When Less is More: Cognitive Behavior Therapy vs Psychoanalysis for Bulimia

American Journal of Psychiatry published a noteworthy report of a randomized clinical trial (RCT) comparing cognitive behavior therapy to psychoanalytic therapy for bulimia.

trophy-lTwenty sessions of cognitive behavior therapy over 5 months reduced binge eating and purging better than 2 years of weekly psychoanalytic psychotherapy. This was true for assessments both at five months (42% versus 6%), marking the ending the cognitive behavior therapy (CBT), and two years (45% versus 16%), marking the ending of the psychoanalytic psychotherapy. Overall, psychoanalytic psychotherapy did not do well, despite the greater intensity of treatment.

If that’s all that you needed to know, you can stop reading here. But continue on if you are interested in finding out more about good conduct and reporting of clinical trials, what’s special about this trial. Then, the next blog post will put this trial into the context of  a larger struggle to secure insurance funding of long-term psychoanalytic or psychodynamic psychotherapy (LTPP)—with exceedingly weak and limited evidence.

What was done in this trial.

The trial was conducted in a University clinic patients recruited through advertisements and referral.

Seventy patients with bulimia nervosa received either 2 years of weekly psychoanalytic psychotherapy or 20 sessions of CBT over 5 months. The main outcome measure was the Eating Disorder Examination interview, which was administered blind to treatment condition at baseline, after 5 months, and after 2 years. The primary outcome analyses were conducted using logistic regression analysis.

Experienced therapists were monitored for their adherence to the model of therapy to which they were assigned.

What was found.

Aside from the primary outcome of reports of binge eating and purging, the secondary outcome was improvement in general psychopathology, measured by standardized measures including self-reported anxiety and depression symptoms. At five months the outcome was better for CBT, but the difference was no longer statistically significant at 2 years.

How the authors explain their results.

CBT is a symptom-focused treatment that is designed to produce a rapid reduction in the frequency of binge eating (10), a change that is highly predictive of the patients’ eventual response (32). In contrast, the psychoanalytic psychotherapy tested in this trial was designed as a nondirective therapy with no specific behavioral procedures directed at the control of binge eating. The more indirect approach to symptoms in psychoanalytic psychotherapy may be insufficient, because binge eating and purging can both be viewed as maladaptive coping strategies that provide an immediate, albeit short-term, relief from negative emotions (5–7). Accordingly, to enable the patient to let go of the symptoms, a directive approach providing concrete alternative problem-solving techniques may be needed.

What was so impressive about this trial.

This trial was designed and reported in ways that substantially reduced risk of bias:

  • Patients randomized to the two therapies were equivalent on key baseline variables.
  • Both treatments were manualized.
  • Authors of this article included developers of both manuals.
  • Both treatments were administered by experienced therapists supervised for their fidelity to the model and manual to which they were assigned.
  • Outcomes were evaluated by raters who were blind to treatment assignment.
  • Analyses were intent to treat, i.e., conducted with all patients who were originally randomized.
  • Timing of outcome assessments were tailored to end of both treatments, conducted at five months when the CBT ended and again at two years two years when the psychoanalytic psychotherapy ended.

Not perfect.

Limited funding for the study left it underpowered to detect the size of differences that are typically found between two established treatments. If there had been only a small difference between the two treatments—usual for comparisons of two credible treatments—the effect would not have been detected as statistically significant.

There was no neutral comparison/control condition. When a trial is underpowered to detect the expected small differences between two active, credible treatments, it’s good to have the fallback of comparing each of the treatments to the comparison/control condition.

For instance, if the difference between CBT in psychoanalytic psychotherapy was too small to achieve significance, secondary comparisons could still be made to determine if either or both were superior to the comparison/control condition. Otherwise, no differences between the two treatments would leave us undecided whether they were equally good or equally bad, compared to the change that would occur in the absence.

The trial was apparently not preregistered. We cannot independently verify whether the final sample was what was originally intended or that the analysis were as planned before the data were available.

Most patients receiving neither therapy had reduced binge eating and purging, but that apparently was not improved by the considerably larger number of sessions of psychoanalysis versus cognitive behavior therapy.

And it is only one study.

The trial is noteworthy for a number of reasons.

It is a rare head-to-head comparison of psychoanalytic therapy to another credible psychotherapy under conditions which proponents of both treatments would agree are fair.head to head

Proponents of psychoanalytic psychotherapy and mainstream empirically-based therapies have great difficulty agreeing how to conduct a study in terms of characterizing patients, length of treatment, and selecting outcomes. The divide is great. Proponents from one camp can typically object to the other’s recruitment and diagnosis of patients, the manner in which psychotherapy is implemented, and for what length of time.

Proponents of psychoanalytic psychotherapy often question whether the randomized control trial is even appropriate for its evaluation. For instance:

The thesis advanced here is that the privileged status this movement accords such research as against in-depth case studies is unwarranted epistemologically and is potentially damaging both to the development of our understanding of the analytic process itself and to the quality of our clinical work. In a nonobjectivist hermeneutic paradigm best suited to psychoanalysis, the analyst embraces the existential uncertainty that accompanies the realization that there are multiple good ways to be, in the moment and more generally in life, and that the choices he or she makes are always influenced by culture, by sociopolitical mind-set, by personal values, by countertransference, and by other factors in ways that are never fully known.

The comparative psychotherapy literature consists of a few studies in which a reader can readily predict the outcome of comparisons of psychoanalytic psychotherapy versus other treatments between by simply looking at investigator allegiance. Yup, which therapy will produce the largest effect is better predicted by which treatment the investigator advocates, not the particular brand of therapy.

It is highly unusual in finding that any credible psychotherapy has a substantial advantage over any another.

Credible psychotherapies typically acquire their evidence-based status in randomized trials in which they are compared to wait list, no treatment, or a routine care that has not been shown simply inadequate care or no care at all. The evidence mustered to show particular psychotherapies are effective does not typically address whether they are effective versus other credible treatments.

There are also few instances of one credible treatment besting another in a head-to-head comparison. Where this does occur, it is typically due to an investigator allegiance effect.

It is just the kind of comparison that is needed to address the important question of whether investing time and money in longer-term psychoanalytic psychotherapy leads to substantially better outcomes than shorter psychotherapies.

In this single, modest, even if well done trial, a lot fewer sessions of CBT produced greater change than considerably more sessions of LTPP. In terms of calculating cost effectiveness, the less expensive CBT treatment needs to be interpreted in the context of its greater efficacy. In this instance, less and cheaper is better.

As we will see in my next blog post, the question of whether the added expense of LTPP can be justified has been the subject of meta analyses that were wretchedly done with almost no quality evidence. The meta-analyses are nonetheless heavily promoted by those who seek to secure insurance coverage of LTPP.

6 thoughts on “When Less is More: Cognitive Behavior Therapy vs Psychoanalysis for Bulimia”

  1. Cognitive behavior therapy is great tool to use in the right moment, I use it also when i am treating some of my own paitents.

    The lees is more idealogy is great – but I also thing it is important to remember follow up on the patients, and sometimes the patient needs a bit more than less 🙂



  2. “In-depth case studies” = in-depth anecdotes.

    Anecdotes/case studies are not a valid substitute for science. Psychologists, of all people, should understand the multitude of psychological biases that permeate anecdotes/case studies. The scientific method is needed to overcome this.

    If a therapist prefers to operate as a nonobjectivist hermeneuticist, then that should exclude them from making any claims to be part of the healthcare system, and should not be paid for with taxpayer dollars or insurance premiums. And, of course, it is the ethical obligation of such practitioners to disclose the unscientific, non-evidence-based status to all of their clients for meaningful informed consent.



Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s