Psychotherapy Research

Research on psychotherapy has taken many different directions over time, as discussed in my previous blog. Although a bit of an oversimplification, most outcome research tends to fall into one of two camps: 1) Empirically Supported Treatments and 2) Common or Contextual Factors Research. I will discuss each approach briefly analyzing the strengths and weaknesses of each. 

The focus of the Empirically Supported Treatment, at its heart, is to evaluate the effectives of a narrowly defined treatment approach. This is a noble effort. First, it provides information that gives the appearance of being rather clear and straightforward. Oftentimes, it facilitates the development and improvement of very structured therapy approaches that are fairly easy to follow. These approaches often are very appealing to insurance companies and managed care, allowing psychotherapy to have an inroad in to reimbursement options. 

However, this approach has too often attempted to become something more than it is, leading to several major problems. First, this approach has often gone down the path of comparing psychotherapies trying to 'prove' which approach is best. I have concerns about the motivations, utility, and philosophy of science of such attempts. Researchers often are rooted in one of the therapy approaches reflecting a particular philosophical system and design or utilize measures that are best suited for the particular approach to therapy. Stated more clearly, different outcome measures are better suited to measure different types of therapy. 

Let me provide a brief example. The Beck Anxiety Inventory (BAI) and the State-Trait Anxiety Inventory (STAI) are two common measures of anxiety, both reflecting different theories. The BAI focuses mostly on one's perceptions of anxiety (i.e., cognitions) and little on visceral symptoms (i.e., bodily symptoms). The STAI, conversely, focuses more on visceral or bodily symptoms. The BAI is typically used in evaluating the effectiveness of cognitive behavioral therapy (CBT), which makes logical sense in that in focusing on defining anxiety primarily by cognitions and perceptions. In CBT therapy, the therapist is working on helping the individual change cognitions and the BAI evaluates this. However, it assumes that cognitions and experience will be equivalent. This creates a few problems, such as demand characteristics, in which clients are encouraged to change their cognitions in a direction with the assumption that the accompanying emotion will follow. This is not always the case. Sometimes cognitions change, the BAI results change, and the experience does not. Other forms of therapy focus on targeting the visceral symptoms more directly, without being as concerned about the cognitions. This therapy could potentially be effective by the STAI and not the BAI. In other words, as soon as an outcome measure is introduced, so is a bias. 

Studies rooted in the Empirically Supported Therapies tradition tend to emphasize a fairly controlled research setting, which also introduces limitations. This approach is biased to more structured, solution-focused therapies and against the more fluid, relational depth psychotherapy approaches. However, even many solution-focused practitioners point out that the therapy conducted in these research studies often do not reflect therapy in the real world. It takes the relationship out of even the solution focused approaches. In particular, therapists often follow structured guidelines on how to intervene at a particular point in treatment and are discouraged from following clinical judgment and experience about when adaptations are necessary. 

The controlled aspect introduces another serious limitation. First, careful screening is conducted to assure the client meets the right criteria, which often means not including individuals with complex issues or dual diagnoses. This is a large portion of the population entering therapy! Studies also have to carefully consider which factors to include and control for as potential confounding variables, but cannot include too many factors as if enough factors are controlled for, the natural co-variance across influences would reduce any unique variance to non-significant levels. Variables such as the therapeutic alliance and client motivation, which meta-analytic studies suggest are very important, are generally not included. Thus, it is very difficult to say that it is the techniques, or what is intended to be measured, that is actually being measured. These serious limitations do not suggest we should throw away this type of outcome research, but rather we need to acknowledge there are many very significant limitations. We are only getting a small bit of truth at most in these studies. 

Common Factors Research emerged from broad, critical reviews of the research, meta-analyses of previous research, and new research across therapy approaches into other factors not accounted for in the empirically supported treatment paradigm, such as empathy and therapist factors. This includes many very rigorous research approaches that are better able to take into consideration factors, such as the therapy relationship, that the empirically supported treatment paradigm frequently ignores.

The results are consistent: The two most important factors predicting therapy success are 1) client factors and 2) the therapy relationship. Both of these factors are generally not given much consideration in the empirically supported treatment paradigm! These two factors alone account for most of the therapeutic change (see Humanistic Psychology: A Clinical Manifesto, 2009, by David N. Elkins, for an in depth review of this research). 

One of the most fascinating aspects of this research is regarding technique. In the various meta-analytic studies, the role of technique generally accounts for somewhere between 1.5% to 7% of the variance in therapy outcome. However, even with this, it is not a particular technique that accounts for the variance, but that techniques are used. This means that the efficacy of any particular technique would be much less than the 1.5 to 7%! What accounts for a greater portion of the therapy effectives could be labeled as the therapist and client belief in the therapy modality used. In other words, the therapist believing in what they are doing and being able to inspire hope in their clients is more important than the type of therapy the therapist employs! The moral of this story is to make sure you believe in what you are doing and choose a therapy orientation that fits. From this perspective, we should be less interested in coercing young therapists into a particular orientation, but more interested in helping them find the orientation that is the best fit for them. 

In conclusion, outcome research is complex. Anytime you heard that it is clear and simple, be suspicious. I believe that the most rigorous and complex research on therapy outcome strongly suggests that what we should be most concerned about is 1) client factors, 2) the therapy relationship, and 3) helping therapists find a therapy modality that they believe in. This last factor, although not in the two factors identified by meta-analyses as most predictive of successful therapy outcomes, I think is related. When therapists are doing something they believe in and are passionate about, and are concerned with helping clients find the best fit (as opposed to trying to convince every client they are the best therapy for them), they will build better therapy relationships with their clients. 

Comments

[MP]

Nice post! But still too many researchers and CBT-therapists are convinced that CBT is the only evidence based and the best therapy for each... The myth of CBT and a bias of thinking?

[Focus Factor]

I have looked at many sites on this subject and not

come across a site such as yours which tells

everyone everything that they need to know. I have

bookmarked your site. Can anyone else suggest any

other related topics that I can look for to find out

further information?