The Therapeutic Bond in Suicidal Crisis

Journal of Counseling Psychology| 2024, Vol. 71, No. 2, p. 115-125

The Role of the Therapeutic Bond When Working With Clients in Suicidal Crisis

Daniel W. Cox; Counseling Psychology Program, The University of British Columbia

Halina M. Deptuck; Counseling Psychology Program, The University of British Columbia

Olivia Fischer; Counseling Psychology Program, The University of British Columbia

Katharine D. Wojcik; Menninger Department of Psychiatry and Behavioral Sciences, Baylor College of Medicine

The desire to die by suicide has been linked with interpersonal difficulties and impeded clinical outcomes. Despite the emphasis on the therapeutic relationship in clinical guidelines for working with suicidal clients, little is known about how suicidal clients’ interpersonal difficulties manifest in clinical contexts. Additionally, there is limited understanding of the therapeutic relationship in single-session suicidal crisis contexts. Our aim was to examine the trajectory of the therapeutic bond in mediating clients’ suicidal desire and outcome in single-session suicidal crisis intervention. Single-session online text-based crisis intervention sessions (N = 354; Mage = 29.43, SD = 9.15; 64.5% women) were coded for suicidal desire, therapeutic bond (each quarter), and outcome. We examined the proposed sequential mediating model (suicidal desire to early bond to bond change to outcome) using structural equation modeling. The proposed sequential mediation model fits the data well, χ2 (11) = 22.030, p = .0241, root-mean-square error of approximation = .053, 90% CI [.019, .085], comparative fit index = .983, Tucker–Lewis Index = .977, and was a better fit than several alternative models. Further, the indirect effect from suicidal desire to outcome through early bond and bond change was significant (b = −0.474, 99% CI [−0.782, −0.203]). Our findings indicate that therapeutic bonds were beneficial for clients with elevated suicidal desire—and—elevated suicidal desire was negatively associated with therapeutic bonds. These findings highlight the importance of training clinicians to navigate the unique challenges of developing therapeutic bonds with acutely suicidal clients.

Therapeutic bond; therapeutic relationship; suicidal desire; suicide ideation; helping process

“There is compelling theoretical and empirical evidence that those who are suicidal have difficulty developing interpersonal connections and that connecting with clinicians is particularly important for interventions to be successful with suicidal clients. Thus, while trusting and caring therapeutic relationships (i.e., therapeutic bonds) may be particularly beneficial for clients with elevated suicidal desire, elevated suicidal desire may impede the development of trusting and caring therapeutic relationships. The purpose of our study was to explicitly examine this interpersonal conceptualization of suicidal desire and effective suicide intervention” (115 - 116).

“Developing a therapeutic relationship with suicidal clients is uniquely embedded with challenges and contention. While clients are contemplating ending their lives, a central goal of clinicians is to prevent suicide, often eschewing prioritizing developing therapeutic relationships with their clients and prioritizing collecting clients’ information to categorize suicide risk (e.g., low, medium, high) for predicting suicidal behavior. The challenges and distress for clinicians working with suicidal clients are likely further potentiated when they are conducting single-session suicidal crisis interventions where there are no preexisting therapeutic relationships…In single-session therapeutic approaches, therapeutic progress is expected to happen quickly, and substantial time at the outset of sessions cannot be set aside to develop therapeutic relationships. Relative to traditional (i.e., multisession) psychotherapy, in single-session therapy, clinicians must take a more active role—quickly evaluating the needs and motivations of clients and working within those needs and motivations… The unique demands of single-session suicide intervention are further extended in text-based communication mediums. In text-based contexts, clinicians cannot attend to or apply vocal or body language cues. Thus, they must rely on text to perceive clients and to demonstrate understanding and responsiveness” (p. 116 -117). 

This study conducted a meta-analysis with the following inclusion criteria: “clients had to be suicidal (stated in the session), 18 years or older, first-time crisis center users (due to differences between first-time and repeat users), and sessions had to be 30 min or longer… The sampling of text-based suicidal crisis intervention sessions occurred in two steps. First, [they] used search filters in the clinical records system to identify sessions that met the inclusion and exclusion criteria. Second, [they] randomly sampled (SAMPLE function in R) and then manually screened 400 of these sessions for inclusion and exclusion criteria. Of these 400, 46 did not meet inclusion criteria due to not being first-time clients of the center... The measures used in this study included the Segmented Working Alliance Inventory– Observer-Based Measure (SWAI-O), three items from the Beck Scale for Suicidal Ideation, and a “previously developed taxonomy of crisis counseling outcomes to code client outcome based on the level of solution that clients were able to achieve” (p. 119). 

“The goal of the present study was to examine if therapeutic bond trajectories mediated the association between suicidal desire and client outcome in a single-session online suicidal crisis intervention… Our findings indicate that those with greater suicidal desire have weaker therapeutic bonds with clinicians. This link between suicidal desire and impeded relationship development is consistent with previous work demonstrating that those with suicidal desire have elevated interpersonal coldness—behaving in ways that keep others at a distance or push others away” (p. 122). 

“Our findings indicate that the early relationship sets clinicians and clients on a course for continued productive therapeutic relationships, which can help facilitate good client outcomes. While we did not investigate how the relationship facilitates good outcomes, one hypothesis is that relationship development—in and of itself—is the therapeutic agent… The two studies that have examined the association between the therapeutic relationship and outcome for clients receiving clinical services explicitly designed for clients with suicide-related thoughts and behaviors have indicated notably larger effects of the therapeutic relationship than has been found—on average—in non-suicide-specific studies of the relationship” (p. 122).

“First, it is important for clinicians—across therapeutic approaches— to conceptualize their work with clients in suicidal crisis as similar to much of their other clinical work—aiming to understand clients’ suffering and motivations to facilitate collaborative therapeutic processes, regardless of treatment approach. We contrast this collaborative model with a model in which clinicians prioritize collecting clients’ information to categorize suicide risk (e.g., low, medium, high) for predicting suicidal behavior. Categorizing with the aim of predicting risk places tremendous expectations on clinicians that likely impedes their ability to develop a strong therapeutic relationship. Further, categorizing suicidal risk with the aim of predicting attempts is of little practical value due to our (i.e., clinicians) inability to accurately predict suicidal behavior. However, taking a collaborative approach likely reduces clinicians’ feelings of pressure and anxiety, facilitating their ability to develop stronger and more effective therapeutic relationships. Our second recommendation is practice. A task force of clinical suicidologists identified 24 core competencies for working with suicidal clients. However, for many clinicians, it may be months or years between when they work with clients in suicidal crisis. It is unreasonable to expect clinicians to maintain competency while doing such infrequent work with suicidal clients. By practicing with peers, supervisors, or as part of continuing education, clinicians’ work with suicidal clients will become more intuitive, increasing their ability to maintain a collaborative stance and facilitate therapeutic relationships. Our third recommendation is that clinicians use—and practice with—semistructured clinical interviews. With so many expectations on clinicians working with suicidal clients, it is unreasonable to expect clinicians to remember everything that they hope to discuss. Having semistructured interviews—that clinicians have practiced and have become comfortable with—can be a useful tool for facilitating clinicians’ attention on their clients rather than on what they need to remember to ask. Of note, we follow the lead of others and caution against clinical interviews being used as definitive scripts. Rather, clinical interviews should be conceptualized as a tool for facilitating collaborative work” (p. 123). 

“Our study had several limitations that should be considered when conducting future research. In the present study, we examined suicidal desire as a predictor of the therapeutic bond and client outcome; however, future work would do well to consider the specific cognitions or behaviors that frequently co-occur with suicidal desire (e.g., thwarted belongingness,interpersonal coldness). This would facilitate more direct examinations of the psychological phenomena that impede therapeutic relationships and provide more direct targets for intervention. Future work should also consider clinicians’ effects. Samples that contain several clients per clinician would allow for examining how clinicians differentially impact therapeutic processes. In particular, how much of the early relationship—and its association with relationship growth—is due to clinician effects (e.g., therapist effectiveness or therapist priorities) versus how much is due to client effects. It is also important that future work include studies that are less than 30 min. Generalizing the evidence that poor therapeutic relationships are associated with premature termination, excluding sessions less than 30 min may have biased our sample. Another limitation is that we did not consider any within-session predictors of change in the relationship. For example, while most longitudinal investigations have concluded that the therapeutic relationship is predictive of client outcome—and not the other way around—some studies have identified reciprocal model where the relationship predicts outcome and outcome predicts the relationship. In many ways, this reciprocal model is consistent with the single-session sequential model that we proposed—bond facilitates therapeutic progress, which facilitates bond, which facilitates therapeutic progress, and so forth. Future work that directly examines the proposed within session gains would facilitate more direct tests of the sequential model. Additionally, studying micro processes such as alliance ruptures may facilitate a more nuanced understanding of how the relationship impacts client outcomes and how clinicians can utilize the relationship for therapeutic gains. There would also be value in future research examining the association between the therapeutic relationship and client outcome in the online crisis context to examine outcome in a variety of ways. Our outcome measure was ordinal, observational, and short term. Studies using continuous, self-report, or longer term outcome measures would facilitate a better understanding of the multifaceted impacts of the therapeutic relationship as well as measurement-related method variance. It is also important that future research examines the generalizability of our findings. We conducted our study in the context of single-session online suicidal crisis intervention. While the theory and research that grounded our study was from breadth contexts, empirical examination of the generalizability of our findings is important to better understand their contextual parameters. There would also be value in considering the impact of sociodemographic variables in studies of suicidal crisis interventions. These data were not collected by the community organization we partnered with; therefore, we were unable to consider them in our study. However, the impact of race, ethnicity, and other sociodemographic variables could certainly impact the associations between suicidal desire, the therapeutic relationship, and client outcome” (p. 123 - 124).