Premature termination of treatment is a significant problem in individual psychotherapy as well as in group therapy for its high prevalence and numerous negative consequences. A meta-analysis conducted of 125 studies on psychotherapy dropout revealed that the average outpatient psychotherapy dropout rate was 46.8% across a wide range of settings, diagnoses, and treatment patients [1]. In individual psychotherapy, early termination has been associated with reduced treatment efficacy [2,3]. Also, the fact that issues related to the termination are not explored can lead to feelings of dissatisfaction, failure and discouragement in patients and ultimately to their functional deterioration [4,5]. In group psychotherapy, dropping out and irregular attendance can have negative repercussions on the other members and disrupt group cohesion, itself an essential factor in treatment continuance [6]. Members who attend regularly do not want to repeat themselves and may experience feelings of insecurity, anxiety or anger [7]. There are several reasons why a patient might initiate premature termination such as anxiety about self-disclosing or disagreement with therapist about which problems should be addressed, but the specific factors remain largely unknown [7]. Among socio demographic variables, minority racial status, less-educated and lower income groups have been related with dropout rates with a moderate effect size [1]. However, when other related variables are taken into account such as patient expectations, the relationship between socioeconomic status and dropout is diminished [8]. Another issue in the premature termination literature concerns the various definitions of dropout, each method of operationalizing it having its own strengths and weaknesses. Studies that defined dropout in terms of failure to attend a scheduled session reported lower dropout rates than did studies defining dropout based on therapist judgment or the number of sessions attended [1]. Additionally, factors related to dropout depend on clinical contexts and diagnoses.

Individuals with Borderline Personality Disorder (BPD) exhibit high dropout rates [9] and utilize health care services more frequently than any other psychiatric group [10]. The reported dropout rates in BPD patients range from 16% to 67% depending on treatment modalities and definitions [11-15]. A younger age has been found to be a consistent predictor of dropout from outpatient psychotherapy in BPD patients [13,16]. Also, higher [17] or lower baseline psychopathology [11], higher levels of anger and hostility [11-13,18] behavioral impulsiveness [18] and lower interpersonal distress [16] were all found as predictors of dropout rates in BPD individuals. Examining the relationship between specific personality disorder criteria and individual psychotherapy attendance, Hilsenroth et al. [19] found that criteria associated with low levels of interpersonal distress (e.g. lack of remorse in antisocial personality disorder) were associated with fewer sessions whereas criteria reflecting greater interpersonal distress (e.g. frantic effort to avoid real or imagined abandonment in BPD) were associated with greater attendance.

There is a paucity of studies focusing on the relationship between premature termination psychotherapy and aspects of personality measured by projective techniques such as the Rorschach and the Thematic Apperception Test [20-23]. And yet, these aspects of personality are particularly important from clinical and research perspectives for at least two reasons. First, these measures of personality can reveal elements directly relevant for psychodynamic psychotherapy planning. Secondly, these measures of personality variables such as object relations might predict different behaviors associated with psychotherapy termination from those predicted by self-reported measures such as interpersonal distress. Although promising predictors of dropout rates, current findings on the quality of object relations remain inconsistent.

Object relations, which refer to the cognitive-affective representations underlying interpersonal relationships, are known to be related to a person’s capacity to establish stable relationships in different contexts, including in the therapeutic context [24]. Given the importance of relational issues in treatment models for BPD, the object relations model offers a rich conceptual framework for identifying the factors associated with termination of psychotherapy. Ackerman et al. [23] used a measure of object relations, the Social Cognition and Object Relations Scale [SCORS; 25], applied to TAT narratives, to predict the number of individual psychotherapy sessions that would be attended by patients with a personality disorder. Their results indicated that a low score on the affective quality of representations and a high score on the emotional investment in relationships scales of the SCORS were predictive of the number of sessions attended by patients. The authors concluded that patients with poor object relations, which are the ones who perceive and experience interpersonal relationships as malevolent and abusive, but who also manifest a capacity/desire to invest themselves more deeply and more positively in relationships, were more likely to stay in psychotherapy. In contrast to patients with more mature object relations and who are better able to meet their relational needs outside of therapy, these patients would have a greater need to contact with an understanding individual (e.g. therapist) and improve their interpersonal functioning, would be more motivated to work toward this end, and therefore would attend more psychotherapy sessions. These interesting findings were partially supported by a recent study examining the relationship between the quality of object relations as measured by the SCORS and individual and group psychotherapy sessions [20]. Results showed that malevolent affective expectations of interpersonal relationships, but not the capacity to emotionally invest in others, were predictive of individual psychotherapy attendance in patients suffering from severe mental illness. There were no object relations dimensions related to group therapy attendance. Consequently, the question remains as to whether findings obtained in individual psychotherapy can be applied in a group therapy context. Moreover, given the particular aspects of relational issues in group psychotherapy as opposed to individual psychotherapy, it is possible that these two object relations dimensions (affective quality of representations and emotional investment in relationships) are distinct but not independent variables in their associations with various aspects of group therapy. In such a case, they would end up being in conflict in the prediction of group therapy termination. For example, a multicollinearity relationship between these variables could lead to a non-significant result even though both of them are meaningful predictors of attendance rates. In this case, a more suitable approach would be to examine their respective associations independently from each other.

The aim of the present study was to examine the relationship between the quality of object relations and the group psychotherapy attendance among patients enrolled in a specialized program for borderline personality disorder. More specifically, affective quality of representations and emotional investment in relationships dimensions of object relations were examined in relation to attendance rates while controlling for the effect of age of the participants. Based on previous studies conducted on individual psychotherapy, we hypothesize that: 1) scores on the affective quality of representations scale of the SCORS will be negatively correlated with the rate of attendance at group psychotherapy, and 2) scores on emotional investment in relationships scale will be positively associated with the rate of attendance.

Participants

The sample of the present study consisted of 41 outpatients who were diagnosed with borderline personality disorder. All patients were diagnosed by a senior psychiatrist based on DSM-IV criteria. Also, in their initial admission assessment by a senior psychologist, they all obtained a score of 8/10 or more on the French version of the Diagnostic Interview for Borderline – Revised (DIB-R) [26]. The average age of the patients was 37.3 years (SD=10.3, range=20-62). Eighty-eight percent were women. The sample consisted of two cohorts of patients who participated in a group psychotherapy program at the Douglas Mental Health Institute (DMHI) in Montreal, Canada. The first cohort (n=26) was constituted based on the DMHI’s patient database. All patients participated in a group psychotherapy program between September 2003 and June 2004, and their records included the TAT protocols. They signed a written consent form, at the time of their admission assessment, authorizing the use of the collected data for research purposes. The second cohort (n=15) was made up of patients who participated in a group psychotherapy program in 2007–2008. All the patients in this group were informed of the study by their primary therapist. Only those who volunteered were included in the study.

Group therapy and therapists

Each patient received a psychodynamically-oriented group therapy. In this therapy, the primary objective is to enhance the patients’ insight about anxiety related to separation-individuation issue [27]. Groups consisted of five to eight persons maximum. Sessions were held weekly and the treatment contract stipulated that patients were expected to attend every week. Patients could join the group during the course of the year. However, all the participants in this study had been attending the group from the start of the psychotherapy. Two therapists were present at each group session. In the groups of this study, the therapists were all women; they were three psychologists and one social worker with more than ten years of experience working with patients with borderline personality disorder. At the DMHI, group psychotherapy programs ran between mid-September and mid- June of the following year.

Social Cognition and Object Relations Scale (SCORS)

The SCORS is a scale that measures the quality of object relations on eight dimensions [25]. The study focused specifically on two dimensions, affective quality of representations (AFF) and emotional investment in relationships (EIR). The AFF scale measures the extent to which a subject expects malevolence and pain in contrast to benevolent and enriching relationships from others. The EIR scale measures the extent to which a subject tends to be preoccupied with his own needs, turbulent relationships, and limited or non-existent relationships, as opposed to engaging in interdependent relationships and emotional intimacy with respect and appreciation for others. Each scale from the SCORS is scored from 1 to 7 using the Thematic Apperception Test [28]. A mean score is obtained from the stories’ scores on each scale, the lowest score corresponding to the most maladaptive response and the highest, the most adaptive. Several studies demonstrated good to excellent interrater reliability and convergent validity with psychiatric, occupational and interpersonal functioning, and with personality features of many populations including BPD [29-35]. In the present study, each scale was scored using a five-card TAT protocol (cards 1, 2, 3BM, 4 and 13MF) following earlier studies [23,34,36].

Procedure

All the study protocols were first transcribed verbatim from audiorecordings. Clinical psychologists were responsible for the administration of the TAT and the scoring of the SCORS. Interrater reliability with a second rater was obtained on a random selected series of 45 stories (22%) with a two-way, mixed-model intraclass correlation coefficient (ICC). Consistency agreements for reliability scores were excellent on AFF (ICC=0.87, p<0.001) and EIR (ICC=0.90, p<001).

We utilized the number of group therapy sessions attended during the year to calculate an attendance rate for each patient.

Regarding the SCORS variables, the average scores were 2.59 (SD=0.42, range=2-3.4) and 1.84 (SD=0.46, range=1-3.4) for the AFF and EIR dimensions respectively. The average group therapy attendance rate was 0.77 (SD=0.16, range=0.33-1.0).

Partial correlation was used to explore the relationship between each SCORS variable and group therapy attendance rate, while controlling for age. Preliminary analyses were performed to ensure no violation of the assumptions of normality, linearity and homoscedasticity. There was a moderate positive partial correlation between EIR and group psychotherapy attendance rate (r=0.32, p<0.05), with higher levels of EIR associated with higher attendance rates. There was a trend to a significant positive partial correlation between AFF and group psychotherapy attendance rate (r=0.28, p<0.07), with higher levels of AFF associated with higher attendance rates. There was a large and positive correlation between the two SCORS variables (r=0.55, p<0.001). These results confirmed the hypothesis regarding EIR but showed an inverse relationship to what was expected for AFF.

The participants in this study who showed the greatest attendance in group psychotherapy were found to have a higher EIR score. Two possible hypotheses might explain this result. First, on the object relations level, it could be that patients with pathological EIR scores experienced the group differently than did those with healthier scores. Second, on the interpersonal level, patients with pathological EIR scores may have had a different impact on the group than did those with healthier scores.

With regard to the object relations aspect, results of the present study support previous data obtained in individual psychotherapy [20,23]. Ackerman et al. [23] explained this positive correlation between EIR and the attendance rate by suggesting that, for patients with higher EIR scores, the therapeutic relationship might provide an opportunity to satisfy desires for emotional investment that is not available to them elsewhere. In the context of group psychotherapy, this explanation could be even more far-reaching, given that the group offers many more possibilities for emotional investment and development of relational capacities than would individual psychotherapy. Patients with healthier EIR scores can turn to each other, to the therapists or to the group at large to satisfy their desires for a relationship. Conversely, the eminently relational situation of the group may not be suitable for patients who have no capacity/desire to invest emotionally in relationships. Because the group does not adequately respond to their specific personal needs, they would be less inclined to attend on a regular basis. This interpretation is in line with the conclusions of another study about the importance of matching the therapy to the particular needs and expectations of patients [37]. Indeed, it is possible that the needs and expectations of patients with a pathological EIR score are different from those of patients with a healthier score. Patients with low EIR may have difficulty seeing how the group situation could be helpful to them, if they are not conscious of any need for psychotherapy around issues related to emotional investment in relationships. The qualitative study by Hummelen et al. [38] explored the dynamics leading to dropout among women with borderline personality disorder. The authors interviewed both patients who had discontinued long-term group psychotherapy and their group therapists. The majority of the patients and therapists identified the inability to make use of the group as having led to the termination. These patients reported that they did not see the point of being in group psychotherapy. In fact, the way they described themselves (unable to open up in the group and difficulty being empathetic) echoed certain pathological elements of the EIR scale. Moreover, they tended to perceive the group as being responsible for their symptoms and interpersonal difficulties. Most patients also expressed conflicts regarding sharing time and attention in the group. In sum, we suggest that the capacity to make therapeutic use of the group is influenced by the capacity to emotionally invest in relationships. More research is needed to investigate this hypothesis.

At the interpersonal level, it is possible that patients with low EIR would provoke negative reactions among the other participants in group therapy, for example, by being more focused on themselves, using others for their own personal gratification, and being disinclined to show any interest in others. This could lead to their coming less often to group sessions, in order to avoid having such negative reactions directed at them. Future studies could explore the associations between EIR levels in group participants and other variables such as group cohesion.

With regard to affective quality of representations, a trend to a significant positive correlation between the AFF variable and group therapy attendance obtained in the present study was inconsistent with what was expected from the negative correlation obtained by Ackerman et al. [23] in individual psychotherapy. Taken together, these data suggest that the relationship between this dimension of object relations and therapy attendance varies according to the treatment modality (i.e., individual or group psychotherapy). In individual psychotherapy with a therapist who is able to regulate projective identifications as well as his/her own countertransference reactions, patients with poor object relations may find an opportunity to satisfy unmet relational needs or to repair unresolved developmental issues. This new corrective relational experience might improve treatment retention in individual psychotherapy among patients with poor object relations. In group psychotherapy, the situation might be different for patients with pathological AFF scores. Indeed, if patients are more likely to experience intense dysphoric emotions in group situations, then their tendency to greater absenteeism could be understood as an attempt to avoid experiencing negative affects. The qualitative results of Hummelen et al. [38] support this view. The patients with BPD who discontinued group psychotherapy reported intense negative feelings during the treatment, ranging from rage, contempt, powerlessness and guilt to strong anxiety. Most also reported a desire to escape the negative judgments of other members of the group. In sum, we suggest that patients with pathological malevolence affective expectations of interpersonal relationships can unknowingly reproduce their unconscious relational dynamics in such a way that the group situation becomes nearly intolerable and leads to more absenteeism.

The results of the present study should be interpreted with caution for three main reasons. First, only two dimensions of the quality of object relations were assessed while other dimensions could contribute to the prediction of group therapy attendance. Second, there was no control variables included in the analyses except for age while other sociodemographic and personality variables should be taken into account (e.g. therapists’ characteristics, history of psychotherapy treatment, motivation, participants’ perceptions of treatment, socioeconomic status, other mental disorders, etc.). Finally, other studies with larger samples are needed to replicate and validate our results with more powerful parametric statistics.

This study is the second one, after Fowler & DeFife’study [20], to examine the quality of object relations in relation to the rate of attendance in psychodynamic group therapy by patients with borderline personality disorder. Given the relevance of the predictors for psychodynamic therapy planning, more research is needed to identify their utility according to different treatment modalities and their clinical implications. For example, patients who are incapable of investing emotionally in relationships may be better candidates for individual psychotherapy, which would be more suitable to their needs and developmental issues. For its part, the affective quality of representations underscores the clinical importance of therapists paying particular attention during the process to any dysphoric emotions triggered in participants by the group situation, in order to reduce absenteeism and prevent termination.

We also wish to express our appreciation to Anne Henry for help with an earlier version of this manuscript and to Sacha Daelman with the inter-rater reliability assessment.