Blog

The Great Psychotherapy Debate: Since in April, 2015 I review parts of The Great Psychotherapy Debate (Wampold & Imel, 2015) in the PPRNet Blog. This is the second edition of a landmark, and sometimes controversial, book that surveys the evidence for what makes psychotherapy work. You can view parts of the book in Google Books.

Wampold, B.E. & Imel, Z.E. (2015). The great psychotherapy debate: The evidence for what makes psychotherapy work (2^nd edition). New York: Routledge.

Some therapists achieve better patient outcomes than others. This seems obvious on the surface and yet few people talk about it, and the research literature seems to downplay or ignore this fact. To illustrate the differing outcomes achieved by therapists, I reviewed a unique study in the September 2013 PPRNet Blog. In that study, 10 cases were randomly selected from 700 therapists (N = 7000 patients), and therapist outcomes were assessed by averaging their patient outcomes. Depending on the presenting problem, as many as 67% of therapists were reliably effective, but as many as 16% were reliably harmful. Clearly therapists differ. Yet psychotherapy research typically treats therapists as if they are uniformly effective. In their chapter, Wampold and Imel review some of the research that estimates the therapist’s effect on outcomes. In other words, what is the impact of the particular therapist on the patient outcomes? Even in studies in which: (a) therapists are selected as experts to provide a specific type of therapy (i.e., CBT, psychodynamic, interpersonal, etc.), (b) therapists are highly trained to be adherent to a manual with repeated supervision, and (c) patients are randomly assigned to treatments, there remains a significant amount of variability in therapist outcomes. Indeed in many studies the therapist effect is as large or larger than the effect of the intervention that is being delivered. In other words, which therapist a patient gets in a treatment study matters just as much or more than what type of therapy they receive. This is also true in medication trials. Better psychiatrists (i.e., those with overall better patient outcomes) who gave a placebo had better patient outcomes than poorer psychiatrists who gave the active medication. A recent large meta analysis found that about 5% of patient outcomes in controlled psychotherapy trials was attributable to the therapist, and the effect is as high as 7% in naturalistic settings. For treatment of PTSD, therapist effects are as high as 12%. On the surface these look like small effects, but in reality they can have a big cumulative impact. Therapists with the best and worst outcomes differ dramatically. For example in one large study, the best performing therapists had a patient response rate of 80% compared to the worst performing therapists who had only 20% of their patients improve. Which therapist would you want a loved one to see?

Practice Implications

Wampold and Imel reported that that therapist effects generally exceed the effects of the specific treatment that is being tested or provided. Some therapists consistently achieve better patient outcomes than others. What are the characteristics and actions of effective therapists? Factors like therapist allegiance to the therapy, empathy, and the ability to form and maintain an alliance with clients appear to differentiate therapists who consistently have good patient outcomes versus those whose patients tend to have poor outcomes.

Boswell, J.F., Kraus, D.R., Miller, S.D., & Lambert, M.J. (2015). Implementing routine outcome monitoring in clinical practice: Benefits, challenges, and solutions. Psychotherapy Research, 25, 6-19.

Routine outcome monitoring (ROM) refers to: (1) systematically assessing patient outcomes at every session, (2) comparing patient scores and progress to a database of similar patients, (3) using algorithms or decision tools to identify patients who are not improving or deteriorating, (4) providing regular and immediate feedback to therapists about the patient, and (5) in some cases providing clinical decision aids to help therapists improve outcomes for patients who are not improving or who are deteriorating. Boswell and colleagues review the research related to ROM. Generally, about 30% to 50% of patients do not respond to treatment, and 8% of patients tend to get worse during treatment. Therapists tend to overestimate their patients’ improvement, and so therapists may not always identify patients who do not respond or get worse. Therapists may need assessment aids to help them make decisions about patient progress and treatment. Boswell and colleagues point out that ROM have a proven ability to predict treatment failure and other negative outcomes. In a meta analysis of over 6,000 patients, the patients at risk of a negative outcome whose therapist received ROM feedback prior to every session were better off than 70% of at-risk patients whose therapist received no feedback. When therapists are provided feedback and suggestions for interventions, their patients had almost four times higher odds of achieving clinically significant improvement. Boswell and colleagues list a number of barriers that psychotherapists and agencies experience to implementing ROM in their practices. Many therapists are not aware of or have no experience with ROM, and so they may not be aware of its benefits to their practice and patients. Time and money are two practical issues that may arise. This type of assessment is not always reimbursed and the average clinician may feel that they do not have enough time to reflect on routine assessment and feedback so as to reconsider their interventions. Agencies may not understand the value of allocating resources to this type of testing (although medically oriented agencies would not hesitate to order a blood test or an x-ray). Finally, some therapists might experience ROM as intrusive, as impeding the therapeutic relationship, and as a means for an agency to control therapist decisions.

Practice Implications

Routine outcome monitoring (ROM) has clear benefits to patients, therapists, and agencies. To overcome barriers, therapists and agencies could implement ROM as part of routine clinical care, and advertise this as an evidence-based practice that will benefit prospective patients. Clients generally appreciate knowing that they will receive the best possible care. ROM can enhance the therapeutic relationship if it is presented to clients as a collaborative endeavor. For example, if a patient is not improving or is deteriorating, therapists can discuss this with patients as well as a plan to alter aspects of the treatment in order to improve the prospects for a better outcome. Therapists can choose from a number of possible ROM options to best tailor the approach to their clients based on cost, time, and relevance. Currently, there are several outcome monitoring systems available to clinicians including: the Partners for Change Outcome Management System (PCOMS), the Treatment Outcome Package (TOPS), the Clinical Outcomes in Routine Evaluation (CORE), and the Outcome Questionnaire (OQ) system.

Safran , J. D. & Kraus , J. (2014). Alliance ruptures, impasses and enactments: A relational perspective. Psychotherapy, 51, 381-387.

In this clinically oriented review, Safran and Kraus discuss the evidence related to alliance ruptures, repairing alliance ruptures, and methods of training in alliance rupture repair. Safran’s work represents “second generation” research on the therapeutic alliance. The therapeutic alliance refers to the relational bond between client and therapist and their agreement on tasks and goals of therapy. A positive alliance is associated with good client outcomes across a variety of therapeutic approaches. Therapeutic alliance ruptures in psychotherapy are inevitable, such that the alliance is continually being re-negotiated, both implicitly and explicitly, throughout the therapy. Such ruptures might include strains, tensions, or breakdowns that could interfere with the ongoing collaboration between therapist and client. Ruptures are associated with re-enactments of dysfunctional relational patterns, but they also may provide opportunities for change and growth in therapy. Safran’s model of alliance ruptures and repairs sees the processes in the client-therapist relationship as key to understanding the client’s relationship problems. Collaboratively addressing tensions in the alliance allows the client to develop more flexible ways of being in relationships and of experiencing themselves. Research by Safran and Muran (2000) suggest that it is rare not to have some minor strain occurring in the therapeutic alliance. Ruptures may occur in half of therapy cases within the first six sessions. Research indicates that unresolved ruptures are associated with deterioration in the alliance, poor outcome, and patients dropping out. In a meta-analysis, repairing alliance tensions by using evidence-based strategies was associated with improved patient outcomes and the effect was large. Alliance ruptures occur across theoretical orientations. For example, research on cognitive therapy showed an improvement in therapist-client interpersonal processes after therapists were trained in techniques to resolve alliance ruptures.

Practice Implications

Alliance ruptures can range in intensity from minor tensions to major rifts in collaboration. They may occur at any time in treatment, and may be present in single or across multiple sessions. Safran and Kraus describe two general types of ruptures. First, withdrawal ruptures occur when clients deal with ruptures or misunderstandings by falling silent. The resolution may involve the therapist exploring the client’s interpersonal fears, reasons for inhibiting negative feelings, and providing the client with an opportunity to communicate their needs. Second, confrontation ruptures occur when clients directly express anger, resentment or dissatisfaction with the therapist or therapy in a blaming manner. The resolution may involve the therapist empathically engaging with the client to facilitate feelings of disappointment, hurt, and vulnerability. Key to this process is the therapist’s meta-communication or mindfulness abilities. The therapist must be aware of the behavior associated with the rupture, collaboratively explore the rupture experience, help the client overcome avoidance of feelings related to the rupture, and explore the client’s needs and wishes that emerge while working through the rupture.

Hoglend, P. (2014). Exploration of the patient-therapist relationship in psychotherapy. American Journal of Psychiatry, 171, 1056-1066.

In this overview of patient-therapist relationship factors, Per Hoglend reviews research on transference in psychotherapy. He argues that transference and transference work is a specific technique that focuses on exploring the patient-therapist relationship. Hoglend takes a broad definition of transference as: the patient’s pattern of feelings, thoughts, perceptions, and behaviors that emerge in the therapeutic relationship and reflect the patient’s personality functioning. Hoglend also defines transference work as any therapist intervention that refers to or explains the patient’s experience of the therapist and their interaction. These interventions include the therapist: (1) addressing transactions in the patient-therapist relationship; (2) encouraging exploration of feelings and thoughts about the therapy or therapist; (3) encouraging the patient to discuss how he or she believes the therapist might feel or think about the patient; (4) including him or herself in interpreting the patient’s dynamics; and (5) interpreting repetitive interpersonal dynamics and linking these to the therapy relationship. More than 30 studies have been published on providing empirical evidence for the relationship between transference work in psychotherapy and positive patient interpersonal outcomes. Effect sizes of the association between transference work and patient outcomes tend to be large. Some of the research indicates that low frequency of transference interventions is useful, but that a higher frequency may lead to negative effects on the patient. Research on transference-focused psychotherapy indicates that it is as effective as dialectical behavior therapy and supportive psychotherapy for borderline personality disorder, but that transference-focused therapy produced better outcomes for attachment related functioning like mentalizing. In the First Experimental Study of Transference Work (FEST), Hoglend found that patients with low quality of object relations (i.e. a poorer ability to maintain close relationships and to regulate affect) benefited most from transference focused therapy. However, those with high quality of object relations did not require the transference work to get better. Also, women responded better to transference work than men. There are some studies of therapeutic approaches like cognitive behavior therapy, in which patients with depression had better outcomes when the patient-therapist relationship was explicitly discussed.

Practice Implications

Hoglend argues that transference work in psychotherapy is an active ingredient that can lead to specific change in some patients. Most studies that Hoglend reviewed showed significant and large associations between transference work and interpersonal changes in patients. Exploring the patient-therapist relationship appears to be most useful for female patients, those with difficult interpersonal relationships, and those with more severe personality pathology. Patients with more mature relationships may not benefit as much from transference work. Although generally effective, if transference work is used too frequently in a session it can also lead to negative patient outcomes.

Tracey, T.J.G., Wampold, B.E., Lichtenberg, J.W., & Goodyear, R.K. (2014). Expertise in psychotherapy: An elusive goal? American Psychologist, 69, 218-229.

As I have reported many times in this blog, there is substantial evidence for the efficacy of psychotherapy. However, the quality of psychotherapy differs across therapists – that is, some therapists achieve better client outcomes than others. Tracey and colleagues (2014) ask: is it possible to demonstrate expertise in psychotherapy? They define expertise as “increased quality of performance that is gained with additional experience”. Professions that can demonstrate expertise include: astronomers, test pilots, chess masters, mathematicians, and accountants. But several professions may not demonstrate expertise, including: psychiatrists, college admissions officers, court judges, personnel selectors, and psychotherapists. The difference is that the former group has predictable outcomes and has access to quality feedback. In addition, Tracey and colleagues argue that psychotherapy lacks adequate models for how interventions produce benefits. As a result, adherence to treatment protocols (i.e., manuals) is not reliably associated with better patient outcomes. Further, more experienced therapists are not more effective than less experienced therapists. Experienced therapists might have more complete conceptualizations of client problems, but these conceptualizations may not be accurate. Finally, although therapists affect outcomes, client variables (e.g., motivation, severity of symptoms, expectations) likely explain the largest proportion of outcome variance. Tracey and colleagues argue that part of the problem is that psychotherapists do not engage in “deliberate practice”; that is, practice of a specific task (e.g., identifying a rupture in the alliance), receiving specific feedback (e.g., that a rupture was not identified), opportunity for repetition (e.g., to identify a subsequent rupture in the alliance), and opportunity for improvement afforded by error (e.g., better able to identify a future rupture and repairing that rupture). Generally the practice of psychotherapy provides little feedback about the accuracy of past clinical decisions. In other words there is a lack of quality information to help therapists develop into experts. Further, for a whole host of reasons, psychotherapists are notoriously poor at assessing client progress (i.e., like other humans, therapists engage in a number of biased evaluations of their performance). Quality information might be available from progress monitoring (i.e., continuous feedback to therapists about client outcomes), which has been shown to improve client outcomes. However, this may not aid therapists in developing expertise, since progress monitoring provides little information about what therapist behaviors are necessary to improve performance and client outcomes.

Practice Implications

Tracey and colleagues conclude that currently psychotherapy does not provide evidence that it is a profession with expertise. To achieve expertise, therapists need quality information not only about their patients’ outcomes but also about their own average outcomes (i.e. performance) relative to other therapists working with similar clients. And therapists need information on how to manage specific events in psychotherapy. Tracey and colleagues suggest therapists set aside time to generate hypotheses about one’s practice that can be disconfirmed, and then test these hypotheses. For example, if a therapist is experiencing a higher than average number of premature client terminations (which may follow a misunderstanding with the client), the therapist may hypothesize that he or she is not identifying key alliance ruptures. To test this hypothesis, the therapist could repeatedly assess the alliance (with a validated instrument) with some clients, use this information and not clinical judgement alone to identify alliance ruptures (i.e., a week to week severe downward trend in alliance scores), and implement an intervention to repair the alliance with these clients. Do clients with whom a therapist has implemented this procedure drop out at a lower rate? Does this process of deliberately identifying alliance ruptures and repairing them lead to enhanced therapist performance regarding alliance ruptures? This form of deliberate practice (testing disconfirmable hypotheses based on quality information) might lead to greater expertise in identifying alliance ruptures.

Handbook of Psychotherapy and Behavior Change: Starting in March 2013 I will review one chapter a month from the Handbook of Psychotherapy and Behavior Change in addition to reviewing psychotherapy research articles. Book chapters have more restrictive copy right rules than journal articles, so I will not provide author email addresses for these chapters. If you are interested, the Handbook table of content and sections of the book can be read on Google Books.

Crits-Christoph, P., Connolly Gibbons, M.B., & Mukherjee, D. (2013). Psychotherapy process-outcome research. In M.E. Lambert (Ed.), Bergin and Garfield’s Handbook of Psychotherapy and Behavior Change, 6^th Edition (pp. 298-340). New York: Wiley.

In their chapter in the Handbook, Crits-Christoph and colleagues (2013) review research in which psychotherapy processes are related to patient outcomes. I reported in the July 2013 PPRNet Blog that therapeutic alliance is reliably correlated with treatment outcomes in a variety of disorders and treatment types. Alliance refers to an agreement on tasks and goals of therapy, and the bond between therapist and client. The common assumption is that alliance is a necessary condition that in part causes change in client symptoms. However therapeutic alliance studies tend to be correlational, that is, the studies show a relationship but the study designs do not allow one to say that alliance causes good outcomes. What if the opposite were true; what if early experiences of symptom reduction caused the therapeutic alliance to improve? If that were the case, then alliance would be an artificial and not particularly important aspect of psychotherapy. Crits-Christoph and colleagues (2013) review the literature on this topic. Some studies of cognitive behavioural therapy (CBT), for example, found that prior change in symptoms predicted later therapeutic alliance, but prior alliance did not predict later symptom change. In a more sophisticated study, Crits-Christoph and colleagues (2011) found that previous change in the alliance was related to later change in outcomes, but not vice versa. In the same study, the authors noted that measuring patient alliance at a single early session accounted for only 4.7% of the outcome variance at post treatment, whereas averaging assessments of alliance across 6 early sessions accounted for almost 15% of the outcome variance. In other words, averaging assessments across many sessions produced a more dependable measurement of alliance. Several studies now report a reciprocal relationship between alliance and outcome, indicating that change in alliance and change in outcomes across therapy sessions progress in a mutually reinforcing spiral. That is, early change in alliance causes subsequent change in outcome, which in turn results in further change in alliance, which precipitates more change in symptoms, etc. The review by Crits-Christoph and colleagues (2013) also noted that the importance of alliance seems to be greater for patients with a disorder like depression, compared to anxiety disorders.

Practice Implications

Developing an early alliance with a client is related to treatment outcomes. Measuring alliance repeatedly (not just once) will give the best indicator of the state of the therapeutic relationship. Patients and therapists who have a genuine liking for each other, who agree on how therapy will be conducted and on the goals of therapy will improve the chances that psychotherapy will be successful. Alliance and symptom change may work together throughout therapy so that improvement in one will cause change in the other on an ongoing basis across therapy sessions. Alliance may be particularly important for patients with depressive disorders that are characterized by isolation from others, loneliness, and low self esteem.