Clinical Practice Guidelines

This post is co-authored with Scott Lilienfeld

Recently, the American Psychological Association (APA) released clinical practice guidelines for the treatment of posttraumatic stress disorder, or PTSD (Courtois, et al., 2017). After carefully surveying the available research, the developers of these guidelines recommended several treatments for trauma-related psychopathology, namely cognitive-behavior therapy, cognitive processing therapy, cognitive therapy, prolonged exposure, and less strongly, brief eclectic psychotherapy, eye movement desensitization and reprocessing, and narrative exposure therapy. To individuals familiar with different approaches to psychotherapy, this is a diverse range of therapies, surely with something that most practitioners could find acceptable. As the guidelines also make clear, no recommendations are made about some of the most widely practiced types of therapy (particularly dynamic and experiential approaches), not because we know that they do not work but because they have not been adequately tested. “Absence of evidence is not evidence of absence,” but it does make it difficult to make claims of efficacy based on anecdote, clinical supposition, and clinical experience alone. These three sources of evidence can be enormously useful in generating hypotheses, but as historians of medicine have long noted, they can readily lead to erroneous inferences of treatment effectiveness.

The development of clinical practice guidelines represents a significant advance for psychotherapy practice. Consumers benefit as they can identify recommended interventions when seeking treatment. Practitioners benefit as the guidelines offer specific instructions to assist them in selecting appropriate interventions for their clients. And finally, psychotherapy at large benefits as the guidelines place clinical practice on a comparable plane as that in other health care professions, where standards of practice are routinely developed and revised in accord with the latest scientific evidence.

In the psychotherapy domain, some individuals have recently criticized practice guidelines as inadequate or even harmful. In one particularly telling illustration published on the Psychology Today blog, Shedler (2017) criticized the newly released practice guidelines for PTSD. He raised particular concerns with randomized controlled trials (RCTs) and the problem of potentially harmful effects of PTSD therapies that are presently regarded as empirically supported. .

In the spirit of Rapaport’s rules for appropriate argumentation (see https://www.brainpickings.org/2014/03/28/daniel-dennett-rapoport-rules-…), we begin with several points of agreement with Shedler. First, we concur with him that RCTs are not the only source of evidence that bears on the efficacy of treatments. There are numerous other research methodologies to evaluate treatments, several of which can be helpful in ascertaining which of them work well. Second, we agree with Shedler that RCTs are not needed to address all questions in science (although we are unaware of any serious scholar who has argued that this is so). Third, we agree with Shedler that RCTs, like all research designs, can be poorly conceptualized and ineptly implemented. In all other respects, though, we part ways with Shedler.

Shedler’s comments neglect a critical point, namely, that RCTs, although not panaceas, eliminate more sources of error in psychological inferences - such as placebo effects, the tendency of conditions to improve on their own accord, and the tendency of observers to inadvertently attend to outcomes that confirm their hypotheses - than do other sources of research evidence, including naturalistic designs and accumulated clinical experience. Put somewhat differently, RCTs are imperfect but crucial safeguards against human error and both researcher and practitioner hubris (Lilienfeld, Ritschel, Lynn, Latzman, & Cautin, 2014). The histories of medicine and psychology teach again and again that without RCTs, we can be led to conclude erroneously that ineffective or even harmful interventions are beneficial. In these respects, Shedler’s critique is surprisingly nonpsychological and ahistorical.

As we have often found with critiques of clinical practice guidelines in particular, and empirically supported treatments generally (e.g., Shean, 2016), Shedler misrepresents the available evidence. To illustrate his point that clinical practice guidelines are based on faulty data, he discusses research showing that flossing does not lead to significant improvements in dental health. He maintains that for dental researchers to demonstrate the efficacy of flossing, they would need to carry out a RCT spanning many years or decades, with one group flossing and the other not over the entire time of the trial and thereby violating ethics given the long-term risk of harm to the control group. Nevertheless, this comparison fails because in psychotherapy, psychiatric symptom reduction can often be established in a far shorter time period. His complaint applies only if one were to test his favored approach, psychodynamic psychotherapy, against a comparison treatment. According to Shedler (2010), the most efficacious approach to treatment emphasizes the therapeutic relationship and de-emphasizes or ignores altogether any active or directive interventions. The challenges inherent in relying solely on a therapeutic relationship as a mechanism of change has been discussed at length elsewhere, but this approach effectively oversimplifies therapy and leaves practitioners with no additional recourse should that alone fail to provide relief to the client (McKay, 2011; Pilecki & McKay, 2016).

Shedler misrepresents the science process more generally, maintaining that “hard” sciences such as physics, chemistry, and astronomy do not make use of RCTs. Nevertheless, this point is irrelevant and incorrect. As far as we are aware, there are no astronomy-based health care interventions, save perhaps for the public health recommendation that eclipse watchers avoid staring directly at the sun. There are, however, numerous interventions based on chemistry, as the entire field of pharmacotherapy is based on developing and testing chemical compounds for later drug development purposes. Tests of the effectiveness of these compounds are based, at least in part, on pre-clinical studies (mostly experiments) and the subsequent findings from RCTs.

In criticizing the clinical practice guidelines, Shedler complains about the recommendation of CBT, which he regards as emanating from a clear bias toward this treatment. His assertion is patently incorrect and reflects a highly selective reading of the guidelines. If one reads only the abstract of the guidelines, it is immediately clear that these guidelines endorse a wide range of approaches, some of which do not fall within the traditional CBT umbrella.

It is also worth noting that, as the APA was preparing to release its first practice guideline, a specific aim was to highlight putative harmful interventions (Hollon et al., 2014; Lilienfeld, 2007). At the present time, few psychological interventions are labeled as harmful, an omission that implies erroneously that at worst an intervention is benign. This cannot be a tenable proposition, as failure to improve alone can often be construed as harm from treatment (Dimidjian & Hollon, 2010).

Health Essential Reads

Taking Health Literacy to the Next Level

Are Women Really Better Doctors?

In Shedler’s zeal to criticize CBT in general, and exposure treatments in particular, he further argues that treatment trials of trauma are marked by unduly high dropout rates. This critique is erroneous on two counts. First, recent evidence not cited by Shedler suggests that PTSD dropout rates are no higher for CBT in general and exposure in particular than for other psychotherapies (Imel et al., 2013). Second, treatments for many medical conditions are similarly marked by high dropout rates. Nevertheless, these rates have resulted not in changes in the treatment guidelines but rather in adjustments in how the treatment is applied to maximize compliance. For example, although cardiac rehabilitation is considered essential following bypass surgery, dropout from this post-operative treatment is as high as 41% (Yohannes, et al., 2007). As another example, individuals with diabetes frequently have poorly managed symptoms and high dropout rates from interventions (Shrivastava, et al., 2013). In each of the aforementioned medical examples, the remedy is not to discard the treatments, but instead to identify methods for improving compliance, such as through health psychology interventions. That is, the approach involves ‘this and that’, rather than the ‘this, not that’ advocated by Shedler, an error that embodies the familiar “either-or” logical fallacy (see https://psychologenie.com/explanation-of-either-or-fallacy-with-examples). What serious scholar has suggested that a relationship with a client is irrelevant to treatment when applying prolonged exposure? Shedler presents his argument as if the relationship between client and therapist is unnecessary when conducting exposure, when in fact competent practitioners from all schools of therapy focus on maintaining a sound relationship (Taylor, Abramowitz, & McKay, 2010).

We welcome constructive criticisms of psychotherapy methodology and of the clinical practice guidelines. Both can and will benefit from thoughtful feedback. Nevertheless, Shedler’s recent commentary is critical, but it embodies a lack of critical thinking. He lays out a variety of straw person arguments and engages in a highly selective reading of the practice guidelines and the research evidence undergirding them. . He encourages practitioners to ignore the guidelines, but given that these guidelines are based on the best available research evidence – imperfect as it necessarily is – practitioners would be wise to ignore his nonpsychological and ahistorical recommendations.

The authors are, respectively, President-elect and President of the Society for a Science of Psychological Science (SSCP), section 3 of Division 12, the Society for Clinical Psychology, within APA. The views expressed in this article are those of the authors and do not necessarily represent those of the SSCP or APA.