The beginning of the end: A comparison of treatment completers and early dropouts in trainee-provided time-limited Cognitive Behavioral Therapy

Abstract

About one in five clients drops out of treatment prematurely. Premature termination has been found to correlate with patient, therapist, and treatment factors, as well as complex interpersonal processes, including ruptures in the therapeutic alliance. This study examines the therapeutic alliance using a qualitative approach to patient-, therapist-, and observer-based data. The sample includes five trainee therapists, each of whom worked with one patient who terminated after the first or second session, and one who completed a cognitive-behavioral therapy protocol. The session(s) preceding premature termination in the drop case and the corresponding session(s) in the completer case were examined. Rupture resolution process was prevalent in both groups, though confrontation ruptures seemed more prevalent with completers and withdrawal ruptures were more clinically impactful with dropouts. Therapist awareness of process and responsiveness or contribution to rupture were identified as potential factors contributing to patient retention.

Despite its well-documented effectiveness (Lambert, 2013), psychotherapy does not work for all patients. In addition to the large group of patients who complete therapy without experiencing clinically significant improvement in their presenting symptoms (35-40%) or even deteriorate over the course of treatment, (5-10%, see Lambert, 2013), many who start therapy drop out prior to its completion. Premature termination, or dropout, has been defined as the unilateral discontinuation of therapy by a client before their therapy goals are achieved (Garfield, 1994; Swift et al., 2009). Several operational definitions have been used to identify dropouts, such as quitting treatment unilaterally, prior to completing a given number of sessions, or prior to achieving clinically significant change. Most dropouts occur within the first few therapy sessions (Swift & Greenberg, 2012). Recent meta-analyses suggest that dropout rates range from 19.7% (Swift & Greenberg, 2012) to 21.9% (Swift et al., 2017).

In comparison to completers, patients who drop out of treatment display worse outcomes (Cahill et al., 2003; Lampropoulos, 2010), are less satisfied with their therapy (Björk et al., 2009; Kokotovic & Tracey, 1987), and are more likely to engage in treatment and drop out again (Reis & Brown, 1999). Accordingly, premature termination represents a significant obstacle to the implementation of effective treatments (Barrett et al., 2008; Garfield, 1994; Hatchett & Park, 2003; Swift et al., 2009).

Client sociodemographic characteristics such as self-identified race or ethnicity (Arnow et al., 2007; Austin & Wagner, 2010), age and education level (Swift & Greenberg, 2012) have been found to predict dropout. Higher dropout rates were also found among patients presenting with eating disorders, personality disorders (McMurran et al., 2010; Swift & Greenberg, 2012), and lower intrapsychic functioning (Rubin et al., 2018). Among therapist and treatment factors, therapists’ experience was found to be negatively associated with dropout, while therapists who conducted manualized treatments and/or participated in efficacy studies were less likely to experience dropouts (Swift & Greenberg, 2012).

The relationship between alliance and dropout has received increasing attention. Several studies found that patients who quit therapy tended to report negative exchanges with their therapists, disagreement on the tasks and goals, and the sense that therapy did not progress (Burton & Theriault, 2020; Hundt et al., 2020; Norcross & Wampold, 2011). In two studies, the absence of agreement between patients and therapists on therapy tasks and goals, conceptualized as a key component of the alliance (Bordin, 1979), was associated with dropout (Gabbay et al., 2003; Westmacott et al. 2010). Further efforts have also demonstrated that patient- and therapist-rated alliance were negatively correlated with dropout (Eubanks et al., 2018; Sharf et al., 2010; Roos & Werbart, 2013), and that the alliance mediated the relationship between other predictor variables and early dropout (Anderson et al., 2019). Furthermore, more confrontation ruptures, more instances of therapists contributing to ruptures, and less rupture resolution have been observed among dropout than completer CBT cases (Eubanks et al., 2019).

Aims of the present study

Based on recommendations to evaluate the session(s) preceding premature termination (Swift et al., 2017), some promising and innovative work has been conducted with Turkish students (Gülüm et al., 2018) and adolescents (O’Keefe et al., 2020), and we hope to contribute to this growing literature. The present study’s aim is to utilize an exploratory approach utilizing primarily qualitative analyses, as well as multiple perspectives to advance our understanding of the processes that take place immediately prior to dropout (Swift et al., 2017). It has been found that up to one-third of adults who begin psychotherapy do not return for a second visit (Simon & Ludman, 2010; Simon et al., 2012), highlighting a sub-sample of early dropouts and suggesting there may be differences among those who drop out early and those who drop out later in treatment. To add to the moderate body of literature that distinctively explores dropout after the first or second psychotherapy visit (e.g., Hamilton et al., Horvitz-Lennon, 2003; Simon et al., 2012), we focus on those who ended treatment after the first or second session.

Methods

Participants

Dyads included in the present study were drawn from the archives of an ongoing psychotherapy research program at a major metropolitan medical center in the United States. Therapists were all trainees in psychology, reporting 0 to 3 years of clinical experience at the time, who had consented to participate in the research as part of their clinical training. The Cognitive Behavioral Treatment (CBT) condition, from which these data were collected, was largely based on the Beck et al. (1990) adaptation of cognitive therapy to the treatment of personality disorders, incorporating Persons’ (1989) case formulation approach. The focus on the therapeutic relationship was consistent with standard approaches to CBT (see Persons, 1989); the standard practice included dealing with problems in the therapeutic relationship and eliciting feedback at the end of each session regarding potential problems in communication. Trainees’ supervision, provided by a Fellow of the Academy of Cognitive Therapy, consisted of a weekly 75-minute group supervision that focused on case-specific formulation, treatment planning, role-playing, and videotape analysis and feedback.

Self-report data from five cases in this study have been previously published (Zilcha-Mano et al., 2017; Zilcha-Mano et al., 2019, 2016a, 2018). Eubanks et al. (2019) used self-report data and session video of one of the cases in this study, but that session is not included in our dataset. As such, no prior publication has reported findings derived from the session videos analyzed here.

Dyad selection

Therapists who provided CBT were considered for inclusion in the study if they met both of the following criteria:

1. Early dropout: Therapist was assigned a patient who unilaterally ended treatment prematurely after session 1 or session 2. At the time of dropout, members of the research team made efforts to follow up with patients and their therapists. Patients who reported external factors (e.g., moving) as reasons for termination were not included in these analyses.

2. Completer: Therapist was assigned a patient who successfully completed treatment, operationalized as the completion of at least 27 out of 30 sessions. Due to the training nature of the research program (e.g., trainees finishing their internship), some dyads ended treatment before reaching 30 sessions.

We further filtered our selection by the availability of session video. This search resulted in the identification of five therapists, each of whom worked with one early dropout patient and one completer patient.

Demographic characteristics

Therapists ranged in age from 27 to 41 (M = 31.60, SD = 5.94). Patients ranged in age from 24 to 51 (M = 33.70, SD = 8.74). Additional demographic and diagnostic information is described in the Results.

Treatment

Patients in this sample consented to receive 30 sessions of weekly CBT that focused on symptom reduction and schema change and included strategies such as self-monitoring, cognitive restructuring, behavioral exercises, and experimentation (see Beck et al., 1990 and Persons, 1989). Patients paid a nominal per-session fee determined using an income-sensitive sliding scale. All sessions were video-recorded, and a post-session questionnaire was administered to both patient and therapist after each session.

Measures

Post-Session Questionnaire

Working Alliance Inventory-Short (WAI).

The Working Alliance Inventory (WAI: Horvath & Greenberg, 1989; Tracey & Kokotovic, 1989) is a self-report measure of the therapeutic alliance along the three dimensions (goals, tasks, and bond) conceptualized by Bordin (1979). The WAI has demonstrated sound psychometric qualities and is widely used in psychotherapy research (see Flückiger, et al., 2018). Items were rated on a 7-point Likert scale, ranging from “Never” (1) to “Always” (7).

Session Evaluation Questionnaire (SEQ).

The SEQ is a measure of session impact used in studies of psychotherapy and supervision process. The SEQ Form 4 (Stiles et al., 1994) includes 12 items in a 7-point bipolar adjective format, evaluating two dimensions of impact: depth (a session’s perceived power and value) and smoothness (a session’s comfort, relaxation, and pleasantness).

Direct Report Measure.

The post-session questionnaire also included three questions regarding the presence of ruptures and their resolution scaled on a 5-point Likert scale with an open-ended option to describe the experience. To assess for rupture presence: Did you experience any tension or problem, any misunderstanding, conflict or disagreement, in your relationship with your [therapist/patient] during the session? To assess for rupture intensity: If yes, please rate how tense or upset you felt about the problem during the session. To obtain a description of the rupture: Please describe the problem. Finally, to assess for rupture resolution: To what degree to you feel this problem was resolved by the end of the session? For further description of this measure, see Muran, et al. (2009).

Session Progress Note.

Each therapist also completed a progress note after each session. There was no standardized prompt, and the data were included in analyses in order to provide supplementary narrative-based data.

Observer Measures

Rupture Resolution Rating Scale (3RS).

The Rupture Resolution Rating System (3RS: Eubanks et al., 2015) is an observer-based measure for identifying alliance rupture markers and resolution strategies in five-minute intervals. The measure began as an adaptation of a coding system by Harper (1989a, 1989b) and differentiates between two types of rupture: confrontation, in which a patient moves against the therapist or the work of therapy (e.g., complaints/concerns about progress in therapy) and withdrawal, in which a patient moves away from the therapist or the work of therapy (e.g., avoidant storytelling). Within each rupture category, 3RS identifies seven rupture markers defined by observable patient behaviors. The 3RS also identifies 10 rupture resolution strategies defined by observable therapist behaviors. Using five-point Likert scales ranging from “no significance” to “high significance,” coders rate the extent to which each marker and strategy had a clinically meaningful impact on the alliance over the course of the session, as well as the overall impact of withdrawal markers and of confrontation markers. Coders also rate the extent to which ruptures were resolved over the course of the session, and the extent to which the therapist caused and/or exacerbated ruptures. 3RS has been applied to different populations in various setting and multiple types of psychotherapy, including CBT (Coutinho et al., 2014; Eubanks et al., 2019). The 3RS aims to be transtheoretical and, as such, coders, who had been previously trained by the eighth author to a reliability of ICC (1,2) > .70, were trained to use the measure across theoretical orientation.

A team of four graduate students (Authors E.H., J.K., J.T.W., and G.T.W.) completed the 3RS coding for this study. Each session was coded by two randomly paired coders. Coders independently rated videos, and codes were averaged. In the event that codes diverged by 2 or more points on the five-point Likert scale, coders met to discuss and improve future coding. All coders were blinded to termination status (dropout vs. completer case), as well as session number (1 or 2).

Coders’ Process Notes.

Coders were asked to reflect on the sessions they coded and write a brief narrative commenting on the dyadic process and any content they found striking or believed was not captured via 3RS. Unstructured observation has been argued to be a potentially important research method in qualitative research and alongside more structured methodology (see Mulhall, 2003).

Overview of Analysis

Within each therapist, we focused on the session(s) available for the prematurely terminated case (session 1 or sessions 1 and 2), and the corresponding sessions for the completer cases. In total, 16 sessions were examined.

Five authors who were not part of 3RS coding (A.A.U., L.M.L., S.B., J.C.M., and C.F.E) employed a modified form of the Sort and Sift, Think and Shift method (Maietta et al., in press) to analyze the data. This qualitative approach allows for fluid movement between the “diving in” and “stepping back” phases of data analysis and has been used in various fields of research, including nursing and behavioral health (e.g., Bush et al., 2019; Fryer et al., 2015).

Data Inventory.

The analysis began with data inventory in which all available data – treatment status, session numbers, participant demographics, WAI and SEQ scores, therapists’ progress notes and direct reports, 3RS scores, and coder’s process notes were collected. The qualitative team members began by noting datapoints within the qualitative data that added to the description of the dyad and/or process. For example, one coder described patient C1 (described below) as “presenting for treatment for […??? No reason, because everything is actually ok?]” Such a statement is noteworthy because it begins to paint a picture of a patient who consistently says that everything is okay and potentially leaves the observer confused.

Diagramming.

The collected data were presented visually using diagrams and summary tables. Each therapist had their own page with a column for each of the two patients, showing available scores and highlighted quotations. Miles and colleagues (2018) refer to this process as “displaying the data” with the goal of providing a convenient, contained, and holistic picture of the story.

Memoing.

Memoing in qualitative research serves to assist the researcher in making conceptual leaps from raw data to abstractions that explain research phenomena in the context in which it is examined (Birks et al., 2008). Guided by the questions of “What have I learned from this data?” and “How do I make sense of it?,” each team member participated in the memoing process and provided written reflections to facilitate discussion. The memos served to syndicate what we know from the data and what we know from our understanding of the literature, as well as formulate questions that may be worth exploring. For example, one team member memoed, “Overall, very high rupture exacerbation across the board: what’s up with that?”

Episode Profiles.

Maietta and colleagues (in press) state that the goal of an episode profile is “to tell a holistic, vertical story of each…qualitative data collection episode. Across a project, these accessible and representative stories of each case serve as diagnostic and comparative tools that demonstrate…the essence of each episode” (p. 12). Team members’ memos were added to the inventory in order to facilitate the creation of profiles, which are presented below, under Results.

Bridging and Threading.

Maietta et al. define bridging as “creating a bridge between two or more ideas or themes” and threading as “discovering themes in the dataset that weave component ideas together” (p. 4). This process was done through team meetings in which memos and data were reviewed and further “mined” to identify recurring themes and subthemes. The questions that stemmed from memoing were also reviewed. In response to the aforementioned question of therapist exacerbation, for example, the team had a discussion about therapist exacerbation of rupture process in this sample. See Discussion for the themes and subthemes that emerged through this process.

Results

Tables 1–5 summarize data from each of the therapists’ two cases. With regard to 3RS scores, we focus on markers that received at least a mean clinical impact rating of 2.5. As the anchor for a rating of 2 on the 3RS significance items is “Rupture markers/resolution strategies have a minor impact on the alliance,” a mean rating of 2.5 suggests that at least one of the coders rated the marker as having more than a minor impact (Eubanks et al., 2019).

Table 1.

Therapist A

Process Measure	Pt A1 (D)		Pt A2 (T)
	P-Rated	T-Rated	P-Rated	T-Rated
WAI Total
Session 1	6.92	4.83	5.92	4.00
SEQ/Smoothness
Session 1	6.60	4.00	6.60	2.60
SEQ/Depth
Session 1	4.20	3.00	4.40	2.60
3RS	Session 1 (M)		Session 1 (M)
Total Withdrawal	3.0		3.0
Content/affect split	4.0		N/A
Deferential and appeasing	3.5		3.0
Minimal response	3.0		N/A
Abstract communication	2.5		N/A
Self-critical / hopeless			3.5
Total Confrontation	2.0		3.0
Defends self	2.5		3.5
Rejects intervention	N/A		2.5
Resolution Strategies
Illustrate task/provide rationale	2.5		2.5
Degree to which ruptures were resolved	2.0		2.5
Degree to which therapist caused or exacerbated ruptures	2.5		2.0

Note. D = Drop; T = Term. WAI = Working Alliance Inventory. SEQ = Session Evaluation Questionnaire. M = Mean of 3RS clinical impact (significance) scores; only markers rated 2.5 or higher are included. N/A = Not Applicable.

Table 5.

Therapist E

Process Measure	Pt E1 (D)		Pt E2 (T)
	P-Rated	T-Rated	P-Rated	T-Rated
WAI Total
Session 1	3.83	4.83	4.50	4.33
Session 2	4.75	5.42	5.08	4.33
SEQ/Smoothness
Session 1	2.80	5.00	5.60	3.60
Session 2	5.00	5.00	4.40	4.80
SEQ/Depth
Session 1	4.20	3.80	4.00	3.80
Session 2	4.80	5.00	4.20	4.80
3RS	Session 1 (M)	Session 2 (M)	Session 1 (M)	Session 2 (M)
Total Withdrawal	3.0	2.0	2.5	4.0
Content/affect split	3.5	2.5	3.5	3.5
Avoidant storytelling / shift	N/A	N/A	N/A	3.5
Abstract communication	4.5	N/A	3.0	N/A
Minimal response	N/A	N/A	N/A	4.5
Total Confrontation	2.0	1.5	1.0	2.0
Rejects intervention	2.5	N/A	N/A	2.5
Defends self	3.0	N/A	N/A	2.5
Complaint of activity	N/A	N/A	N/A	2.5
Resolution Strategies
Illustrate task/provide rationale	N/A	2.5	N/A	3.0
Degree to which ruptures were resolved	1.5	3.0	3.0	2.0
Degree to which therapist caused or exacerbated ruptures	4.0	2.0	3.0	4.5

Note. D = Drop; T = Term. WAI = Working Alliance Inventory. SEQ = Session Evaluation Questionnaire. M = Mean of 3RS clinical impact (significance) scores; only markers rated 2.5 or higher are included. N/A = Not Applicable.

Therapist A

The therapist was 27-year-old female Master’s level clinician identifying as Hispanic/Latinx and Agnostic. She reported three years of clinical experience and identified with an integrative/eclectic orientation.

Patient A1 (Drop)

The patient was a 24-year-old, single, college-educated and employed male identifying as Asian or Pacific Islander and Catholic. At intake, he was diagnosed with Major Depression, recurrent (Axis I) and Personality Disorder Not Otherwise Specified (Axis II). The patient dropped out after the first session.

Session Summary.

The patient rated the session high in alliance and session smoothness, with a lower score in session depth, and did not report any problems in the therapeutic relationship (see Table 1). Relative to the patient, the therapist rated the session lower across all three measures, noting in the direct report that the patient “seemed overly accommodating and preoccupied with a toothache,” which the therapist addressed by recommending an over-the-counter analgesic. In the progress note, the therapist recorded a brief history of the patient’s presenting problem and the goals for therapy, which included “1) Be more assertive and confident, 2) Learn how to cope with negative work environment, 3) Make more friends.” The therapist also assigned a homework task, asking the patient to buy a notebook and write down thoughts he has while at work.

3RS codes indicated a primarily deferential/appeasing patient presentation with withdrawal markers having “some impact on the alliance.” Coders’ process notes suggested that the patient “appears active and disclosing with therapist, and/but highly cooperative and compliant” with frequent apologies, and that the patient “seems preoccupied with intense toothache.” The only therapist resolution strategy that received a significant clinical rating was the therapist illustrating a task / rationale for therapy. The coders highlighted that the therapist “only nominally expresses concern over [the patient’s] pain or seems to explore it with him.” Coders rated the extent to which ruptures were resolved in this session as “below average,” and they rated the therapist as “maybe” and “somewhat” contributing to the ruptures in session.

Patient A2 (Completer)

The patient was a 35-year-old, single, college educated and employed female identifying as Asian or Pacific Islander and Jewish. At intake, she was diagnosed with Major Depression, recurrent (Axis I) and Obsessive-Compulsive Personality Disorder (Axis II). The dyad completed 29 sessions, and for the purposes of this study, the first session was analyzed.

Session Summary.

The patient rated the session higher than the therapist in alliance and session smoothness, with a lower score in session depth, not reporting any problems in the relationship (see Table 1). The therapist, on the other hand, noted that the patient “did not like the structured aspects of CBT and just wanted to talk… she was getting annoyed at the agenda and explanation of CBT.” In the progress note, the therapist recorded a brief history of patient’s presenting problem and therapy goals, which included “1) Achieving better interpersonal relationships, 2) Gaining higher self-esteem, 3) Identifying and coping with emotions.”

3RS codes displayed a complex patient presentation, with both withdrawal and confrontation markers having “some impact on the alliance.” Coders’ notes suggested that the patient was “very difficult” and “had an assertive style of communicating that might be experienced as domineering.” In this session, the therapist was slightly more active, illustrating a task / rationale for therapy and also redirecting when the patient changed topics. The coders added that the therapist “allowed space for [patient’s communication style] … was working hard to remain responsive and flexible… adjusting the focus of the session to build rapport rather than sticking rigidly to the CBT protocol.” Coders rated the extent to which ruptures were resolved in this session between “below average” and “average,” and they rated the therapist as “maybe” contributing to the ruptures in session.

Analysis

An important distinction between these two patients is their different interpersonal styles (“compliant” vs “domineering, as derived from the coders’ process notes). While the therapist seemed to be aware of Patient A1’s accommodating style (as evidenced by the progress note) it is possible that she did not feel well-equipped to address it. Patient A2, on the other hand, was more explicit in her concerns (evidenced by both 3RS scores and coders’ process notes), which allowed the therapist not only to address them via resolution strategies but also to focus on alliance-building, which may have created a stronger foundation for ongoing treatment.

Therapist B

The therapist was a 28-year-old female Master’s level clinician identifying as White (not Hispanic) and Jewish. She reported no prior clinical experience and identified with a dynamic/analytic orientation.

Patient B1 (Drop)

The patient was a 51-year-old, single, college-educated and employed female identifying as African-American (not Hispanic) and Buddhist. At intake, she was diagnosed with Other Problem Related to Employment (Axis IV). The patient dropped out after the first session. Neither therapist nor patient completed post-session questionnaires, thus we could only analyze the observer perspective.

Session Summary.

See table 2. 3RS codes displayed a primarily abstract and deferential patient presentation with withdrawal markers having “some impact on the alliance.” One coder’s process note suggested that the patient was “interpersonally guarded and slick in her communication style, which did not leave much space for real connection or exploration… [Patient’s] narrative in session is superficially sunny, intellectualized, distanced, and anecdotal.” Another coder noted that the patient, a social worker, “immediately displays expertise, experience, and familiarity with the field of therapy / mental health.” No therapist resolution strategies were identified, with coders describing the session as “pervaded by an overall tone of imbalance in expertise/experience between patient and therapist” where the therapist “did not explicitly draw attention or seemingly specifically intervene in response to any interpersonal process.” Coders rated the extent to which ruptures were resolved in this session between “poor” and “below average,” and they rated the therapist as “maybe” and “somewhat” contributing to the ruptures in session.

Table 2.

Therapist B

Process Measure	Pt B1 (D)		Pt B2 (T)
	P-Rated	T-Rated	P-Rated	T-Rated
WAI Total
Session 1	**	**	6.92	5.75
SEQ/Smoothness
Session 1	**	**	6.00	4.00
SEQ/Depth
Session 1	**	**	6.60	4.80
3RS	Session 1 (M)		Session 1 (M)
Total Withdrawal	3.5		4.0
Content/affect split	3.0		N/A
Deferential and appeasing	3.5		5.0
Abstract communication	3.5		N/A
Avoidant storytelling / shift	3.0		N/A
Minimal response	N/A		2.5
Total Confrontation	1.0		1.5
Complaint of parameter	N/A		2.5
Resolution Strategies
N/A
Degree to which ruptures were resolved	1.5		1.0
Degree to which therapist caused or exacerbated ruptures	2.5		4.0

Note. D = Drop; T = Term. WAI = Working Alliance Inventory. SEQ = Session Evaluation Questionnaire. M = Mean of 3RS clinical impact (significance) scores; only markers rated 2.5 or higher are included. N/A = Not Applicable.

^**Missing data

Patient B2 (Completer)

The patient was a 28-year-old, married, college educated and employed male identifying as White (not Hispanic) and areligious. At intake, he was diagnosed with Obsessive Compulsive Disorder (Axis I) and Depressive Personality Disorder (Axis II). The dyad completed 30 sessions, and for the purposes of this study, the first session was analyzed.

Session Summary.

The patient rated the session higher than the therapist in alliance, session depth and smoothness (see Table 2). In the direct report assessing problems in the relationship, he listed symptoms (i.e., “my obsessive-compulsive tendencies and my generalized anxiety with regards to my daughter”) but noted that “putting goals and techniques for addressing my problems helped me to feel hopeful that I can make progress in the future.” The therapist also noted that the patient “was very eager and encouraging – sometimes got the sense that he was overly accommodating to me.” In the progress note, the therapist recorded a brief history of the patient’s presenting problem and the assigned homework. She also noted that the patient “is very motivated to do the work and seemed grateful for the opportunity to talk about his symptoms.”

3RS codes displayed a primarily deferential/appeasing patient presentation with withdrawal markers having “moderate impact on the alliance.” Coders’ notes suggested that the patient was “quite deferential and eager… was withdrawing from anything going on within the dyad… [The therapist] also pushed at times, which left me with a sense of a real power differential.” Here, too, the therapist did not engage in any resolution strategies, with coders reporting that “the amount of anxiety coming from both the patient and therapist was incredibly uncomfortable.” Coders rated the extent to which ruptures were resolved in this session as “poor,” and they rated the therapist as “moderately” contributing to the ruptures in session. Of note, one coder highlighted that “there’s a chance such an anxious patient might feel comfortable with such an anxious therapist.”

Analysis

Both dyads were described by the coders as having palpable power differentials, with patients somewhat withdrawing from the process of therapy, though in different ways. The therapist, who, according to the coders’ process notes, seemed anxious herself, was relatively passive in her employment of resolution strategies, which suggests the importance of dyadic “fit”: as one of the coders noted, Patient B2 may have felt more comfortable with this style; he also appeared motivated for treatment (as noted by the therapist in the progress note), which could have served as a protective factor against dropout. Patient B1, as noted by the coders, seemed to struggle to fully engage in therapy in the context of her own clinical background and may not have found as much value in the therapist’s interpersonal style. It is notable that this patient did not complete the post-session questionnaire, which we presume could have provided valuable information about the patient’s experience.

Therapist C

The therapist was a 41-year-old female Master’s level clinician identifying as White (not Hispanic) and areligious. She reported one year of clinical experience and being undecided regarding clinical orientation.

Patient C1 (Drop)

The patient was a 43-year-old, single, high school-educated and employed female identifying as White (not Hispanic) and Christian. At intake, she was diagnosed with Personality Disorder Not Otherwise Specified (Axis II). The patient dropped out after the second session.

Session Summary.

Across both sessions, the patient rated the session higher than the therapist in overall alliance, session smoothness, and depth (see Table 3). Neither patient nor therapist reported any problems in the relationship.

Table 3.

Therapist C

Process Measure	Pt C1 (D)		Pt C2 (T)
	P-Rated	T-Rated	P-Rated	T-Rated
WAI Total
Session 1	6.42	4.58	5.75	5.00
Session 2	6.42	5.00	4.83	5.25
SEQ/Smoothness
Session 1	7.00	3.80	3.60	4.60
Session 2	6.80	4.40	2.60	4.40
SEQ/Depth
Session 1	5.40	4.20	3.40	4.00
Session 2	5.60	5.20	3.60	4.60
3RS	Session 1 (M)	Session 2 (M)	Session 1 (M)	Session 2 (M)
Total Withdrawal	3.0	4.0	4.0	4.0
Content/affect split	N/A	2.5	5.0	4.0
Deferential and appeasing	N/A	N/A	4.0	N/A
Abstract communication	3.0	3.0	3.5	4.5
Denial	3.0	2.5	N/A	N/A
Avoidant storytelling / shift	3.5	4.5	N/A	N/A
Self-critical / hopeless	N/A	2.5	N/A	N/A
Minimal response	N/A	N/A	N/A	2.5
Total Confrontation	2.5	2.0	3.0	3.5
Control/pressure	2.5	2.5	3.0	4.0
Rejects intervention	3.5	2.5	3.0	N/A
Complaint of activity	N/A	N/A	3.0	N/A
Defends self	N/A	3.0	N/A	3.0
Resolution Strategies
Clarify misunderstanding	N/A	N/A	2.5	N/A
Change task / goal	N/A	N/A	3.0	N/A
Invite thoughts / feelings	N/A	N/A	2.5	N/A
Redirect	N/A	2.5	N/A	3.0
Disclose internal experience	N/A	N/A	N/A	2.5
Degree to which ruptures were resolved	1.5	1.0	2.0	1.0
Degree to which therapist caused or exacerbated ruptures	2.0	3.0	2.5	3.0

Note. D = Drop; T = Term. WAI = Working Alliance Inventory. SEQ = Session Evaluation Questionnaire. M = Mean of 3RS clinical impact (significance) scores; only markers rated 2.5 or higher are included. N/A = Not Applicable.

3RS codes displayed a more complex patient presentation in the first session, with both withdrawal and confrontation markers having “some impact on the alliance” in the first session, and withdrawal markers having a “moderate impact” in the second. Coders’ notes suggested that the patient was “avoidant, intellectual, and dismissive… Despite [patient’s] presentation as affiliative and pleasant, she iteratively concludes each narrative in session as something that she already understands.” Redirecting was identified as the only impactful therapist resolution strategy (in the 2^nd session). One coder, who happened to code both sessions, reported that the therapist “tried in session 1 but… completely checked out for session 2.” Coders rated the extent to which ruptures were resolved in these sessions between “poor” and “below average,” and they rated the therapist as “maybe” contributing to the ruptures in the first session and “somewhat” contributing to ruptures in the second.

Patient C2 (Completer)

The patient was a 40-year-old, married, college educated and employed female identifying as White (not Hispanic) and Atheist. At intake, she was diagnosed with Major Depression, recurrent (Axis I). The dyad completed 28 sessions; the first two sessions were analyzed.

Session Summary.

The patient rated the first session higher than the therapist in alliance and rated both sessions lower than the therapist in session smoothness and depth (see Table 3). Neither patient nor therapist directly reported any problems in the relationship. The first session progress note reviewed the established goals for therapy, which included, “1) not quitting therapy and 2) working to lessen depressive and anxious symptoms.” The second session progress note briefly reviewed covered topics (i.e., “discussed key people in her life, including friends she’ll be spending the holidays with”).

3RS codes displayed a complex patient presentation in both sessions, with withdrawal markers having “moderate impact” on the alliance and confrontation markers having a “some-to-moderate impact.” Coders’ notes suggested that the patient was “remarkably uncomfortable, highly defended, and deeply ambivalent / rejecting / devaluating of therapy… keeping the therapist at an arm’s length.” The therapist was more active with this patient, employing several resolution strategies. The coders noted that the therapist was “highly didactic/explanatory” and also “clearly invites the patient to express negative sentiments / personal reactions.” Still, coders rated the extent to which ruptures were resolved as “below average” with the therapist “maybe” or “somewhat” contributing to ruptures in the first session. They rated the second session as “poor” in rupture resolution with the therapist “somewhat” contributing to ruptures.

Analysis

Both patients were described by coders as “ambivalent” about psychotherapy, and both presented with a mix of withdrawal and confrontation rupture markers. Based on 3RS scores, similar to patient A2, Patient C2 was more explicitly rejecting of therapy, giving the therapist a clear signal that a rupture was unfolding and the opportunity engage in resolution strategies. Moreover, according to the progress note, the dyad established “not quitting therapy” as an explicit goal, which perhaps also served as a protective factor against premature termination. Patient C1, on the other hand, demonstrated more avoidance and abstract communication on the 3RS and, as the therapist’s post-session scores are higher than the patient’s in this case, it is possible that the therapist viewed the session as smoother with more agreement and collaboration than the patient felt it was.

Therapist D

The therapist was a 34-year-old female Master’s level clinician identifying as White (not Hispanic) and Jewish. She reported one year of clinical experience and identified with a cognitive/behavioral orientation.

Patient D1 (Drop)

The patient was a 28-year-old, married, college-educated and employed male identifying as White (not Hispanic) and areligious. At intake, he was diagnosed with Social Phobia (Axis I) and Avoidant Personality Disorder (Axis II). The patient dropped out after the second session.

Session Summary.

The patient rated the alliance and session depth higher than the therapist across both sessions (see Table 4). Neither therapist nor patient reported any problems in the relationship in either session. Therapist progress notes reflected the topics of the session, which included psychoeducation about CBT (e.g., “discussion of how [a therapy notebook] might be used to help him internalize and implement therapeutic work”) and identification of goals and homework assignment (“Continue noticing and recording his thoughts, also considering which of the cog. distortions might characterize him”).

Table 4.

Therapist D

Process Measure	Pt D1 (D)		Pt D2 (T)
	P-Rated	T-Rated	P-Rated	T-Rated
WAI Total
Session 1	6.00	4.83	5.00	4.58
Session 2	6.25	5.33	5.75	5.33
SEQ/Smoothness
Session 1	**	4.80	3.80	3.00
Session 2	5.00	5.00	2.20	4.00
SEQ/Depth
Session 1	4.00	4.40	5.20	4.20
Session 2	4.60	4.40	6.00	4.60
3RS	Session 1 (M)	Session 2 (M)	Session 1 (M)	Session 2 (M)
Total Withdrawal	4.0	3.5	3.0	3.0
Content/affect split	3.5	3.0	N/A	2.5
Deferential and appeasing	N/A	3.5	N/A	2.5
Abstract communication	4.0	N/A	N/A	4.0
Minimal response	3.0	2.5	N/A	N/A
Avoidant storytelling / shift	N/A	N/A	3.5	3.5
Self-critical / hopeless	N/A	2.5	2.5	N/A
Denial	N/A	N/A	N/A	3.0
Total Confrontation	1.5	2.0	1.5	1.0
Complaint of activity	3.0	N/A	N/A	N/A
Complaint of parameter	N/A	2.5	N/A	N/A
Resolution Strategies
Illustrate task/provide rationale	3.0	N/A	3.5	N/A
Validate defense	2.5	N/A	2.5	N/A
Redirect	N/A	N/A	3.0	N/A
Degree to which ruptures were resolved	3.5	2.0	3.5	1.5
Degree to which therapist caused or exacerbated ruptures	2.5	3.5	1.5	3.0

Note. D = Drop; T = Term. WAI = Working Alliance Inventory. SEQ = Session Evaluation Questionnaire. M = Mean of 3RS clinical impact (significance) scores; only markers rated 2.5 or higher are included. N/A = Not Applicable.

^**Missing data

Per the 3RS, the patient spoke in a vague or abstract way, and withdrawal markers had “moderate impact on the alliance” in both sessions. Coders’ notes suggested that the patient “appears to want to collaborate with the therapist but finds it difficult to do with his anxiety.” Coders’ notes also highlighted that in the first session, the patient pushed back against homework assignments but in the second, seemed more deferential and described the homework as helpful.

The therapist engaged in illustrating task / rationale for therapy and validating defenses in the first session, though there were no impactful therapist resolution strategies in the second. Coders described the therapist as “authoritative about psychoeducational components of CBT” while noting that she “also appeared extremely anxious” and potentially, “there was some fear of not being liked by the patient.” Coders rated the extent to which ruptures were resolved between “average” and “good” in the first session, with the therapist “maybe” or “somewhat” contributing to the ruptures. They rated rupture resolution as “below average” in the second session, with the therapist “somewhat” or “moderately” contributing to ruptures.

Patient D2 (Completer)

The patient was a 31-year-old, single, college educated and employed female identifying as White (not Hispanic) and areligious. At intake, she was diagnosed with Major Depression, single (Axis I) and Avoidant Personality Disorder. The dyad completed 30 sessions; the first two sessions were analyzed.

Session Summary.

With the exception of smoothness in the second session where the patient scored lower than the therapist, the dyad had comparable scores of alliance and session impact across both sessions (see Table 4). Neither reported any problems in the relationship in either session. The first session’s progress note reflected a brief history of the patient’s presenting problem and agreement on homework (i.e., “[Patient] agreed to get a therapy notebook and to begin noticing and writing down thoughts she has while feeling anxious”). In the second session, the therapist stated that “much of the session was spent gathering a more detailed history of the chronology of client’s anxiety.”

3RS codes displayed another primarily abstract communicative patient presentation with withdrawal markers having “some impact” on the alliance in both sessions. Coders’ notes suggested that the patient “has great difficulty staying with or elaborating her experience” and seemed to be “dealing with heightened anxiety and a lot of sadness.” The therapist was active in illustrating task / rationale for therapy and validating defenses in the first session, with no clinically impactful resolution strategies in the second. Coders described the therapist as “flexible with the CBT protocol… did a good job adjusting to the patient’s needs” but also appeared a bit anxious herself. One coder noted, “[Patient] could possibly appreciate the high anxiety levels of the therapist as she had a matching state of anxiety.” Coders rated the extent to which ruptures were resolved between “average” and “good” in the first session, with the therapist “maybe” or “somewhat” contributing to the ruptures. They rated the second session “below average” in rupture resolution, with the therapist “somewhat” or “moderately” contributing to ruptures.

Analysis

According to coders’ process notes, the patients’ and the therapist’s anxiety dominated the therapeutic process across both dyads, highlighting the importance of “fit” as well as therapist flexibility. With Patient D2, who, according to the coders, appeared almost disoriented by her anxiety, the therapist showed more activity in adjusting the treatment protocol (per 3RS), her post-session WAI scores also demonstrating an awareness of some disagreement about therapy tasks. As the coders suggested, the therapist’s anxiety may have also been “received” well by the patient and may have even facilitated their bond. With Patient D1, the therapist seemed to adhere more to the protocol, despite the patient’s explicit push-back about homework (as evidenced by both 3RS scores and coders’ notes). Interestingly, the patient returned to the second session with less confrontation but more withdrawal markers, particularly deference, which suggests he may have been giving therapy “another try” before dropping out.

Therapist E

The therapist was a 28-year-old female Master’s level clinician identifying as Asian or Pacific Islander and Sikh. She reported one year of clinical experience and identified with a dynamic/analytic orientation.

Patient E1 (Drop)

The patient was a 33-year-old, single, graduate-level educated and employed male identifying as Hispanic/Latinx and Baptist. At intake, he was diagnosed with Major Depression, single (Axis I) and Obsessive-Compulsive Personality Disorder (Axis II). The patient dropped out after the second session.

Session Summary.

The patient rated the three domains relatively low in the first session, with a slight increase in the second, though all scores were lower than the therapist’s (see Table 5). Therapist’s progress notes were short and factual, reviewing the topics of the sessions (e.g., “Family issues, relationship issues, feelings of loneliness, worries of rejection”).

3RS codes displayed a complex rupture presentation with both withdrawal and confrontation markers present in the first session, the former having “some impact” on the alliance, and only content/affect split present in the second session indicated “minor impact” on the alliance. Coders’ notes suggest that the patient “appears fearful about entering therapy and potential rejection from the therapist.” The therapist was fairly inactive in impactful resolution strategies, with only illustrating task / rationale for therapy in the second session. Coders rated the extent to which ruptures were resolved between “poor” and “below average” in the first session, with the therapist “moderately” contributing to the ruptures. They rated rupture resolution as “average” in the second session, with the therapist “maybe” contributing to ruptures. Coders had strong reactions to the therapist’s interpersonal style, which was “active” but at times felt “mocking.” One coder noted, “There was some sarcasm and intellectualization that they both participated in during the session. It was difficult to discern if this was alliance building or colluding.”

Patient E2 (Completer)

The patient was a 24-year-old, single, college-educated and employed female identifying as White (not Hispanic) and Jewish. At intake, she was diagnosed with Major Depression, recurrent (Axis I). The dyad completed 27 sessions; the video for session 2 was missing, so the first and third sessions were analyzed.

Session Summary.

The patient rated the three domains higher than the therapist across both sessions (see Table 5). Though the patient did not report any problems in the relationship, the therapist noted in the first session that she “felt as though I wasn’t able to understand what [the patient] was going through,” which she addressed by “letting [the patient] talk more without trying to give my own inferences.”

3RS codes displayed a complex presentation with only withdrawal markers present in the first session having a “minor impact” on the alliance, and an increase in both withdrawal and confrontation markers in the third session, the former having a “moderate impact” on the alliance. Coders’ notes suggest that the patient appeared both “somewhat compliant and affectively defended” and “cognitively engaged and motivated.” The therapist did not employ many resolution strategies, only illustrating task / rationale for therapy in the third session. Coders rated the extent to which ruptures were resolved as “average” in the first session, with the therapist “somewhat” contributing to the ruptures. They rated rupture resolution as “below average” in the second session, with the therapist “moderately” or “mostly” contributing to ruptures. Coders reported that the therapist “was trying too hard to offer solutions… without gauging how the patient was responding to them.” Coders highlighted that this was a confusing dyad in which it felt like “two different conversations were being had.”

Analysis

According to the coders’ process notes, this therapist was among the most directive within this sample; with both patients, this seemed to contribute to rupture process, as her comments may have been construed as insensitive. As such, Patient E1’s premature termination did not come as a surprise to coders. Moreover, this patient rated the alliance and session impact as lower than the therapist, which, again, may indicate that the therapist believed the sessions to be going better than patients experienced them to be. With Patient E2, the first session, though not rupture-free, seemed to fare better for the dyad. Their third session came as a surprise with increased ruptures and increased therapist rupture exacerbation; as we were unable to observe the second session, we can only speculate that the alliance established in the first session may have allowed for more rupture process by session 3.

Discussion

Several themes emerged through the data consolidation of these sessions, which we discuss below.

Prevalence of Ruptures

We found that rupture markers were prevalent across all sessions, regardless of completion status, and sometimes even more so in completer cases than dropouts. This finding is consistent with the notion that ruptures are common events in the alliance (Eubanks et al., 2018b; Muran, 2019).

Across both groups, withdrawal markers dominated the rupture profiles, though the clinical impact of withdrawal markers tended to be higher in dropout cases. Withdrawal rupture markers are typically more subtle rupture markers, and it is not unusual for therapists to feel stuck or confused or to overlook them altogether (see Eubanks et al., 2018b). This is consonant with Gulum et al.’s (2018) findings that patients and therapists were more disengaged from the activity of therapy in dropout cases compared to completers. In a withdrawal rupture, patients often may have difficulty acknowledging their concerns or expressing them to the therapist, due to worries about disapproval or retaliation (Boritz et al., 2019). Moreover, sometimes patients may be withdrawing from themselves (e.g., denying a feeling state), which could further complicate a therapeutic interaction. Further investigation of withdrawal rupture profiles in larger and more diverse samples could shed light on whether different withdrawal markers are more or less linked with dropping out of psychotherapy.

In contrast, we noticed that completer patients exhibited more confrontation markers. While some confrontation markers (e.g., control/pressure) are subtle and more difficult to recognize, more overt confrontation markers (e.g., rejection of interventions) often create tension in the room and are more readily identifiable. Completer patients may have felt more comfortable expressing apprehension or skepticism, and that comfort, in itself, was protective against premature termination.

Eubanks and colleagues (2021b) highlighted the importance of the therapists’ response to the confrontation as it could contribute to a corrective experience if approached with curiosity and compassion rather than counter-hostility. Similarly, in this sample, patients exhibiting more confrontation markers may have opened “windows of opportunity” by allowing concerns to be named and addressed and potentially repaired. Among the patients who dropped out, this was often not the case, either because the patient was more withholding or because the therapist did not allow enough space for it.

Across both groups, we observed several complex rupture profiles that included confrontation and withdrawal markers. Patient C1, for example, communicated her concerns in an intellectualized way, often shifting topics, while also dismissing her therapist’s interventions. This kind of presentation may contribute to the therapist feeling like they just “can’t win.” In this case, Therapist C “checked out” of the therapeutic relationship during the session preceding the patient dropping out.

These results suggest the importance of CBT trainee therapists attending closely to withdrawal markers and more complex ruptures which, again, can be more difficult to identify and subsequently address. While therapists tend to “look out for” confrontation rupture markers and believe withdrawal markers to be less impactful (see, e.g., Eubanks et al., 2018a), our results point to the contrary – confrontation markers can be a part of an active and mutual engagement, while withdrawal markers may be a sign of trouble ahead.

Prevalence of Rupture Resolution

We found no meaningful differences in the clinical impact of therapist resolution strategies. Across both completer and dropout cases, the degree to which ruptures were resolved did not surpass the “good, above average” threshold on the 3RS. As our sample consisted of trainee therapists, it is possible that therapists felt less equipped to repair ruptures, particularly complex ones. However, while we did not find differences in the clinical impact of resolution strategies, we did notice more resolution strategies being used with the completer cases. Similar to rupture, we believe this was a co-constructed process in which patients may have been more disclosing and confrontational, which opened the door for therapists to be more responsive. There is extant research supporting that in order to repair ruptures, therapists must be first able to recognize rupture markers (see Muran & Eubanks, 2020, p. 4), and confrontation markers in this sample may have been easier to identify and respond to.

Therapist Responsiveness.

Therapist flexibility and responsiveness were recurring themes in the completer cases. Responsiveness includes behaviors that are influenced by and appropriate to the context of the therapeutic relationship (Stiles & Horvath, 2017) and is a prerequisite for recognizing and repairing a rupture (Eubanks et al., 2021a; Muran & Eubanks, 2020). For example, coders perceived Patient A2 to be a difficult patient with a “domineering” style of communication; she seemed defensive but also explicitly expressed her apprehension about the more structured aspects of CBT. In response, the therapist actively implemented resolution strategies and remained flexible, adjusting the focus of the session to alliance-building. It is often easier to be attuned to explicit rupture markers, and the recognition of the rupture and the attempt to resolve it by encouraging the patient’s commitment to therapy may have contributed to the patient continuing treatment. Moreover, for someone like Patient A2, a therapist who responds compassionately and non-defensively to her style of communication may not only be comforting, but also shape her expectations of interpersonal interactions.

Therapist Awareness.

We found that compared to dropouts, completer dyads were more “on the same page” in their early sessions based on WAI and SEQ data. Congruence between patient and therapist alliance ratings has been found to predict patients’ symptom levels (e.g., Zilcha-Mano et al., 2016b). While we did not look at symptom levels, we found that dyads who completed treatment demonstrated more congruent scores in their alliance and session impact ratings. Therapist D, for example, had almost identical alliance and session impact scores with her completer case, but rated the first session’s depth higher than did her dropout patient. Similarly, Therapist E rated the overall alliance and session impact higher than Patient E1, who dropped out after two sessions. Not only is this noteworthy, as patients generally report higher alliances than therapists (Tryon et al., 2007), but it suggests incongruence in the perception of alliance and session impact.

Therapist Contribution to Ruptures.

All therapists in this study were rated as at least “maybe” exacerbating the rupture process (i.e., all received a 2 or greater on the therapist rupture contribution item). 3RS’s implicit assumption that patients cause ruptures and therapists initiate resolution is a limitation of the measure (Eubanks et al., 2019); as such, coders’ process notes offered meaningful insights into therapists’ contributions to ruptures. With Patient D1, for example, Therapist D appeared anxious and concerned about being liked by the patient, while Therapist E’s utilization of humor with Patient E1 was experienced by coders as “mocking.” These findings are consistent with the view of rupture and resolution as inherently dyadic (Safran & Muran, 2000).

Training Implications. While there are standardized efforts to train therapists to identify and address difficult moments in psychotherapy (see Alliance-Focused Training [AFT]; Eubanks-Carter et al., 2015), there are easy and practical ways that therapists can be encouraged to pay more attention to inter- and intra-personal process in the therapy room. Check-ins at the end of a session, which are already often integrated into CBT protocols, can serve as a space for the therapist to more actively inquire about reservations or concerns the patient may be having about the psychotherapy. Moreover, therapists are encouraged to pay attention to their own feelings. For example, if a therapist finds themselves talking more than usual, feeling bored when they usually don’t, or wanting the session to end, it could be a sign of a significant relational shift within the dyad. Adequate clinical supervision and supervisory elements (e.g., video review of sessions) could also facilitate a trainee’s ability to critically examine their interactions and identify their blind-spots. Muran and Eubanks (2020), who have written extensively on alliance and rupture processes, provide extensive therapist training recommendations, that include both didactic and experiential strategies.

Strengths, Limitations, and Areas for Future Research

The use of multiple perspectives (patient, therapist, observer) in this study underscores the importance of utilizing multimethod evaluations of complex phenomena. Alliance and outcome research has historically relied on patient self-reported data (see Eubanks et al., 2018a; McLeod, 2001 for critique), shifting recently to include more observer-based measurement. While this is a notable development, much is still assumed when we rely on a single assessment method. For example, Patient A1, who dropped out of treatment, rated the alliance particularly high, but it was the therapist’s progress note and 3RS codes that revealed this patient’s deferential stance. To our knowledge, this is the first application of the Sort, Sift, Think, and Shift method in examining psychotherapy process. This methodology is novel and promising and, as such, would benefit greatly from continued utilization and investigation with different datasets and types of data. Relatedly, ongoing exploration of psychotherapy and alliance also requires continued rigor and innovation to better understand the multi-faceted nature of complex phenomena, particularly since our findings underscored the intricacies of even early therapeutic interactions.

Similarly, our utilization of coders’ process notes as data aligns with recent developments in qualitative research. Such notes can provide essential context to inform data analysis, encourage researcher reflection and identification of bias, and supplement language-based data (see Phillippi & Lauderdale, 2017). Coders in this study pointed out factors not typically considered, such as therapists’ anxiety levels and how well humor was utilized.

It is important for us to meet the recommendations for qualitative research in psychology (see Willig, 2014) and recognize our positionality. 3RS coders were all doctoral students in clinical psychology, most of whom have had at least some exposure to alliance-focused training and related concepts. All four have also had extensive exposure to and formal training in CBT through coursework and clinical supervision; one coder has conducted manualized CBT and two others have previously coded adherence to a CBT protocol. However, most of the authors identify with a psychodynamic and/or integrative approach to psychotherapy. While there may be something missed or misunderstood when a coder does not identify with the same theoretical orientation as what they are coding, things can similarly be missed when the opposite is true, due to shared “blind spots” and allegiance effects. Moreover, several authors have published on alliance rupture repair and view it as an important therapeutic process.

Our sample size did not allow for in-depth exploration of differences in demographics or clinical presentations. These were relatively inexperienced therapist trainees; thus, the findings may not generalize to therapists at large. Within the five dyads, three of the dropouts preceded the completer case; it is possible that therapists were impacted by a patient dropping out and thus were more active with their subsequent cases (Barrett et al., 2008). Moreover, these dyads were all part of a psychotherapy research program, the additional demands of which (e.g., thorough intake, video-recording, and post-session questionnaires) could have also contributed to psychotherapy process.

Previous qualitative examinations of dropout have identified lack of buy-in to the rationale or specific therapy tasks as a barrier to treatment continuation (see Hundt et al., 2020). While the most common rupture resolution strategy utilized in this sample was illustrating tasks or providing a rationale for treatment, it is possible that a CBT approach was not a good fit for some of these patients.

Personality style match could have played a role in alliance bond formation in our sample (see e.g., Taber et al., 2011). Therapist D was identified as an anxious therapist with both of her patients, exhibiting also almost identical rupture and resolution patterns in the two cases. However, Patient D2 was also dealing with heightened anxiety, and part of our analysis centered on the possibility that the patient either did not notice her therapist’s anxiety due to her own or perhaps even appreciated the matching state.

Some have argued for a distinction between dissatisfied dropouts and “got-what-they-needed” dropouts (e.g., O’Keeffe et al., 2020, p.3). Since the latter is possible even after one or two sessions (Simon et al., 2012), we realize that it is impossible to capture all the factors that could contribute to patients’ choices around continuing psychotherapy. Patient E2, for example, had an abrasive therapist; in fact, the coders were confident that this was a dropout case. Both patient and therapist reported fairly average alliance and lower session impact scores for both sessions, and the therapist was rated as moderately exacerbating ruptures. It is unclear what made Patient E2 stay for an additional 28 sessions, while the therapist’s other patient did not.

Research suggests that patients avoid expressing dissatisfaction to their therapists (e.g., Hill, 2010), but it is important to note that those who dropped out in this sample and still completed the post-session questionnaire after their last session did not disclose their dissatisfaction with treatment in the questionnaire. This leaves the question of whether patients knew when leaving the session that they would not come back or whether that was a decision made later.

Conclusion

We found that rupture process was prevalent across both completer and dropout cases and that multimethod evaluation allowed a nuanced understanding of complex psychotherapy process. While the clinical impact of withdrawal markers was high for both groups, it was usually higher in dropout cases. Conversely, those who completed treatment typically exhibited more confrontational markers, showing more complex rupture profiles. Therapists were more active in employing resolution strategies with those patients, often responding with flexibility and mutuality, highlighting the importance of being “on the same page.” Our findings suggest that therapists, particularly trainees, could benefit from training in alliance rupture identification and resolution to better navigate these challenging moments and prevent dropout.

Practical Implications:

Alliance rupture process may play a critical role in dropout. Therapist flexibility, awareness, and responsiveness to process may serve as a protective factor against premature termination. Training in rupture identification and resolution could help trainee therapists remain attuned to psychotherapy process and prevent early dropout.