Examining Goal, Task, and Bond in Therapeutic Alliance Ruptures

Extant literature has identified two types of therapeutic alliance ruptures—withdrawal and confrontation—with withdrawal ruptures being categorized as a distancing of oneself from the treatment process and moving towards isolation or appeasement behaviors and confrontation ruptures being signaled by hostility and aggression externalized onto another person (Muran et al., 2021). Despite the harmful implications of a ruptured therapeutic alliance, ruptures can also represent important events in treatment where the client and psychotherapist can identify maladaptive interpersonal schemas held by the client.

Repairing therapeutic alliance ruptures in psychotherapy can lead to a more robust therapeutic alliance than previously experienced, a reduction in client distress, and better overall treatment outcomes (Eubanks et al., 2018; Westerman & de Roten, 2017). Conversely, therapeutic alliance ruptures left unrepaired can lead to poorer treatment outcomes (Flückiger et al., 2018), and eventual treatment dropout (Waddington, 2002).

Diverse methods have been utilized to aid in rupture repair and reduce the impact of ruptures in treatment. A psychotherapist merely recognizing that a rupture occurred post-session can lead to a stronger therapeutic alliance in the next session as well as a reduction in the harmful impact of a rupture on client symptom change (Chen et al., 2018). Additionally, changing the therapeutic technique implemented by the psychotherapist when facing an alliance rupture has been shown to be effective in improving client functioning (Chen et al., 2020). Rupture resolution training has also been applied to increase treatment outcomes, though the evidence of the efficacy of said trainings has been inconsistent (Eubanks et al., 2018). What is known is that negative countertransference patterns experienced by the psychotherapist may be associated with poorer treatment outcomes and more therapeutic alliance ruptures (Tishby & Wiseman, 2020).

A rupture in the alliance can be described as a disagreement of treatment tasks and goals and a weakening of the relational bond (Muran, 2019). Utilizing this definition, the purpose of the current study was twofold. For the first goal, we sought to empirically investigate the disagreement of goals and tasks and the deterioration of the therapeutic bond as the driving forces behind alliance ruptures, as put forth by Muran. We hypothesized that disturbances in some facets of the therapeutic alliance—goal, task, and bond—would be more frequent and more commonly lead to alliance ruptures, though we did not specify which facet we believed to occur most frequently. The second goal of the current study was to investigate the predictive utility of the presence and frequency of alliance ruptures as a predictor of treatment outcomes. We hypothesized that the presence and frequency of alliance ruptures would predict treatment outcomes and negatively affect treatment trajectory.

Methodology

Participants

Participant data came from archival clinic records of adult clients (N = 399) who had terminated from outpatient psychotherapy. Clients were majority female (54.3%; male = 43.2%; gender minority = 2.5%) with a mean age of 34.2 years (SD = 14.7). Slightly over half of the clients (51.3%) identified as non-Hispanic White with 18.4% identifying as Hispanic/Latinx, 9.0% identifying as Black/African American, 3.0% identifying as Asian/Pacific Islander, 2.2% identifying as Middle Eastern or Southeast Asian, 0.4% identifying as Native American/Alaskan, and 15.7% identifying as multiracial or other.

Psychotherapists in this study were doctoral trainee clinicians enrolled in one of two accredited programs (clinical psychology and counseling psychology). Due to risk of identification, the archival data custodian excluded clinician gender and race/ethnicity variables but noted the majority of clinicians self-identified as non-Hispanic White females. Prior research in this clinic suggests male representation was likely around one-third of the clinicians, while minority race/ethnicity likely characterized approximately 20% of the clinicians. All cases are assigned by the Clinic Director and all clinicians receive at least 3 h of supervision weekly by a doctoral-level psychologist.

Measures

Outcome Questionnaire-45.2 (OQ)

The Outcome Questionnaire-45.2 (OQ) is a standard self-report outcome measure (Lambert et al., 1996) used to track client progress and outcomes. Prior to each session of psychotherapy, clients reflect on their experiences during the preceding week and provide responses to 45 items assessing symptomatic distress, with responses ranging from 0 (never) to 5 (almost always). The OQ relies on cutoff scores to indicate high and low levels of global client functioning. The OQ has also been shown to have high internal consistency (α = 0.93) as well as high test-retest reliability (r = .84; Lambert et al., 1996; Lambert et al., 2004; Snell et al., 2001). Additionally, no gender or racial differences have been found in client OQ scores. The internal consistency of the OQ in our study (α = 0.93) was on par with that reported by the measure authors.

Outcome Questionnaire-Assessment for Signal Clients (OQ-ASC)

The OQ-ASC (Lambert et al., 2007) is a self-report questionnaire that consists of 40 items, each on a 5-point Likert scale. The OQ-ASC measures four distinct domains—therapeutic alliance (11 items), social support (11 items), motivation (9 items), and life events (9 items). For the purposes of this study, only client responses on the alliance scale were examined. The OQ-ASC relies on cutoffs to indicate high or low scores in each domain. An alpha coefficient is not computed for the questionnaire, but instead for each of the four domains (therapeutic alliance = 0.87; White et al., 2015). The same alpha value was found for the internal consistency of the OQ-ASC’s therapeutic alliance scale in our sample (α = 0.87).

Working Alliance Inventory – Short Revised (WAI-SR)

The WAI (Horvath & Greenberg, 1986) is a self-report questionnaire that consists of 36 items, each on a 7-point Likert scale. The Short Revised version of the WAI (WAI-SR; Hatcher & Gillaspy 2006) consists of 12 items and includes three subscales—tasks, goals, and bonds—each containing 4 items. The WAI-SR does not rely on a normative sample, rather the developers recommend comparing alliance ratings within and between clients to identify differences in therapeutic alliance development. An internal consistency of 0.91–0.92 has been found (computed with Cronbach’s alpha) and the WAI-SR has shown high correlation with other measures of alliance (CALPAS, r = .80; Penn Helping Alliance Questionnaire, r = .74; Hatcher & Gillaspy 2006). The internal consistency of the WAI-SR in our sample was similarly acceptable (0.78). Correlations among the three subscales are provided in Table 1.

Table 1 Working alliance inventory-short revised (WAI-SR) subscale correlations
Full size table

Psychiatric Diagnostic Screening Questionnaire (PDSQ)

The PDSQ (Zimmerman & Mattia, 2001) is a self-report assessment of psychopathological symptoms, used in this study to better characterize the sample. The PDSQ screens respondents for DSM-IV axis 1 mental disorders. Prior to their intake session, clients in this study completed the PDSQ in the waiting room of the clinic. The PDSQ is comprised of 13 subscales and 125 dichotomous yes/no questions. The average internal consistency for each subscale, found by the measure’s authors is α = 0.85. (Zimmerman & Mattia, 2001). The test-retest reliability of the PDSQ is 0.81.

Subscales that clients in this study most commonly met screening criteria for (with 90% confidence) were Social Phobia (72.4%), Generalized Anxiety Disorder (63.8%), Posttraumatic Stress Disorder (59.8%), and Major Depressive Disorder (59.8%). Over a quarter of clients met the screening criteria for Obsessive-Compulsive Disorder (48.6%), Somatization Disorder (39.6%), Agoraphobia (39.6%), Hypochondriasis (29.1%), and Alcohol Abuse/Dependence (26.6%). The majority of clients (90.4%) met the criteria for more than one subscale.

Procedures

The OQ-ASC and WAI-SR were completed by clients on a monthly basis and the OQ was completed prior to each session of psychotherapy. Adult clients receiving services had the option to consent to have their data used for research purposes, though service utilization was not dependent upon research consent. To be included in the study, clients must have attended at least one psychotherapy session prior to termination of services and closure of their file. All client data was deidentified and treated in accordance with ethical guidelines (American Psychological Association, 2017) and with the approval of the clinic’s Executive Committee as well as the Institutional Review Board.

Results

To identify and define ruptures with the WAI-SR, we used the precedence of prior therapeutic alliance rupture research (Chen et al., 2018, 2020). We first calculated the average unit of alliance fluctuation for each client by using the square root of the mean square successive difference score (RMSSD; von Neumann et al., 1941). The therapeutic alliance was identified as ruptured when the session rating was more than one average unit of alliance fluctuation below the mean of the previous three session therapeutic alliance ratings. For this reason, only ratings after the third therapeutic alliance ratings were evaluated for ruptures with the WAI-SR. Ruptures in the individual WAI-SR subscales were determined using the same approach. Therapeutic alliance ruptures were not defined as occuring only when a subscale ruptured, representing that while facets of the therapeutic alliance may be weakened or fractured, the overarching therapeutic alliance had remained intact—possibly through the bolstering of another undamaged facet. Ruptures in the therapeutic alliance identified by the OQ-ASC were defined as scores falling below 43 on the measure (Lambert et al., 2007).

The first goal of the current study was to investigate the frequency of disagreement of goals and tasks in treatment and the weakening of the therapeutic bond as causes of therapeutic alliance ruptures. A total of 988 sessions were evaluated for ruptures using the WAI-SR in this analysis, with 6.9% of sessions containing a therapeutic alliance rupture. The same previously mentioned method for identifying WAI-SR ruptures was used for each WAI-SR subscale (goal, task, and bond) to determine if ruptures in a subscale were co-occurring and driving the overall alliance rupture. Alliance ruptures were coded by the type of subscale ruptures that co-occurred with the overall rupture (goal, task, bond, goal + task, goal + bond, etc.). As hypothesized, there was a significant difference in the type of ruptures experienced by clients. Shown in Fig. 1, therapeutic alliance ruptures characterized by a rupture in the goal or task subscales were both more likely to occur than therapeutic alliance ruptures characterized by bond subscale ruptures (χ2(5) = 15.47, p = .009).

Fig. 1
figure 1

Frequency and type of therapeutic alliance ruptures

Full size image

The second goal of this study was to investigate therapeutic alliance ruptures as predictors of treatment outcome. Client treatment outcome was determined by subtracting the client’s distress at termination (final OQ score) from their distress at intake (initial OQ score). A total of 859 sessions were evaluated for ruptures using the WAI-SR and 1,619 sessions were evaluated using the OQ-ASC in this analysis. Approximately 5.70% (n = 49) of sessions evaluated for ruptures using the WAI and 5.06% (n = 82) of sessions evaluated using the OQ-ASC contained therapeutic alliance ruptures. We found that the magnitude of overall change in OQ scores was significantly predicted by therapeutic alliance ruptures identified by the WAI-SR (ß = − 0.16, t(179) = − 2.11, p = .036), though not by the OQ-ASC (ß = − 0.08 t(397) = − 1.56, p = .119). An increase in ruptures, measured with the WAI-SR, was associated with a reduction in psychotherapy gains. See Table 2 for information regarding the amount of variance in outcomes explained by each therapeutic alliance rupture measure.

Table 2 Predicting Treatment Outcome with Therapeutic Alliance Ruptures
Full size table

Analyzing the three types of alliance ruptures evaluated—alliance ruptures containing a bond rupture, a goal rupture, or a task rupture—we found that the frequency of individual rupture types was not predictive of treatment outcomes. Furthermore, we found that when comparing the change in treatment of clients who experienced no ruptures (M = 10.44) to clients who experienced at least one rupture (M = 3.24), clients who experienced no ruptures made significantly more change in treatment (t(178) = 2.27, p = .024, d = 0.38), demonstrating a medium effect size.

Discussion

A robust therapeutic alliance is important to predicting treatment outcomes (e.g., Baier et al., 2020; Flückiger et al., 2018). Given the importance of the therapeutic alliance in treatment, it stands to reason that damage to the alliance could be detrimental to treatment. The current study sought to expand upon Muran’s (2019) theorization of therapeutic alliance ruptures as a disagreement of goals and tasks and the deterioration of the therapeutic bond. We used two disparate methods to track alliance ruptures and used these methods to examine the predictive properties of the frequency of alliance ruptures as a predictor of outcomes. These results are important in helping to reconceptualize the role of the agreement of goals and tasks and the relational bond in alliance ruptures, as well as contributing meaningful findings to the current literature pertaining to alliance ruptures and treatment outcomes. More specifically, in examining the prevalence of ruptures in the WAI-SR subscales (Hypothesis 1), disagreement of goals and tasks in treatment were found to occur significantly more often than deterioration in the therapeutic bond. In short, not all facets of the therapeutic alliance rupture equally. This is the first study to our knowledge examining this issue in this manner and they carry important implications for future research.

Accepting that not all ruptures are the same, our findings suggest that supervisors could possibly be more successful if they concert supervision efforts on teaching techniques for collaborating with clients when determining the tasks and goals of treatment. Similarly, based on our findings it could prove beneficial for supervisors to place an added emphasis on rupture resolution training, specifically ruptures that occur due to a disagreement of the tasks and goals of treatment. Conversely, it is within the realm of possibility that an increased focus on goal/task collaboration in trainee supervision proves unproductive, as the ruptures could be a result of countertransference or a circumstance unrelated to the therapeutic relationship. Although improving the efficacy of trainee supervision was out of the scope of the current study, this is an area where the application of our findings may prove fruitful.

Our results demonstrate that it is more likely for an alliance rupture to be attributed to a disagreement regarding the goals and tasks of treatment than damage in the relational bond between psychotherapist and client. By identifying that a rupture occurred and in which facet of the therapeutic alliance, psychotherapists can increase efforts to repair ruptures and mitigate the negative effects (Chen et al., 2018). Furthermore, it may be beneficial for psychotherapists to devote more attention to collaborating with clients pertaining to the treatment goals and tasks, as these areas are the most likely to result in a rupture.

The second aim of the current study was to investigate the effect of ruptures—notwithstanding repaired and unrepaired status—on treatment outcomes (Hypothesis 2). We found that more ruptures in the therapeutic alliance significantly predicted less change in treatment, with clients who experienced no ruptures experiencing the largest psychotherapy gains. This finding was somewhat expected, given evidence suggesting the detrimental effects of an alliance rupture (e.g., Flückiger et al., 2018). An unexpected finding was the similar treatment outcomes between individuals who experienced different types of alliance ruptures. Researchers theorize that one of the key elements in repairing and recovering from a rupture is a strong bond between the client and psychotherapist (Eubanks et al., 2018). Therefore, it would be logical to assume that alliance ruptures occuring primarily due to a deteriorating bond would be the most difficult to repair or recover from, leading to the poorest treatment outcomes. This, however, was not consistent with our findings, as clients experiencing ruptures due to a disagreement of tasks, disagreement of goals, or a deterioration in the bond all had treatment outcomes that were not statistically significantly different.

Interestingly, we found that ruptures identified by the WAI-SR appeared to explain more of the variability in treatment outcomes than ruptures identified by the OQ-ASC. This has an important implication for clinics that routinely measure alliance ruptures in treatment, as these results suggests that alliance measured with the WAI-SR may better explain client distress at termination. The harmful effects of an unrepaired rupture—poorer outcomes, reinforced harmful schemas, and a weaker alliance—adds urgency to the need for psychotherapists to develop skills in identifying and repairing ruptures in the therapeutic alliance, as both methods have been shown to mitigate the impact of alliance ruptures (Chen et al., 2018; Eubanks et al., 2018; Westerman & de Roten, 2017). Despite the treatment trajectory implications of an unrepaired rupture, alliance ruptures provide unique opportunities in psychotherapy. As just one example, psychotherapists and clients might explore and work through maladaptive interpersonal schemas held by the client. Taken together, the results of the current study highlight the importance of identifying the source of the therapeutic alliance rupture and implementing steps for rupture repair.

Limitations and Future Directions

Our study had three primary limitations. The first limitation was the frequency in which alliance tracking measures were administered. Clients completed the WAI-SR and the OQ-ASC on a monthly basis. Increasing the frequency in which we administered both measures could have improved our ability to identify ruptures occurring from session to session, rather than from month to month. While acknowledging this limitation, the frequency in which we administered the WAI-SR and OQ-ASC provided sufficient power for our analyses and was actually more frequent than is commonly found in therapeutic alliance literature (Eubanks et al., 2018). An additional limitation was the lack of study focus on rupture resolution. By incorporating this aspect into the study, we could have better understood which types of ruptures are more likely to be resolved. A final limitation of the current study was the lack of more in-depth conceptual analyses, such as using a hierarchical model to determine if some psychotherapists are more likely to experience therapeutic alliance ruptures than others. Despite this limitation, the analyses use in this study were sufficient to answer our primary research questions. A strength of the current study is that it is highly generalizable. Our study was conducted in a naturalistic setting with a treatment as usual approach—psychotherapy—from a diverse group of providers.

This is the first study to our knowledge to examine the role of treatment goals and tasks and the relational bond as driving forces in therapeutic alliance ruptures, providing insight for informing rupture resolution interventions while opening the door for several future research possibilities. Future research should be focused on examining further differences in the goal, task, and bond scale regarding treatment. Furthermore, ruptures caused by these different facets may differ in the amount of time taken to repair and in the effectiveness of repair efforts. It may be beneficial to test the efficacy of rupture resolution techniques that specifically target the disagreement of goals and tasks. Research has identified withdrawals and confrontations as common types of therapeutic alliance ruptures. Future research should examine if ruptures in different facets of the alliance are more likely to co-occur with a withdrawal rupture or confrontation rupture. Finally, session-by-session qualitative research may shed light on why ruptures are likely to be attributed to a disagreement in goals or tasks than a weakening relational bond.

Our study focused only on the differences between the goal, task, and bond facets of therapeutic alliance and the function of therapeutic alliance ruptures as outcome predictors. The current study was unable to address the pertinent question of whether alliance ruptures are more problematic in certain types of treatments. This is an area of future study which would both shed more light on the nature of alliance ruptures as well identify techniques and therapies which may reduce their likelihood of occurrence.

Through identifying facets of the therapeutic alliance most vulnerable to fracturing—agreement of tasks and goals—our results emphasize the importance of continually collaborating on the content of treatment with clients. Our results also highlight the need for employing rupture repair strategies to mitigate damage that can come with ruptures. Ruptures offer unique events in treatment in which maladaptive interpersonal schemas held by clients can be come to the surface in a safe setting. This study indicates that of the facets of the therapeutic alliance, the relational bond is the least prone to rupturing. Implementing strategies that target disagreement in the tasks and goals of treatment represents the next step in bolstering the therapeutic alliance.