Commentary

In theory, CBT sounds right up my alley. I love the idea of leaning on rationality as a tool to navigate daily life successfully. I love challenging ideas that sound irrational, subjective, outdated, or just too-good-to-be-true. Navigating the world gets so much more comfortable for me when I believe that I can prepare a rationality toolkit that will slice through the parade of bullshit we’re all subjected to on a nonstop basis.

In practice, CBT seems to have become a gold standard or a silver bullet, and it’s impacted the culture of mental health in a way that has little, if anything, to do with the theory.

By the early 1990s, CBT had become the dominant approach to psychotherapy for depression. While there’s a strong base of evidence that supports that CBT itself has positive outcomes, there hasn’t been evidence to suggest that challenging cognitive distortions or correcting cognitive dysfunction was the reason. In short, we knew it helped but didn’t know how or why.

In 1996, Niel S. Jacobson ran an experiment that isolated components of CBT and studied their impacts. The first group would undergo behavioral action, the second group combined it identifying automatic thoughts, and the third group received the complete CBT protocol that involved cognitive restructuring. All three groups showed substantial improvement, and the third group was not meaningfully better than the first group. Many researchers at this point stepped into Behavioral Activation as its own modality independent of the rest of CBT. Other researchers moved away from examining its underlying mechanisms and into analyzing its effectiveness.

Steven D. Hollon and Robert J. DeRubeis continued studying CBT mechanisms, but it became difficult to find consistent patterns in those suffering from depressive symptoms undergoing CBT. Sometimes cognitive changes came before improvement in symptoms, but sometimes the symptoms improved before cognitive changes, and sometimes they both changed at the same time. This made it difficult to claim that challenging cognitive distortions was consistently improving depression.

Nevertheless, there grew an industry of CBT manuals targeting all kinds of mental health illnesses and disorders: panic disorders, generalized anxiety disorder, social anxiety disorder, PTSD, OCD, substance use, etc. Research had empirically validated many of these models, but the problem of excessive manualization emerged. CBT is really easy to study compared to other treatments, which meant it had more evidence, which meant it attracted more funding, which meant it had more evidence, which meant it attracted more funding… so on and so forth. However, having more evidence doesn’t mean other therapies are less evidence-based if the evidence is still high quality, such as peer-reviewed publications.

Ronald Kessler and colleagues in the research space ran the National Comorbidity Survey (NCS) which found that people with one mental health disorder very frequently presented with others. This presented some complications, since the manuals had mapped treatment protocols to individual diagnoses as presented by the DSM. If a patient came in with multiple co-occurring diagnoses, a therapist might find themselves combining multiple manuals instead of applying the fundamentals that worked across them all. Barlow sought to create a Unified Protocol in 2004 that did exactly that, and while it didn’t stick in the long run, other approaches to CBT that transcended diagnoses and manuals kept developing.

Jacobson and Dimidjan led studies into Behavioral Activation separate from Cognitive Therapy as a treatment protocol. In a 2006 publication, on patients with Major Depressive Disorder (MDD), BA did as well if not better than CT especially for the severely depressed subgroups. In fact, BA was comparable to antidepressant medication in how effective it was.

The same year, Hayes et al published a review of Acceptance and Commitment Therapy (ACT), which challenged the conventional thinking about changing cognition by arguing for an expansion of the concept of cognition. Rather than rational or irrational beliefs related to cognitive distortions that required restructuring, ACT was aimed at rigid internal organization around a belief regardless of its rationality or truth value. A patient with improve mental flexibility and a firm value system might be able to accept unwanted experiences or feelings that were previously avoided. Empirical results again showed that it was comparable in effectiveness to CBT, underscoring that changing the content of thoughts was not the primary mechanism for improvement.

In 2009, Hollon & DeRubeis reviewed the literature around CBT and found that, although its efficacy was still established, there was weak evidence to identify the mechanisms of CBT that lead to improved outcomes. They found lots of methodological issues in studies that were trying to pinpoint those drivers. The year after, we see some sparks fly. Lynch, Laws, & McKenna (2010) featured sharp criticisms of the literature up to that point as being overly reliant on comparing CBT to weak control groups like patients on waitlists. They argued that any intervention can look impressive relative to nothing instead of a comparison to established psychotherapy, and they pointed out that meta-studies had conflated both types of controls when describing the significance of CBT’s outcomes.

Specifically, they found a statistically significant difference between blinded and non-blinded studies for CBT for schizophrenia, implying that observer bias from researchers was substantially inflating the results. There was a small improvement in acute depression symptoms from CBT rather than the moderate-to-large effects some studies claimed, but the main benefits were in preventing relapses into depressive episodes. They found no statistically significant evidence that it was effective for those with schizophrenia or preventing episodes from bipolar disorder.

To be clear, Lynch et al were not trying to say that CBT is ineffective. They were clarifying that the literature so far can give the impression it’s far more effective than it really is given how heavily manualized and institutionalized it was becoming. Farchione et al 2012 gives us an empirical attempt at Barlow’s 2004 proposal for a transdiagnostic unified protocol of CBT that set aside the DSM’s diagnoses as a starting point for manuals and interventions.

The last two decades have featured a lot of debate over this, and I’m still in the process of sorting through the avalanche of studies and the consensus around them. Part of the problem is that studies try to answer very specific questions while the debate about protocols demands generalization. It is true that CBT is one of the most evidence-backed therapies, but it is also true that CBT’s effectiveness has been overstated when other therapy modes seem just as effective if not more so depending on the patient. It is true that CBT studies have had lots of issues in methodology, but it’s also true that CBT has generally held up as improving a patient’s outcomes in the follow-up studies with better methodologies.

One of the most vocal critics of the dominance of CBT over other therapies is Jonathan Shelder. A 2018 paper hosted on his website (and included in the links below) has some incisive critiques about the research and therapeutic communities. It’s a paper written in plain enough language for non-academics and non-professionals to read. Shelder is very opinionated on how CBT gets marketed as “evidence-based” or a “gold standard” when we also have evidence for psychodynamic therapies that are excluded from those labels. He doesn’t go as far as to say CBT is wrong or bad as theory in this paper, but he does criticize the widespread use of manualized therapy for short-term symptom reduction method as convenient for institutions rather than the outcome of good science and health policy.

Even ardent CBT advocates have acknowledged that manualized CBT offers lasting help to few. Writing in Psychological Science in the Public Interest, eminent CBT researcher Steven Hollon noted “Only about half of all patients respond to any given intervention, and only about a third eventually meet the criteria for remission… . More over, most patients will not stay well once they get better unless they receive ongoing treatment.”2 Ironically, this was written by the same researcher who declared other forms of psychotherapy “inadequate.” Sadly, such information reaches few clinicians and fewer patients. I wonder what the public and policy makes would think if they knew these are the same treatments described publicly as “evidence-based,” “scientifically proven,” and “the gold standard.”

Lest the reader think this a misrepresentation or caricature, consider that prominent proponents of manualized therapy advocate treatment by minimally trained paraprofessionals. The journal Psycho logical Science in the Public Interest, the house organ of the Association for Psychological Science, published the following: “Many of these [evidence-based] interventions can be disseminated without highly trained and expensive personnel… . CBT is effective even when delivered by nondoctoral ther apists or by health educators with little or no prior experience with CBT who received only a modest level of training … manuals and workbooks are available on user-friendly websites.”22 The devaluation of clinical expertise inherent in these statements requires no interpretation

The researchers examined studies that claimed to compare an “evidence-based” therapy with an alternative form of psychotherapy. The researchers examined more than 2500 abstracts. After closer examination, they winnowed that down to 149 studies that looked like they might actually compare an “evidence-based” therapy with another legitimate form of therapy. But when they finished, there were only 14 studies that compared “evidence based” therapy with a control group that received anything resembling bona fide psychotherapy. These studies showed no advantages whatever for “evidence based” therapies.

From 2500 to 14 actual comparisons between “evidence based” manualized therapy and other psychotherapy. That speaks to how easy it is to run a CBT study compared to a psychotherapeutic alternative. In fact, Shelder proposes that one of the reasons CBT can lead to improvements is the implementation of techniques that come from the domain of psychodynamic therapy. Another argument proposes that the consistent effectiveness of most psychotherapies has less to do with a particular mode. Instead, there are common factors shared by all therapeutic interventions that explain more of the variation in outcomes than a particular modality.

One of these is the therapeutic relationship that forms the shared understanding between the patient and therapist, and forming a trusted relationship can’t be standardized and commodified. There is no manual to relationships. Yet, achieving that relationship can be a more significant factor in the outcome of therapy than whether a CBT or alternative psychodynamic framework was used. Shelder points out that means therapists should be trained like apprentices who learn from experienced professionals modeling successful relationships rather than mass produced through manualized training modules.

What was even more telling than the numbers was this footnote:

Lest the reader think this a misrepresentation or caricature, consider that prominent proponents of manualized therapy advocate treatment by minimally trained paraprofessionals. The journal Psychological Science in the Public Interest, the house organ of the Association for Psychological Science, published the following: “Many of these [evidence-based] interventions can be disseminated without highly trained and expensive personnel… . CBT is effective even when delivered by nondoctoral therapists or by health educators with little or no prior experience with CBT who received only a modest level of training … manuals and workbooks are available on user-friendly websites.”22 The devaluation of clinical expertise inherent in these statements requires no interpretation

As a patient, you may not want to see that CBT can be delivered by people with “little or no prior experience with CBT who received only a modest level of training” or that you received a worksheet from materials “available on user-friendly websites.” So we can surmise that this benefit is being marketed to institutions looking to cut the costs of “highly trained and expensive personnel.” Whether Shelder is too harsh on CBT and the literature supporting its efficacy, it’s not hard to see that there are incentives within the system that are not necessarily aligned with optimizing patient outcomes.

So let’s do a quick recap of the state of CBT research we’ve touched on so far. Shelder challenges us to look at the institutions that adopt therapeutic models to see the influence of their incentives in the quality and delivery of healthcare. Wampold stepped back from the specifics of modalities to study the common factors across all therapies that contribute to success more than the choice of treatment. Cuijpers compares CBT to other credible and evidence-back psychotherapies like the psychodynamic talk therapy Shelder defends. Hayes and Hofmann consider the merits of CBT moving towards processes and away from diagnostic labels from the DSM. Hollon and DeRubeis attempt to figure out the specific mechanisms within CBT that accounts for successful improvements, particularly in exposure therapy where it has some of the biggest effects on patients.

You might be thinking, “Wait, Mank, what’s the actual problem? If CBT can genuinely help people to some extent, can be tried quickly through a standardized intervention, and has a lot of material easily accessible, it sounds like a great first line treatment that institutions can quickly roll out. If it doesn’t work, a therapist can use the relationship formed with the client in the process of CBT to try something else, right?”

Right, much of the time that is exactly the case. Not all of the time. One thing that gets swept under the rug when discussing average metrics is that some people get worse, just like with any treatment.

Michael Lambert conducted research into this in the 1990s and 2000s and found that continuous monitoring of deterioration in patients undergoing active treatment can help therapists adjust to slow or reverse it. This showed two things: deterioration is a detectable pattern, and deterioration can be addressed if caught. Lambert’s work estimated that 5-10% of a population of clinical patients in active therapy were deteriorating, but that could be halved if the therapist was aware of it. Interestingly, in randomized controlled trials by researchers, only 6% reported deterioration at all and the median rate was 4%. This suggests some more support for Shelder’s claim that many studies toss out complex cases that clinicians meet in the real world. Unfortunately, this is still not studied enough for us to have more evidence-based or data-driven insight into the types of populations at highest risk for deterioration, but we have some clues we can look at.

Marsha Linehan developed DBT in response to the observation that CBT could lead chronically suicidal patients with borderline personality disorder into more negative outcomes. Discussing cognitive restructuring or distortions with these patients could inadvertently lead to the message that their experiences with strong emotions is fundamentally wrong. This could lead to feelings of shame and self-blame that impacts engagement in treatment, dropout rates, ruptures in therapeutic relationships, and more emotions that prove difficult to regulate, which forms a kind of vicious cycle. Linehan specifically included validation as a foundational premise for DBT to address that.

In 2007, Scott Lilienfield published Psychological Treatments That Cause Harm which discussed the historical reluctance from the psych research community to examine adverse consequences of therapeutic interventions instead of assuming evidence-based treatments with appreciable improvements in the average case are neutral at worst.

This page is WIP.

MankNotes

Explorer

Cognitive Behavioral Therapy for Substance Use

Commentary

Links

Graph View

Table of Contents

Backlinks