Introduction

Early psychotherapy dropout is a significant phenomenon, frequently encountered by therapists in various theoretical approaches, with rates ranging from 15 to 75% (^{Arnow et al., 2007}; ^{Bados, Balaguer, & Saldanã, 2007}). Among the factors associated with dropout, factors of the therapist-patient relationship stand out, especially the therapeutic alliance (TA). The meta-analysis by ^{Sharf, Primavera and Diener (2010}) revealed a relatively strong relationship between TA and dropout (d: 0.55), indicating that patients with weaker TAs are more likely to drop out of treatment early. The study of moderators suggests that the relationship between TA and dropout is stronger in individuals with less schooling, longer therapies and in the context of hospitalization. Establishing and maintaining a good alliance is, therefore, decisive to avoid treatment dropout and to create the conditions for therapeutic progress to occur (^{Horvath, Del Re, Flückiger, & Symonds, 2011}; ^{Krause, Altimir & Horvath, 2011}).

The concept of the TA originates from the consideration of Freud (1913) on the need for the patient to connect positively with the therapist so that the analytical work can develop. The terms therapeutic alliance and working alliance were coined, respectively, by Zetzel and Greenson, who indicated their character of being conscious, non-conflicting and differentiated from transference (^{Gomes, 2015}). However, the concept evolved and started to be conceived pantheorically, from the concept of ^{Bordin (1979}), as a relationship of conscious and purposeful mutual collaboration between patient and therapist in therapy, characterized by an agreement on aims, an assignment of tasks (agreed in the therapeutic contract) and the development of emotional ties. This pantheoric vision of the TA, with an emphasis on collaboration and consensus, influenced, although in a different and non-consensual way, most of the main current measures of the construct (^{Horvath et al., 2011}; ^{Krause et al., 2011}).

The TA is currently recognized as a dynamic process that varies in intensity, frequency and duration, depending on the patient's diagnosis and the type of theoretical approach employed. These oscillations, called ruptures, can happen during treatment and even during the same session (^{Barros, Altimir & Pérez, 2016}; ^{Safran, Muran, & Eubanks, 2011}). The ruptures are characterized by a deterioration in the TA, manifested by a lack of collaboration in tasks or objectives or a tension in the emotional bond and difficulty in negotiating aspects of the therapeutic relationship (^{Eubanks, Muran, & Safran, 2015}; ^{Safran & Muran, 2006}; ^{Safran, Muran, & Proskurov, 2009}). The patient can avoid the therapeutic work and/or the therapist or confront them directly. These movements of rupture of TA are inevitable in any psychotherapy (^{Eubanks et al., 2015}). However, when not repaired, ruptures can cause dropout from the treatment (^{Safran et al., 2011}).

There is ample evidence that weakened alliances are correlated with the patient's unilateral termination of the therapy (^{Doran, 2016}; ^{Safran et al., 2009}; ^{Safran, Israel, & Einstein, 2006}), while negative interactions (i.e., hostile and aggressive) between patient and therapist are associated with unfavorable outcomes (^{Coady, 1991}; ^{Samstag et al., 2008}; ^{Zilcha-Mano, Muran, Eubanks, Safran, & Winston, 2018}). According to the meta-analysis of ^{Safran et al. (2011)}, there is empirical evidence that individuals with personality disorders present greater intensity of ruptures at the start of therapy than those without personality disorders. Among the former, patients with Borderline Personality Disorders (BPD) present the highest rate of ruptures. This may explain the higher rates of treatment dropout among patients with this disorder (^{Kroger, Harbeck, Armbrust, & Kliem, 2013}; ^{Koons et al., 2001}; ^{Linehan et al., 2007}; ^{McMain et al., 2009}).

Patients with BPD have inflexible and long-lasting patterns of emotional and interpersonal difficulties (^{Benjamin, 1993}; ^{Leichsenring, Leibing, Kruse, New, & Leweke, 2011}a) that require from the therapist to manage central aspects of their pathology (^{Barnow et al., 2009}; ^{Kernberg, 2012}; ^{Lazarus, Cheavens, Festa, & Zachary Rosenthal, 2014}; ^{Skodol et al., 2002}). For example, affective instability and lack of integration of the self and significant others is manifested through feelings of chronic emptiness, contradictory and impoverished perceptions of themselves and others, hindering the empathic experience and the relationship with the therapist. For this reason, it is recommended that interventions with these patients focus on the development and maintenance of the TA (^{Bennett, Parry, & Ryle, 2006}; ^{Geremia, Benetti, Esswein, & Bittencourt, 2016}).

Because the alliance is so critical to the outcome, Safran and colleagues (^{Eubanks, Muran, & Safran, 2014}; ^{Muran et al., 2009}; ^{Safran & Muran, 1996}, ²⁰⁰⁰; ^{Safran et al., 2011}; ^{Safran et al., 2009}; ^{Safran et al., 2006}; ^{Safran, Muran, & Samstag, 1994}) created and tested a model to resolve ruptures in the alliance during therapy. Recently, in order to systematize the codification of ruptures and resolutions of the TA, ^{Eubanks et al. (2015)} developed the Rupture Resolution Rating System (3R’s). The system aims to code ruptures in the TA and resolution interventions, in segments of psychotherapy sessions.

^{Gersh et al. (2017}) conducted a study of 44 young people with BPD, aiming to explore the processes of rupture and resolution of the TA, through the 3RS. The study showed that ruptures occurred in 53% of the sessions and with the passage of treatment they tended to increase, with confrontations being more frequent. The ruptures that occurred at the beginning of treatment were associated with worse results. Conversely, greater resolution of the ruptures was associated with better results and could be opportunities for therapeutic growth.

Considering the relevance of the TA process for the adherence of patients with BPD to psychotherapy, this study aimed to identify the frequency and variations of the processes of rupture in the TA in an interrupted case of psychoanalytic psychotherapy, performed with a patient diagnosed with BPD. A further aim was to describe the characteristics of the overall therapeutic process.

Method

This was a systematic, idiographic, longitudinal and intensive case study. This type of study has similarities with traditional clinical case studies, however, it differs from these in that, among other aspects, it presents greater methodological rigor, such as the use of independent judges, analysis of recorded audio and video sessions and control over biases of the researcher (^{Edwards, 2007}; ^{Serralta, Nunes, & Eizirik, 2011}).

Results

Overall psychotherapy process

The therapy lasted 27 weeks. However, due to frequent absences between sessions, the patient attended only 15 of them. The patient was late for almost half of the sessions (46.7%). The mean duration of the 15 sessions was 34.25 minutes (SD= 11.30). Between the 1^st and the 4^th session, the mean duration was 29.25 minutes (SD= 6.07). Between the 5^th and 11^th sessions, the duration of psychotherapy increased, lasting for a mean of 41.87 minutes (SD= 8.98). The recording failed during session number 7 due to a lack of battery power in the camera.

The overall therapeutic process evaluated with the PQS showed that the therapy had few silences (item 12) and was permeated by significant material (item 88). The dyad discussed topics related to the patient's current life situation (item 69), as well as his aspirations (item 41), generally presenting a specific focus (item 23) on the discussion of the interpersonal (item 63) and loving relationships (item 64) of the patient. During the sessions, the patient felt safe and confident (item 44), was active (item 15) and initiated the subjects (item 25). He generally showed acceptance of the therapist's comments (item 42) and was able to understand the therapist's comments (item 5).

The therapist was self-confident (item 86) and communicated through a clear and coherent style (item 46). In her relationship with Carlos, she was responsive, involved (item 9) and showed tact (item 77). Her interventions were directed toward obtaining more information and elaboration (item 31) and to facilitating the patient's speech (item 3), giving attention to verbal aspects, without referring to non-verbal ones.

The identification of ruptures in figure 4 illustrates the number of rupture markers identified in each treatment session. Ruptures were identified in all the psychotherapy sessions. Over the 15 sessions, 100 rupture markers were found (a mean of 6.66 ruptures per session). Of these, 69% were withdrawal ruptures and 31% confrontation ruptures.

Figure 1: Number of rupture markers identified. 

As shown in Figure 1, the sessions with the highest occurrence of ruptures were 5, 6 and 9 (with 17, 13 and 17 ruptures, respectively). The ones with the lowest occurrence were 7, 12 and 14 with 3 ruptures each, and session 15, with only one. A bimodal frequency distribution was observed, in which two main peaks of increase in ruptures were identified (sessions 5 and 9). Regarding the type of rupture, with the exception of the 1^st, 4^th and last sessions, there was a predominance of withdrawal ruptures in all sessions.

The withdrawal ruptures with a higher mean frequency throughout the treatment were: avoidant storytelling and/or shifting topic (Fr= 11), deferential and appeasing (Fr= 9), and abstract communication (Fr= 6). The predominant confrontation ruptures were: patient rejects therapist intervention (Fr= 12), defends self against therapist (Fr= 3) and efforts to control/pressure therapist (Fr= 3). The frequencies of the ruptures are presented in Table 1.

Table 1: Mean occurrence of rupture subcategory markers identified in the 15 treatment sessions 

The intensity of the impact of the confrontation and withdrawal ruptures was examined. The intensity of the confrontation ruptures (M= 3.61; SD= 0.9) was slightly higher when compared to the impact of the withdrawal ruptures (M= 2.97; SD= 1.1). The most intense confrontation ruptures were: patient defends self against therapist (M= 3.67; SD= 0.5); efforts to control/pressure therapist (M= 3.67; SD= 1.12) and patient rejects therapist intervention (M= 3.11; SD= 1.0). The rupture of complaints/concerns about the parameters of therapy (M= 4.00) presented only one occurrence, and therefore, the 4 degree of impact refers only to this marker. Among the withdrawal ruptures, the markers with the greatest impact were denial (M= 3.33; SD= 1.4); abstract communication (M= 3.17; SD= 1.0) and deferential and appeasing (M= 3.09; SD=0.9). The other means of the impacts of the specific subcategories of the withdrawal and confrontation ruptures are presented in Table 2.

Table 2: Mean impact of each subcategory of the TA ruptures over the 15 treatment sessions 

The ruptures (of confrontation and withdrawal) showed some mean significance for the TA (M= 3.0; SD= 0.62). In all sessions, the ruptures had at least some impact on the TA (score of at least 3.0). In sessions 5, 6, 9, 13 and 14 there were high impact ruptures (score 5.0). The sessions that had the greatest mean impact on the TA were 12 (M= 3.7; SD= 1.10), 13 (M= 4.2; SD= 1.10) and 14 (M= 4.0; SD= 1.10).

During the treatment, 30 contributions of the therapist to causing or exacerbating the ruptures were identified. The sessions in which the therapist most frequently contributed to the occurrence of ruptures were 5 (n= 10), 6 (n= 6), 8 and 9 (n= 5 in each), although this conduct was also observed in sessions 7, 10 and 11. In contrast, there were no contributions by the therapist to the ruptures in sessions 1, 2, 3, 4, 12, 13, 14 and 15.

Discussion

Carlos' psychotherapy was understood as an unsuccessful therapy, as the patient unilaterally abandoned the treatment without the intended therapeutic gains having been achieved. The treatment lasted for 27 weeks. However, despite the weekly frequency contracted, it totaled only 15 sessions. Furthermore, the low number of sessions in relation to the treatment time was mainly due to the absences between sessions (there were 9 absences; of these, only one previously agreed with the therapist). The absences started in the 4^th session, just when, according to the literature, it is expected that the TA starts to develop more effectively (^{Gersh et al., 2017}).

When examining the sequence of attendance and absences in the sessions, it can be seen that, with the exception of the initial sessions, the process is practically all interspersed with absences. This possibly reflects the pair's difficulty in developing a sufficiently good TA at the beginning of the treatment (^{Horvath et al., 2011}). As the patient spontaneously sought psychotherapy, expressing a need for help and apparently agreed with the initial contractual arrangements, there is a clear indication of disagreement over the therapy tasks (^{Bordin, 1979}; ^{Krause, Altimir & Horvath, 2011}) that manifested itself in frequent absences and lateness.

Effective treatment models for BPD share, among other aspects, a focus on emotional experience, greater activity by the therapist and a focus on the therapeutic relationship (^{Weinberg, Ronningstam, Goldblatt, Schechter, & Maltsberger, 2011}). The overall description of the process with the PQS does not capture any of these elements, although there are items in the instrument that describe these characteristics. Furthermore, this description shows an “apparent collaboration” in the exploration of significant material in the sessions, related to the problems that led the patient to treatment. There was also a significant lack of discussion of non-verbal behaviors that could have been due to resistance and difficulties in the interaction.

The fact that the overall analysis with the PQS takes into account the macroprocess in psychotherapy and focuses on the broad observation of how the patient and therapist experience the therapeutic process should be considered. Just like a photograph, a scene is captured as a whole and a panoramic view of the process is obtained, with a low degree of resolution (^{Bucci, 2007}; ^{Cordioli & Grevet, 2018}). Therefore, in order to better understand the TA process, through the 3RS the aim was to address specific, detailed and immediate issues related to how the patient and therapist organized themselves during the psychotherapy session (the microprocess), observing the therapeutic scene with greater resolution to identify possible problems with the TA (^{Barcellos, Cardon, & Kieling, 2018}; ^{Bucci, 2007}).

The evaluation of the TA throughout the treatment, analyzed at a microscopic level, identified problems in the collaboration between the patient and therapist in the present case, already verified at the macro level due to the absences and lateness mentioned. Ruptures were identified in all the sessions, a high number of ruptures (n=100), compared to other studies conducted using the 3RS with patients with BPD (^{Gersh et al., 2017}). The ruptures were predominantly of avoidance, more frequent in the intermediate sessions (with the contribution of the therapist) and had some impact on the TA. Furthermore, it was observed that after the ruptures increased, in the intermediate sessions, there was a decrease in frequency. However, there was an increase in their impact on the TA in the sessions just before the end, in which the therapist contributed to the ruptures.

The difference between the frequencies of ruptures in the present study, when compared with some reports in the literature, is an interesting and unexpected finding. However, the pattern is consistent with the other indicators evidenced (absences and lateness). The fact that this was an interrupted (unsuccessful) psychotherapy with a BPD patient, supposedly more likely to demonstrate difficulties in establishing and maintaining the TA, cannot be disregarded (^{Boritz, Barnhart, Eubanks, & McMain, 2018}; ^{Safran et al., 2009}). The study by ^{Doran, Safran and Muran (2017}), which examined the relationship between the therapeutic process and the negotiation of the TA in 47 patients with different clinical psychopathologies (anxiety and depression) and with personality disorders (48.9%), also highlighted fluctuation in the occurrence of ruptures (confrontation and withdrawal) over time.

^{Bennett et al. (2006}) point out that a greater occurrence of ruptures of the TA and treatment dropout are common in patients with BPD. These patients have flaws in the process of representing the self and others (^{Caligor, Kernberg, & Clarkin, 2008}) and under emotional activation, they tend toward concrete, non-mentalized thinking (^{Bateman, Campbell, Luyten, & Fonagy, 2018}), showing impulsive behaviors related to flaws in emotional regulation, which lead to a rapid distortion of themselves and of the relationship with their therapists (Caligor et al., 2008; ^{Leichsenring et al., 2011}b). This distortion in the relationship with the therapist favors the occurrence and impact of ruptures in the TA (^{Spinhoven, Giesen-Bloo, van Dyck, Kooiman, & Arntz, 2007}).

During the treatment, Carlos used withdrawal ruptures more frequently: avoidant storytelling and/or shifting topic, deference and appeasement and abstract communication. This finding suggests that the patient had difficulty expressing his dissatisfaction or anguish directly. According to the literature, the frequency of withdrawal ruptures is significantly higher in patients that dropout from treatment (^{Boritz et al., 2018}). These ruptures, in comparison to those of confrontation, are more difficult to identify, since the behavior of the patient, in avoiding the therapeutic work or the therapist, is subtle, unclear or even obscured (^{Safran et al., 2011}). Withdrawal ruptures may be disguised as superficial compliance or engagement, for example (^{Boritz et al., 2018}).

Withdrawal behavior can therefore be seen as a strategy to regulate the intense and overwhelming emotion that is activated in the context of the therapeutic relationship (^{Bernecker, Levy, & Ellison, 2014}). The study by ^{Cash, Hardy, Kellett and Parry (2014}) highlighted that the resolution of the rupture did not occur when therapists discussed or explored the ruptures directly, but when they approached the rupture indirectly by changing their approach to exploring issues that were more important to the patient. In the present study, however, rupture resolution strategies were not specifically analyzed. Nevertheless, the results suggest that in the absence of explicit clues for the withdrawal ruptures, the therapist may have positioned herself more passively when faced with them, having difficulty identifying and/or resolving them. The therapist's contributions to the occurrence of the ruptures suggest that she seemed to ignore the rupture markers and was not aware of what was happening at the relationship level, underestimating the impact these ruptures in the therapeutic collaboration had on the process. One study indicates that the therapist's contribution to ruptures is predictive of treatment dropout (^{Eubanks et al., 2018}b). In addition, the literature highlights an apparent paradox in the work of resolving ruptures with BPD patients, as dealing with the rupture directly can be considered excessive and threatening for some individuals, although not addressing it directly can further exacerbate the rupture (^{Boritz et al., 2018}).Despite the predominance of withdrawal ruptures, the confrontation ruptures also deserve attention. In these, the patient mostly rejected the therapist's interventions, defended himself from her and attempted to control the therapeutic process or the therapist. Confrontation ruptures occurred in 13 of the 15 treatment sessions. The judges also noted that, in rejecting the interventions of the therapist, the patient demonstrated a debauched tone, used sarcastic comments and described aggressive behaviors. A study by ^{Gülüm, Soygüt and Safran, (2018}) identified sarcastic humor as a third category of rupture. Sarcasm is a subtle form of aggressive communication, which expresses the difficulty in verbalizing feelings about the therapy or therapist, leading to treatment dropout in later sessions. In the final phase of psychotherapy, the decrease in ruptures is associated with their resolution (^{Eubanks et al., 2018b}). In contrast, in the present case, the last sessions did not constitute a final phase, since in a supposedly long-term therapy the process would ideally have reached an intermediate stage (^{Luz, 2015}). Therefore, the decrease in the occurrence of ruptures observed in the last sessions (sessions 12, 13 and 14), associated with the increase in their significance in terms of the impact on the TA, seems to indicate a deterioration in the TA that may have led to the patient’s dropout from the therapy.

It should be highlighted that the sessions in which the ruptures had the greatest impact on the TA were those that preceded the last session, when the patient talked about his decision to interrupt the treatment. These sessions were also the ones with the least ruptures, and in them there were no ruptures that showed expressions of self-criticism and/or hopelessness, complaints/concerns about the therapist, about the therapy activities or the treatment progress. It appears, therefore, that numerous minor ruptures may precede major ruptures. The hypothesis to be verified is that the non-resolution of less significant ruptures leads to an increase in the worsening of the TA, culminating in dropout. The ruptures can be comprehended as enactments,, a concept that highlights the role of the unconscious in the relationship between patient and therapist (^{Safran & Kraus, 2014}; ^{Safran & Muran, 2000}). For ^{Safran and Kraus (2014)}, ruptures in the TA are essentially unconscious movements between the pair. The therapist is involuntarily involved in the patient's functioning, thus re-enacting a form of dysfunctional relationship that is characteristic of the patient. These processes, when they do not become conscious and the targets of the therapeutic work, obstruct the development of a good therapeutic process. On the other hand, the therapist's awareness of the continuous fluctuations in the quality of the TA can present valuable opportunities to mobilize the process of change in the patient.

The initial or opening phase of treatment may be longer in therapy with patients with BPD, in view of the need to explore the ruptures that may interfere with the establishment and maintenance of the TA. Furthermore, the need to cultivate the TA with the main aim of the treatment becomes important, with the adaptation of psychotherapy to the patient's characteristics and constant monitoring of the fluctuations in the TA. The therapist's flexible and empathetic attitude is essential to deal with the ruptures, both those of withdrawal and confrontation.

The need to be cautious in the interpretation of the data presented is highlighted, in view of the limitations of this exploratory study. The coding of the ruptures in the TA was performed through the 3RS manual, which does not have categories for the assessment and codification of ruptures caused by the therapist. The evaluation of the microprocess measure involved only the patient's perspective. Although the judges received training and obtained substantial agreement, it is not known to what extent the coding is consistent with the coding of the gold standard 3RS.In the item “the therapist's contribution to the ruptures”, only the frequency at which the therapist exacerbated or contributed to them was identified. The assessment of the TA resolutions, which was not the target of this study, could better clarify the oscillatory behavior of the TA throughout treatment.

In future studies, it is suggested to include the coding of strategies that therapists can use to effectively identify and resolve TA ruptures. In addition, the ruptures and resolutions in the TA could be examined in multiple cases, with different psychopathologies and outcomes, with the performance of studies that assess the differences in the nature and predictive value of early and subsequent ruptures.

Conclusion

This study puts forward empirically supported hypotheses regarding the early dropout from therapy by a borderline patient. It reveals a process permeated by many ruptures from the beginning, which increased, first in number and then, in intensity, culminating in the session in which the patient communicated that he would dropout. The therapist contributed to the occurrence or increase in the ruptures. Her actions and interventions focused on the patient's problems, but not on the therapeutic relationship. Taken together, these factors seem to have been related to the dropout. However, considering the limitations of the purely descriptive design and the absence of specific analysis at the microprocessual level of the rupture resolution strategies, this hypothesis was not tested.

It should be emphasized that in Brazil, to date, there are no empirical assessments of ruptures in the TA. The 3RS system was recently translated into Brazilian Portuguese by the team of the Research Laboratory in Psychotherapy and Psychopathology (LAEPSI). Therefore, this is a pioneering study in the national context for the microprocessual examination of the TA, constituting an exploratory study that secondarily aims to introduce studies on the subject.

Nevertheless, the study presents contributions for the clinical practice by highlighting for discussion and analysis the role of the therapeutic relationship in the process of change and in the outcome of psychotherapy. Contemporary psychoanalytic theory emphasizes the mutual influence between therapist and patient and values the therapist's authenticity, flexibility and spontaneity. In this context, the TA acquires a central role, not only being a support for the initial stage of the treatment. With patients with personality disorders, such as borderline patients, who typically have difficulties in developing the alliance, the focus on this difficulty seems to be fundamental to the process. Working through ruptures in the alliance (that is, anticipating, identifying and resolving them) can provide patients with opportunities to learn to negotiate the tension between their own difficulties and the need for help versus the difficulties and needs of the therapeutic relationship, in a new and more constructive way.

Considering the empirical evidence indicating that the TA is a critical ingredient for change in various forms of therapy, it seems important to develop a well-articulated body of research with relevant knowledge about TA microprocessual measures. This study is the first developed in the country to use an approach based on external observers to microprocess the assessment of ruptures in the TA, and to explore how these processes unfold over time in therapy. Further studies should contribute to the expansion of this type of investigation in the country. National and international investigations of other individual cases, as well as multiple and comparative case studies on these microprocesses, developed with patients with different diagnoses, outcomes and therapeutic approaches may, in the future, shed more light on the complex issue of the contribution of the therapeutic relationship and alliance for adherence, dropout, success and/or failure of the psychotherapy.