Abstract
This paper covers ideas presented at a workshop on group-analytic research1. As an introduction, a few studies from the early history of group research are mentioned in addition to excerpts from the group chapter in the ‘Handbook of Psychotherapy and Behaviour Change’, which has been a regularly updated source of research methodology and results during the last 50 years. More specific research on group-analytic and psychodynamic group psychotherapy is partly represented by the Sheffield report, a narrative meta-analysis of studies on Group Analysis (GA) and Analytic/Dynamic group therapies up to 2009. Furthermore, the author presents studies by his own research group: one observational study of 69 outpatients treated in ‘traditional’ GA, plus outcome results from a randomized clinical trial where 167 outpatients were treated in manualized short- and long-term group analytic psychotherapy and followed-up 3 and 7 years after baseline. Inspired by results from this research, a ‘Focused Group Analytic Psychotherapy’(FGAP) suitable for patients with a higher level of personality integration, has been constructed. Conclusion: Although there is a scarcity of high-quality research, outcome studies show that group analytic (and other psychodynamic) therapies, both short- and long-term, are efficacious for patients treated in trans-diagnostic groups (Burlingame and Strauss, 2021: 614). Patients with a higher level of personality organization seem more often to be sufficiently helped by a time-limed group-analytic approach (FGAP; Lorentzen, 2022) while patients with more personality pathology need longer therapies.
Introduction
What is research? A simple answer is ‘to study something in a systematic way’. This ‘something’ may be an object of different sorts, a text, a collection of statements, or aspects of an activity or a craft, like f. ex. group analysis. Focus can be on feelings, on individual overt behaviour, or interactions, be it from an episode in a session, a sequence of sessions or even the whole therapy, lasting for years.
Why do we want to do research? A simple answer is that if we believe in what we ‘preach’ about unconscious motives and conflicts, we take seriously that we as humans tend to deceive ourselves, that we have blind spots that may interfere with an objective evaluation of our own work. This has also been verified in studies showing that some therapists regularly have a larger number of patients who deteriorate or remain unchanged under their care (Lambert and Shimokawa, 2011), and many of these therapists feel that they are doing a good job!
Group research, in general
At the present stage of patient care in the mental health field, an interesting or relevant question could be: What kind of treatment works for whom? —and we could add, and why? It would also make it easier if these areas of interest were shared by most patients, as well as therapists, mental health organizers, and not the least third-party payers!
When I started my training GA some 40 years ago at a time I was about to finish my psychoanalytic training, there were few quantitative studies in both GA and psychoanalysis. There were also strong resistances in both camps against doing research in a more ‘modern’ sense, one reason being that it was seen to interfere negatively with the therapy process. Although the situation on the research front now is somewhat improved, there are still few outcome and process studies of group-analytic therapies, and a lot of resistance when it comes to doing research.
Some of the material presented refers to group psychotherapy in general — that is of therapies with diverse theoretical background. However, it may still belong here since results, interactions, and mechanisms may be similar in different brands of group therapy.
Short history of group research
As to history, group psychotherapy has been practised for more than 120 years and researched for about the last 80 years, since about when Foulkes, the father of GA, started to work with groups in his private practice (around 1940). Early reports of outcome were at first anecdotal, describing single groups. Reviewers summed up numbers of studies with positive, negative or no effect, the so-called box score method (Burchard et al., 1948; Thomas, 1943). Smith and Glass (1977) introduced a new review strategy, called the meta-analysis: The sum of several outcome measures was averaged (effect-size), allowing for direct comparisons with other studies. The efficacy of group-therapy compared to control groups was demonstrated from the 1970s (Bednar and Kaul, 1978; Emrick, 1975; Lieberman, 1976; Luborsky et al., 1975; Meltzoff and Kornreich, 1970; Fuhriman and Burlingame, 1998: 14–20; Burlingame et al., 2016). One general conclusion in controlled studies has been that group psychotherapies of different theoretical orientations are effective treatments for a wide variety of psychiatric populations, often with comparable effect-sizes. Comparisons between group and individual treatments have given the same results: no differences in outcome. One such review included 23 studies where the individual and group formats were used within the same study (McRoberts et al., 1998). There was heterogeneity as to diagnosis and treatment, but most studies were based on group Cognitive Behavioural Therapy (CBT). Average length was 16 sessions. The reasons behind similar outcomes may be that many non-specific (human) factors are effective in all forms of psychotherapy, and/or that our research methodology is of limited value, i.e. not sensitive enough to pick up real differences.
The Bible of psychotherapy research
This name has been used to characterize the Handbook of Psychotherapy and Behaviour Change, which has been a rich source of research methodology and results, appearing regularly with seven editions between 1971 and 2021; four were edited by Bergin and Garfield, two by Lambert and the last one by Barkham et al. Especially important for this presentation is the chapter on group psychotherapy research, which continuously has been updated over the years, last time by Burlingame and Strauss (2021).
In the last editions a modified version of Barlow’s (2013) group typology within the mental health services is presented, the three main categories being self-help groups, psycho-educative groups and psychotherapy groups, the last being in focus here.
Space does not allow too many details, however, so only a few samples of the contents of the group chapter is presented here: The five-factor model, a recent meta-analysis of outcome studies, samples of updated process studies and some thoughts about the use of patient feedback in clinical work (Figure 1).
Factors contributing to change in group psychotherapy (Reworked from Burlingame, MacKenzie and Strauss, 2004).
A quick look at a simple five-factor model (Burlingame et al., 2004), demonstrates how it is possible to conceptualize small group treatments. Therapy outcome is placed in the centre and the model identifies potential variables influencing treatment outcomes (directly as predictors, or in interaction with others, as moderators or even possible mediators). The model groups the variables under headings such as structural factors, patient and therapist (leader) variables, small group processes and formal change theory. Structure can for example be groomed through preparation of patients, variations of leadership style, choice of therapy length or session frequency. Patient variables can be chosen from demographics, aspects of psychopathology or traits of personality. Leader variables are for example personality traits, length and type of professional training and experience. Small group ecosystem (processes) consists of therapeutic elements, aspects of interaction, therapeutic factors, group development, alliance or cohesion. Formal change theories refer to psychodynamic, cognitive behavioural or behavioural therapy, systems theory, existential, etc.
Outcome studies
The number of included randomized clinical trials (RCTs) have increased over the years and the methodology used in meta-analyses is improved, and in the last edition (Burlingame and Strauss, 2021) 10 meta-analyses are reviewed, summarizing 329 RCTs comprising 27000 patients, from the last 30 years. Results: Group psychotherapy is effective compared with non-active treatments and equivalent to other active treatments (including individual psychotherapy), for various mental disorders. Most of the studies are on group CBT, but next to that the psychodynamic/group-analytic models have the longest tradition and are found in studies of the treatment of patients with Borderline PD (15), eating disorders (12) and Post Traumatic Stress Disorders (PTSD; 14). Number of psychodynamic studies are noted in parenthesis. Psycho-educative groups were also widespread, and mainly found effective for patients with schizophrenic and affective disorder (bipolar). It should be mentioned that most of the efficacy studies have been conducted in outpatient settings and that they mainly are focused on short-term groups. The last summary of inpatients (Kösters et al., 2006) demonstrated a small and a large effect-size for controlled and observational studies, respectively. A further limitation regarding the interpretation of the results refers to their validity for a wider range of treatments, because of a gap between clinical-theoretical considerations of psychodynamic and group analytical approaches and other studies in the group therapy literature (Rosendahl et al., 2021: 56).
Process studies
Cohesion (see below) has been the most commonly studied factor in groups, while therapeutic alliance as relational factor traditionally has been studied in individual psychotherapy. More recently, however, the alliance between the therapist and the individual patients has caught more interest, also in group research. One such study found a significant weighted average correlation (r=0.17) between member-leader alliance and treatment outcome in a meta-analysis of 29 studies including 3628 group patients (Alldredge et al., 2021). This is low compared with the correlation (r=0.28) found for alliance in individual treatment (Flückiger, Del Re et al., 2018; Rosendahl et al., 2021: 57). Some moderators were observed in the same study, for example that the heterogeneity in study results could be explained by treatment orientation (lower correlation for CBT than for other group treatments) and the reporting perspective (higher correlation for patient-reported than for mixed or observer-reported alliance). Cohesion refers to the sense of connection or closeness among group members, one relatively recent meta-analysis included 55 studies and more than 6000 patients and demonstrated a significant correlation of moderate effect-size (r=0.26) between cohesion and improvement (Burlingame et al., 2018). This is about the same value reported above for the therapeutic alliance in individual therapy (r=0.28). A moderating effect was found for theoretical orientation, highest correlation observed for interpersonal groups, followed by psychodynamic, cognitive, supportive and eclectic therapy groups. The cohesion-outcome association was stronger when group leaders emphasized member interaction and when groups were long-lasting.
The Group Questionnaire
From the patients’ point of view, cohesion consists of three structural components: the patient-patient relationship, the patient-group relationship and the patient-therapist relationship. Johnsen et al. (2005) sought to achieve a conceptual clarification of cohesion and carried out a factor analysis of the items in the four most commonly used cohesion measures (group climate questionnaire, cohesion, therapeutic alliance and empathy). This resulted in three qualitative factors encompassing two affective aspects (positive emotional bonds and negative relationships) and one work-based facet of group relationships (positive work). These qualitative factors were then intersected with the three structural patient factors mentioned above, to capture the diversity of the group relationships. The factors have been partially validated in follow-up studies (Bormann and Strauss, 2007; Bakali et al., 2009). A user-friendly cohesion measure, The Group Questionnaire (GQ), was arrived at by reducing the number of underlying items (Krogel et al., 2013; see also Lorentzen, 2022: 150).
Feedback systems
Use of feedback systems seems to become increasingly popular in group psychotherapy. Active early proponents were Lambert et al. (2001), who used the Outcome Questionnaire-45 continuously in individual psychotherapy to detect problems early in therapy, a strategy that turned out to result in improved outcome and in a decrease in attrition. A range of methods is also available to monitor the effects of group treatment on individual’s symptoms and other characteristics, such as group cohesion. There has been an increase in use of GQ in clinical practice, and the following two studies demonstrate that continuous feedback to therapists on patients’ experience of group relationships resulted in significantly better treatment outcomes and fewer premature terminations (Burlingame et al., 2018; Gleave et al., 2017).
Attachment
Among patient characteristics that may influence group therapy outcomes, attachment has gradually come more into focus (Marmarosh, 2017). It has also become clearer that interaction with responsive and supportive leaders and cohesive groups, beneficially alter a person’s attachment patterns and psychological functioning (Marmarosh, 2013, 2017). This personality characteristic may very well be a mechanism of change and should therefore be the target for increased and specific group analytic research in the future.
A systematic review of the efficacy and clinical effectiveness of group analysis and analytic/dynamic group psychotherapy, December 2009
Now let us turn to more specific studies of GA research, starting with the Sheffield report (Blackmore et al., 2009, 2012). This review came about because GA seemed gradually to loose its position compared to CBT within the National Health System (NHS) in the UK, possibly because group CBT offered shorter therapies and produced more empirical data supporting its efficacy/effectiveness. Consequently, Institute of Group Analysis (IGA) and Group Analytic Society (GAS), London commissioned University of Sheffield (Centre for Psychological Services Research, School of Health and Related Research) to evaluate the evidence base for GA and Analytic/Dynamic group psychotherapy (A/D).
The mandate was to provide a systematic review of efficacy and effectiveness of these therapies, to look for factors that influenced the results, and there was also a wish for a clearer description of the patients who underwent GA, including duration of their therapies. IGA/GAS also gathered an expert committee who met the Sheffield group several times over 1.5 years to assist with search words, selection of papers, etc.
Results/conclusions
Data search of the literature gave altogether 34 primary studies from the years 2001–2008. Five of the studies were randomized controlled studies (1 GA), and there were seven other controlled studies, 21 observational studies (10 GA) and one qualitative study. In order to catch all previous studies, 19 reviews of outcome studies, some carried out before 2001 were also included. The results showed that all the studies, including earlier reviews, supported the use of psychodynamic group psychotherapy as an effective approach, across diverse conditions, participants, groups, and settings. There might be important effects of age, sex and self-efficacy, psychological mindedness and QOR (quality of object relationships) on outcomes, and attachment style and interpersonal distress had a bearing on group attendance.
Limitations/recommendations
There were also several limitations in the literature: The data search gave no information as to types of patients for whom GA and A/D group therapies were most effective. The different indications for group versus individual psychotherapy, and the comparable cost-effectiveness had not been studied. There was no data on what degree the aspects of heterogeneity versus homogeneity of group membership impacted on outcome. The methodological quality of studies varied, and the Sheffield group had the following recommendations to the field: It is urgent to do trials of GA and A/D group therapies compared with group CBT. It is also important to make a review of the service users’ personal testimony. To improve reporting (publishing) of research, authors were encouraged to use structured abstracts, to give clear definitions of interventions and use agreed keywords in titles and abstracts and develop (and use) a consistent set of outcome measures.
Now to three group analytic research projects carried out with the author as principal researcher:
Long-term analytic group psychotherapy with outpatients. Evaluation of process and change (Lorentzen, 2003)
This project was planned and initiated in a newly established private practice. The therapist (the author) had several years’ experience of clinical work, he was certified as psychoanalyst and well into a group-analytic training programme. The choice of research questions was: What is the effectiveness of long-term group-analytic psychotherapy? What factors predict change? What factors moderate change?
Patients/treatment/evaluation
This is an observational (naturalistic) study of 69 outpatients with the diagnoses of mostly depression and/or anxiety disorders on Axis I, and 68 % with one or more PDs. The patients were treated in three slow-open groups, and the mean duration of therapy = 32.5 months, 90 min/week (range 6 months to 8 years). The therapy approach was group analysis, as developed by Foulkes and successors (Foulkes, 1977; Foulkes, 1984; Foulkes, 1986; Foulkes and Anthony, 1984; Pines, 1983). The patients were evaluated pre-post, and each patient also had a one-year follow-up interview with the therapist and an independent evaluator.
Results
61–86 % of the patients improved or moved into a non-clinical part of the population (recovered) as to symptoms, interpersonal problems and/or psychosocial functioning. Eight patients remained unchanged, and no patients deteriorated.
Ad predictors, treatment time up to 2.5 years predicted a better outcome from group therapy. Patients with chronic symptoms, PD, negative expectations, and more hostile interpersonal relationships usually improve less in short-term group psychotherapy. In this study, however, these patients changed just as much as the others, indicating that predictors of outcome in long-term therapy are different from those found in short-term therapy.
The following moderators indicate which patients needed longer treatment (Lorentzen and Høglend, 2008): those with more serious depression, with higher symptom load (Interpersonal sensitivity, Subscale of SCL-90-R; Derogatis, 1977), PDs in clusters A and B, and patients who were more submissive, exploitable or intrusive on subscales of IIP (Inventory of Interpersonal Problems; Alden et al., 1990).
Short- and long-term group analytic psychotherapy. A randomized clinical trial (Lorentzen et al., 2013, 2015; Fjeldstad et al., 2016)
This randomized clinical trial lasted for several years and has so far resulted in around 25 publications in peer reviewed journals, two books, several book chapters and two doctoral theses: Group processes in short- and long-term psychodynamic group psychotherapy (Bakali, 2013) and The role of treatment duration in psychodynamic group psychotherapy (Fjeldstad, 2019).
The research questions in this project were, among others: How important is time in psychotherapy for outcome? Are there differences in change in patients after having received short- or long-term group-analytic therapy (STG and LTG), respectively? Change in symptoms and interpersonal problems were measured repeatedly until seven years after start of therapy, and global assessment of functioning was measured at therapy start and after three years (see fig. 2). Differences in outcome were studied both for the typical (average) patient, and for patients with or without personality disorders. Patients were interviewed at follow-up three years after therapy start, and the last patient report on symptoms and interpersonal problems was collected at a seven-year follow-up.
Time axis and measures.
Sites/patients/diagnoses/groups/therapists
167 patients were recruited from three urban sites in Norway. Inclusion criteria were at least one diagnosis on Axis I or II (DSM-IV), self-perceived interpersonal problems, motivation to work on these problems in a group setting, and agreement to randomization. The patients were randomized (stratified on gender) to short- and long-term GA and allocated to 18 groups, with 7–8 patients in each group. There were 9 therapists, and each conducted both one short- and one long-term group. The durations of the treatments were 20 versus 80 weekly sessions of 90 minutes each (altogether 6 or 24 months). The sample consisted of 62 % women and 38% men. Mean age was 38.7 (9.4) years. As to diagnoses, 80% had mood and 60% anxiety disorders. Number of axis 1 diagnoses was mean = 3.3. Forty-five % had one or more personality disorder(s), mostly mixed and avoidant PD, and 65 % had been in previous treatment. The patients had a moderate load of symptom distress and interpersonal problems.
Timeline and measures
Concerning statistical analyses, linear mixed models for repeated measures were used to analyze longitudinal data. Both intragroup and between-group differences were calculated.
Manuals and treatment fidelity
The therapies were manualized, describing theory, methodology, techniques and giving clinical examples within the two formats (Lorentzen, 2014). Both therapies were group-analytic, but the short-term group therapy differed in a few respects: more time was used on pre-therapy preparation and to negotiate a therapy focus, the therapist was more active, worked more in the here-and-now, and was more alert to specific stages in the group process (opening, differentiation, interpersonal work and termination). To check treatment fidelity (therapist adherence to manual and therapeutic competence), thirty-nine audio recordings from the first six months of therapy (sessions 3, 10 and 17) were drawn from the pool of 54 recordings. Two evaluators who were masked to group, independently rated sessions on a scale with diverse process variables and one on therapist competence, both Likert scales from 0 (not at all) to 4 (very much). Intraclass correlation on the therapy process scales ranged from 0.70 to 0.94, and the therapist mean competence was similar in both formats, ranging from moderate to high (Lorentzen et al., 2014: 283).
Results in the RCT
It turned out that group-analytic therapies, both short and long, were efficacious for outpatients with mixed diagnoses (depression, anxiety disorders, mild/moderate PD). The typical (average) patient changed quicker and more during STG (6 months) than in LTG, but the change was equal in the two treatments after three years (Lorentzen et al., 2013). The effect was sustained in STG up to seven years, but LTG patients also had a delayed effect (patients continued to change after termination). Patients with one or more personality disorders (PD+) did equally well in STG and LTG during the first six months (STG). However, PD-patients in LTG also demonstrated a delayed effect up to seven years follow-up, while the early effects in the STG groups were sustained after seven years (Lorentzen et al., 2015; Fjeldstad et al., 2016). Patients without PD also changed faster in the STG, and LTG patients used a couple of years to catch up. An important point is that these LTG patients (in addition to having a slower improvement) never exceeded the improvement in the STG patients during the last five years of the follow-up period! These findings may have important implications for what kind of group (short- or long-term) these patients primarily should be allocated to, which underlines the importance of a thorough evaluation of the patients’ personality before therapy starts.
Deteriorations are often not reported in psychotherapy studies. In this study, 5.8 % were deteriorated, across all outcome measures and treatment lengths at three years follow-up. There were no serious adverse events (suicide or suicide attempts).
Focused Group Analytic Psychotherapy (FGAP)
As a consequence of the outcome research presented above, I decided to construct a new, time-limited therapy, Focused Group Analytic Psychotherapy (FGAP; Lorentzen, 2022). The main rationale was to be able to more adequately serve the relatively large group of patients without PD (with less personality pathology) who not only turned out to change faster in STG than patients in LTG, but who also seemed to have the same sustained long-term improvement as patients treated in LTG. Personality pathology and personality disorder as diagnosis may have been treated too lightly in group analytic circles. Maybe the issue should have been to actively try to understand and describe these patients better, in order to clearer differentiate between treatment of patients harbouring and struggling with different types and degree of pathology (both in pre-therapy preparation as well as during therapy). In the observational study described above (Lorentzen, 2003) all patients were treated according to more ‘traditional’ group-analytic guidelines, the therapist being more pending, leaving more initiative and responsibility to the group members. In hindsight, it is reasonable to think that some of these patients harbouring less personality pathology, might have settled for a shorter therapy, if the therapist had worked more actively towards that end. Similarly, some of the participants in the RCT described above who dropped out of the therapy, might have stayed on if their tendency to act out had been properly explored and attended to during the pre-therapy preparation (see inoculation; Whittingham, 2018). Karterud (2025) has in a recent book criticized group analysis for neglecting the challenges represented by patients with personality disorders, particularly borderline patients.
While constructing the new therapy, the manual for the short-term group analytic psychotherapy (Lorentzen, 2014) was taken as a point of departure. Both manuals were already from their inception based on research and clinical experience but could now be scrutinized again to see whether revisions should be done based on new research results and the increased clinical experience we had attained at this stage. During this process, I was struck by the lack of instructions on how to evaluate personality and personality pathology (except for a SCID II-interview, which could result in one or more descriptive/categorical diagnoses). This is unfortunate since level of personality organization is strongly related to how the individual members manage in a group and for the therapeutic quality of the group’s work. Therefore, I found it most urgent to develop the issue of personality integration and pathology further and wrote Focused Group Analytic Psychotherapy (FGAP; Lorentzen, 2022), where evaluation, selection and preparation of patients are described in detail, and where personality issues are explicitly included.
Lack of space only permits a brief mentioning of some main points. Let me first mention the importance of establishing the patients’ level of personality organization (PO) relatively early in the process. Next in line comes the task of developing a case formulation, including a specific psychodynamic hypothesis. Elements of the hypothesis can in concert with aspects of interpersonal problems and symptoms be formulated (or loosely woven together) as a therapy focus. The pre-therapy preparation has also been given a more prominent place, i.e. more time being spent on negotiating different dynamically charged issues (therapy focus). Other topics underlined in the new book are higher therapist activity, work in the here-and-now, and the need of taking the group through four more or less clearly defined phases (opening, differentiation, interpersonal work and termination).
FGAP is based on group-analytic and psychoanalytic theory, and I have included some of Otto Kernberg’s psychoanalytic theories (1980) in order to vitalize FGAP. He has through several decades of work systematically developed his theories on personality integration, including aspects from several areas of psychoanalysis like ego-psychology, object relations theory and self-psychology. Combining group-analytic theory with central aspects of his rich theoretical framework (see f. ex. Caligor et al., 2018), makes it possible to continue Foulkes’ tradition of running heterogeneous (transdiagnostic) groups, at least within certain limits. Kernberg operates with five levels of personality organization: 1. Normal - 2. Neurotic - 3. High - 4. Moderate - 5. Low level borderline. Only patients from the first three levels are suitable for FGAP. The evaluation should be thorough and based on information that is obtained by a combined clinical/psychodynamic interview by a relatively experienced clinician. This evaluation should focus on the quality/impairment in the following personality domains: sense of identity, quality of object relations, maturity of defences, affect tolerance/control (aggression), moral values, and degree of pathological narcissism (Caligor et al., 2018: 552). Each domain is rated on a scale from 1–5, and the mean decides the level. A structured interview (STIPO-R; Clarkin et al., 2016) also exists, but is mostly used in research. This interview is informative, however, as to how questions can be formulated to expose different aspects of the diversity of personality domains and can thus be useful for training of less experienced clinicians. Summing up: Patient suitability for this therapy can be decided through a thorough evaluation/preparation (3–5 sessions). The level of PO should be established, the patient’s case formulation/psychodynamic hypothesis should be teased out, and a treatment focus should be formulated, combining intrapsychic conflicts, interpersonal problems and behavioural tasks. A profile of interpersonal problems, using the Inventory of interpersonal problems-Circumplex (Alden et al., 1990), should also be teased out for each patient, and so should a clinical diagnosis. Again, a more detailed instructions about evaluation, preparation, and group composition can be found in Lorentzen (2022). The book also includes basic theory, describes the stages in the group process, therapist interventions, plus offers a multitude of relevant clinical vignettes.
Conclusion
This narrative presents some cohesive thoughts on group research in general, and specifically on psychodynamic and group-analytic research. Although there is a scarcity of high-quality research, outcome studies show that group analytic (and other short- and long-term psychodynamic) therapies, are efficacious for patients treated in mixed (trans-diagnostic) groups (Burlingame and Strauss, 2021: 614). Patients with a higher level of personality organization are often sufficiently helped by a time-limed approach (FGAP, Lorentzen, 2022) while patients with more personality pathology need longer therapies (Fjeldstad et al., 2016). Many factors that predict or moderate therapy outcomes are known but may vary according to sample composition or length of therapy. Next to nothing is written about mediators of change (mechanisms of change). More detailed knowledge via high quality studies on process-outcome connections is needed. In group analytic therapies evaluation of personality pathology and its destructive impact on the dynamic matrix should be actively addressed before and during therapy.
