call for ratings on self-disclosure, cognitive understanding and confrontation); (b) Avoidance (3 items – to what extent the group member avoids responsibility for their change processes); and (c) Conflict (4 items – measures interpersonal conflict and distrust between group members as well as withdrawal). The instrument is not based on any particular theoretical orientation, and is thus applicable to many different group situations and formats. The GCQ-S has been widely used across different treatment populations, and its construct validity has been tested extensively (Kivlighan and Goldfine, Reference Kivilighan and Goldfine1991; Tschuschke and Greene, Reference Tschuschke and Greene2002). Internal consistency of the GCQ-S subscales has been shown to be high, with alpha coefficients ranging from .88 to .94 (Kivlighan and Goldfine, Reference Kivilighan and Goldfine1991). Patients completed the GCQ-S the week before termination.
Results
Preliminary analyses
First, a one-way analysis of variance (ANOVA) was computed to investigate potential statistical significant differences between the groups in terms of overall quality of group climate. No statistical significant differences were found between the groups in terms of Engagement (F = 2.69, p = .06), Conflict (F = .25, p = .91) or Avoidance (F = 2.39, p = .08). It was thus deemed appropriate to analyse the whole sample, while not controlling for group condition. Descriptive analysis of each subscale on the GCQ-S revealed the following scores: Engagement (n = 27) M = 3.9; SD = 0.7; Avoidance (n = 27) M = 2.3; SD = 0.7; Conflict (n = 26) M = 0.5; SD = 0.3. Further analyses (Pearson's r) revealed that neither pre- nor post treatment scores on any of the outcome variables correlated significantly with any of the group climate variables (p >.05).
Main analyses
Next, in order to assess the relationship between ratings of group climate and long-term follow-up outcome, a series of hierarchical multiple regression analyses was computed. In each regression model, pre- and post treatment scores of the outcome variable were entered in the first two steps to partial out potentially confounding effects. Subsequently, each of the group climate variables (engagement; avoidance; conflict) were entered one at a time in step three in separate analyses for each outcome variable. Due to a large number of analyses, a conservative α of p < .01 was used in the analyses. A summary of step three is presented in Table 2.
Table 2. Prediction of one year follow-up using the Group Climate Questionnaire-Short form
Note: **p < .01. ***p < .001; d = Cohens' effect size (d = 2t/√(df)) (d >.5 = medium effect size, d >.8 = large effect size)
As shown, higher ratings of engagement were associated with better follow-up outcomes on all outcome measure except the BAI. Higher ratings of avoidance predicted a statistically significant reduction in anxiety symptoms at follow-up. Ratings of conflict did not predict follow-up outcome. Moreover, the analyses revealed that pre-scores on all outcome measures were statistically related to follow-up scores (p < .01). For the post-outcome scores, only the SCL-90-R was statistically related to follow-up scores (p < .01).
Discussion
This was the first study to investigate the predictive validity of a measure of group climate (GCQ-S) for long-term (1 year) follow-up in a randomized controlled trial of cognitive-behavioural group therapy. Overall, the findings indicated that higher ratings of engagement were strongly related to a favourable one-year follow-up outcome on nearly all outcome measures. This finding is consistent with the results of some previous studies using the GCQ-S with non-psychiatric samples (Braaten, Reference Braaten1989; Kivlighan and Lily, Reference Kivlighan and Lilly1997; Kivlighan and Tarrant, Reference Kivlighan and Tarrant2001). It also extends the results from the study by Ogrodniczuk and Piper (Reference Ogrodniczuk and Piper2003) by showing engagement or cohesion to be an important predictor for long-term follow-up in a highly structured treatment like CBT-group therapy and with a different diagnostic sample.
Mackenzie (Reference MacKenzie, Dies and MacKenzie1983) has argued that engagement captures many of the essential elements of cohesion, and may be described as an indicator of the positive bonding relationship that is usually thought of as a necessary ingredient in any effective therapy. Further, engagement might also be said to reflect work and self-disclosure among the members, as well as group members' attempts to understand the meaning of their behaviour. Positive change is more likely to occur with these behaviours (Ogrodniczuk and Piper, Reference Ogrodniczuk and Piper2003). Clearly, a positive working atmosphere seems important for a positive outcome, and the usually limited time available in short-term group treatment may well function as a positive catalyst for the development of engagement. A strong-enough sense of engagement may be a necessary and critical ingredient for change to take place, although it is not necessarily an increase in engagement or other relationship qualities throughout the treatment process that causes therapeutic change (Ogrodnizcuck and Piper, Reference Ogrodniczuk and Piper2003; Lindgren, Barber and Sandahl, Reference Lindgren, Barber and Sandahl2008).
Our findings add to the study of Ogrodnizcuk and Piper (Reference Ogrodniczuk and Piper2003) by demonstrating that perceived engagement may not only be predictive of in-treatment change, but also of long-term follow-up in a different treatment modality and with a different diagnostic sample. Although Ogrodnizcuk and Piper (Reference Ogrodniczuk and Piper2003) compared short-term interpretive- and supportive dynamic therapies, which are technically different, one similarity between them is that they both rely heavily on the material that patients bring into treatment. That is, patients are encouraged to contribute actively to both the content and process of therapy, sessions are not highly structured and therapists are usually not very directive. CBT-oriented group treatments, on the other hand, are usually more structured from session to session, therapists are encouraged to be active and a heavy emphasis is placed on the implementation of pre-planned therapeutic tasks as well as homework assignments (Beck et al., Reference Beck, Rush, Shaw and Emery1979). The present study indicates that although CBT-based group treatments may be technically different from dynamic- and experiential therapies, relationship factors such as cohesion or engagement may still be important for treatment outcome.
The question arises, however, in what way engagement is related to outcome and, in turn, how engagement relates to the use of specific techniques. Should engagement be characterized as causing a positive outcome, being the result of prior, in-treatment symptom reduction or therapeutic gains, or simply (spuriously) correlated with outcome. A similar discussion has received much attention in the alliance-research literature in individual therapy, with authors advocating all three positions (Baldwin, Wampold and Imel, Reference Baldwin, Wampold and Imel2007; Barber, Connolly, Crits-Christoph, Gladis and Siqueland, Reference Barber, Connolly, Crits-Christoph, Gladis and Siqueland2000; DeRubeis and Feeley, Reference DeRubeis and Feeley1990). As recently argued by Hatcher and Barends (Reference Hatcher and Barends2006), this debate may be the result of confounding two different levels of analyses since alliance and technique occupy different conceptual levels and cannot be considered to be two different types of activity in therapy. Technique is an activity, alliance is a way to characterize activity. Since engagement largely parallels the concept of the alliance, we caution somewhat against interpreting the results from the present study as indicating that the use of technique is unimportant. Probably, technique and relationship factors interact and are catalytic to each other both in terms of process and outcome (Safran and Muran, Reference Safran and Muran2000).
Ratings of engagement were not predictive of anxiety-symptoms at follow-up, which is an important qualification to the aforementioned general effect of perceived engagement on outcome. This points to a potential limit of the effect of the therapeutic relationship that warrants further research. Moreover, contrary to our second prediction, higher ratings of avoidance were associated with reduced anxiety-symptoms (BAI) at follow-up. This finding came as a surprise, since higher ratings of avoidance are usually thought to capture a sense of lack of personal commitment to the therapy process (MacKenzie, Reference MacKenzie, Dies and MacKenzie1983). Further, although not statistically significant, similar trends between higher ratings of avoidance and better outcomes were apparent at follow-up for all other outcome measures as well (see Table 2).
A closer inspection of the three items comprising the avoidance dimension on the GCQ-S indicates a possible answer, and points to potential important structural differences between CBT based group therapy, on the one hand, and more dynamic- and experiential therapies on the other. For example, item 5 on the GCQ-S asks the patient to what extent the members of the group depended on the group leader for direction in the treatment process. Although a higher score may be seen as avoiding personal responsibility in the change process in dynamically oriented treatments, it is actually in accordance with what one would expect as appropriate patient behaviour in a CBT-oriented group treatment. In CBT, the therapist is supposed to be active and directive, and to structure each session carefully. Moreover, item 9 asks to what extent the group members appeared to be doing things the way they thought would be acceptable to the group. A higher score here is also thought to indicate less commitment to the therapy process, but actually makes sense if group members are supposed to be engaged in a pre-planned structured task. A low score might actually be indicative of being at odds with the rest of the group, or possibly of introducing a personal agenda in the therapy process. Conclusively, this finding indicates that ratings of the avoidance dimension may be partly dependent on the specific treatment structure, and should caution therapists and researchers against interpreting the scores in a context-free fashion.
Lastly, we did not find any support for our third hypothesis, namely that lower ratings of conflict would be related to an overall better one-year follow up. Ratings of conflict were unrelated to follow-up outcome, which may actually reflect an overall low level of conflict at this treatment phase (termination). The non-significant finding may also stem partly from positive therapist behaviours, reducing tension and aggression within the group. However, since the drop-out rate in the original study was 30%, another hypothesis is that patients experiencing a high level of conflict would already have terminated treatment and be considered non-completers. This would obviously restrict the range of scores, making it difficult to establish any statistical significant associations between the variables. Since perceived group climate was measured the week before termination, it was not possible to investigate this hypothesis post-hoc.
It is important to bear in mind that the present study measured patients' perceptions of the group climate, which is subjective and personal, and may or may not reflect “the real” group climate. The results indicate that patients in the same group may have quite different experiences and perceptions of the group climate, which highlights the important intrapersonal elements of the group climate, or member-to-group experience (Burlingame, Fuhriman and Johnson, Reference Burlingame, Fuhriman, Johnson and Norcross2002). This should also caution therapists against using solely their own judgement or “gut feeling” as an indicator of the group climate. Moreover, it is also important to note that groups may differ in terms of overall group climate. Indeed, there was a trend in that direction in terms of engagement (p = .06) and avoidance (p = .08), although not statistically significant. This indicates that although there is an important subjective dimension to the experience of group climate, some groups do probably have an overall more favourable group climate than others.
The finding that perceptions of the group climate may be related to follow-up outcome has important clinical implications. First, developing a group climate where patients are active and engaged may be seen as perhaps a common therapeutic task in all group treatments, but do not necessarily conflict with the use of other modality-specific techniques or structured tasks. On the contrary, the heavy emphasis on structure and pre-planned in-session activities in CBT-oriented group therapy may actually propel engagement as well as reduce anxieties and concerns about the treatment structure and process (Stockton et al., Reference Stockton, Rhode and Haughey1992). Providing a clear treatment rationale, setting an agenda and giving homework assignments are other examples of creating structure and engagement.
Second, therapists should try to identify patients that are relatively less engaged, since they are at risk for a poorer long-term follow-up. This is probably best achieved through collecting data on perceived group climate throughout the treatment process, on a session-by-session basis. Necessary steps should be taken to improve the conditions for patients at risk, depending on the patient's reasons for his or her lack of engagement. For example, if a lack of engagement is due to a perceived mismatch between the demands of a specific task and the personal resources available to the patient, the therapist could either provide the patient with additional help and support to master the task or, alternatively, reschedule for another and more manageable task. Alternatively, if the patient responds negatively to the group treatment structure because of issues related to autonomy, the patient might be rescheduled to individual therapy (Zettle, Halfich and Reynolds, Reference Zettle, Halfich and Reynolds1992).
Limitations
One potential limitation of the present study was that group climate was measured at only one time point (the week before termination). Although we did control for symptoms at both intake and treatment termination, we can thus not completely rule out the possibility that ratings of group climate may have been related to prior symptom reduction. Second, although the patients did not receive any further formal therapy in the follow-up period, we do not know whether the patients had any informal contact with each other in that time period. Third, there is a risk that the results are somewhat overstated due to the fact that the group climate variables were treated as independent data, ignoring the group level. Fourth, dimensions of group climate may wax and wane throughout treatment, and be related to both process and outcome in more non-linear ways than suggested in this study, as hypothesized by MacKenzie (Reference MacKenzie, Dies and MacKenzie1983). Future studies should try to disentangle the relationship between improvement and group climate by including repeated measures. Lastly, patients' ratings of group climate may not necessarily reflect the “real” climate in the group, and it is important to note that the present study only examined patients' perceptions of the group climate as predictors of long-term outcome.
Conclusion
The study found partial support for the use of the GCQ-S as a predictor of long-term follow-up in CBGT for patients with comorbid psychiatric disorders. Perceived engagement was strongly related to most outcome measures, which underlines the importance of relationship factors in an otherwise highly structured treatment approach. However, the lack of support for the avoidance and conflict scales calls into question whether the GCQ-S is a fully appropriate measure for CBGT. Further research with repeated measures is needed to fully evaluate the usefulness of the GCQ-S in CBGT.
Comments
No Comments have been published for this article.