CBT vs. Drugs for Depression

by Milton Spett

                 On March 7, Steven Hollon delivered NJ-ACT’s ninth annual Master Lecture to a record audience of 58.  In his introduction, Arnold Lazarus said that there is no one working in the field of depression who is more capable.  Hollon described his clinical approach to treating moderate to severe depression, and also described his and other research on treatment outcome.

Hollon’s CBT for Depression

             Hollon’s clinical approach includes these CBT components: explaining the CBT model, behavioral activation, cognitive restructuring, identifying and changing maladaptive schemas, relapse prevention, and termination planning.  Before introducing cognitive restructuring, Hollon recommends focusing on behavioral activation, encouraging the patient to engage in pleasure and mastery tasks.  His approach to cognitive restructuring de-emphasizes Ellis’s technique of disputation, and focuses on asking the patient to design behavioral experiments which will confirm or contradict the patient’s dysfunctional cognitions.  With complex or character disordered patients, Hollon advocates working on schemas by exploring both early experiences and the therapeutic relationship.

Hollon: 4 months of CBT  is more effective than 16 months of drug therapy.

             In the videotaped therapy segments he showed, Hollon focused on reviewing the homework forms completed by his patient: the Weekly Activity Schedule, the Thought Record, the Core Belief Worksheet, the Cognitive Conceptualization Diagram, the Relapse Prevention Plan, and the Fear Form.  Blank forms and sample completed forms were distributed at the workshop and NJ-ACT members can receive copies by contacting the office at 908-654-0122 or NJACBT@aol.com.

 Hollon’s Depression Study

            The recently completed Cognitive-Pharmacotherapy Research Project (CPRP) was co-led by Hollon at Vanderbilt University and, at the University of Pennsylvania, by previous NJ-ACT presenter Robert DeRubeis.  The subjects were 240 patients with Hamilton Rating Scale for Depression (HRSD) scores of 20 or above.  For the first four months of the study, 60 patients received CBT, 120 received anti-depressant medication (ADM), and 60 received a placebo.  After eight weeks, 49% of the ADM group achieved a HRSD score of less than 13, compared with 40% of the CBT group and 26% of the placebo group.  But at the conclusion of the sixteen-week treatment period, the CBT and ADM groups were almost exactly equal, with just under 60% of the patients in each group below 13 on the HRSD.  The placebo group did not continue beyond the initial eight weeks.

            After their first sixteen sessions, the CBT group received up to three booster sessions, while half of the ADM group continued on medication and the other half was switched to a placebo.  One year later, 39% of the CBT group remained below 13 on the HRSD, compared with 30% of the ADM group and 16% of the group that was switched from ADM to placebo.  These results support Hollon’s conclusion that four months of CBT (plus up to 3 booster sessions) is more effective than sixteen months of drug therapy for moderately to severely depressed patients.  Note that this is true only while patients continue taking ADM.  Hollon emphasized that according to many studies, once ADM is withdrawn, patients are just as likely to become depressed as they were before taking medication.

 The NIMH Depression Study

            Hollon began his presentation by describing the National Institute of Mental Health depression study, which compared sixteen weeks of CBT, imipramine, Interpersonal Therapy (IPT), and placebo.  Among the less severely depressed patients (HRSD < 20), just over 40% in all three active treatments ended with HRSD scores < 7.  Among the more depressed patients (HRSD  20 or more), 40% of both the ADM group and the IPT group ended with HRSD scores < 7.  But only about 20% of the more depressed CBT patients were below 7 on the HRSD.  Hollon asserted that this occurred  because the CBT therapists were not properly trained and monitored at two of the three NIMH sites, and Hollon’s own study appears to support this assertion.

Can We Generalize From Research

Therapy to Real Life Therapy?

             Real therapists are not as highly trained, supervised, and monitored as research therapists.  Hollon criticized the NIMH study for using CBT therapists who were not so well trained at two of their three sites.  But these less well-trained therapists might be more typical of the CBT that is actually provided to real patients.  Of course you and I know what we are doing, but many practicing CBT therapists may be less competent than the therapists in these carefully monitored research studies. 

The benefits of psychotherapy are permanent, but the benefits of medication end when the patient stops taking the medication.

 On the other hand, research therapists have a very detailed protocol they must follow.  This rigid approach to treatment could hamper therapist flexibility and spontaneity, possibly impeding treatment efficacy.  In a previous NJ-ACT Master Lecture, Marvin Goldfried argued that rigid treatment protocols improve the effectiveness of beginning therapists, but impede the effectiveness of experienced therapists.  So it is possible that patients receiving treatment from experienced therapists in naturalistic settings may benefit more than patients in research studies.  Research therapists treat disorders; real therapists treat patients.  When patients present in private practice, we do not just accept those patients who meet rigid inclusion and exclusion criteria.  We see patients who have multiple problems, uncertain diagnoses, and even problems that do not fit neatly into any DSM category.  We cannot necessarily generalize from disorder-focused research to the treatment of real patients.  And often we cannot treat the complex problems of real patients according to a treatment manual or even according to multiple treatment manuals.

 Real patients are not as compliant as research patients.  Research patients volunteer for studies knowing they may be assigned to any of several treatments or to no treatment at all.  They know they will have to comply with the treatment techniques they are given.  Unlike real patients, research patients complete rigorous pre-treatment screening forms which unintentionally screen out patients unwilling to complete these forms.  And since real patients pay for their treatment, they usually have the ability to choose a therapist who provides a treatment they like.

 Real patients want more than research therapy provides.  No patient has ever presented in my office saying “I want to be less depressed than the average no-treatment control patient at the .05 level of confidence.”  Although statistically significant, the outcome of research therapy is abysmal.  Even in the elegantly delivered therapy of Hollon and DeRubeis, eight months post-treatment, only 39% of the CBT group had HRSD ratings below 13.  Most real patients want to be completely free of depression, and not just for one year.

Four months of any therapy just isn’t enough for the vast majority of patients.  One major depressive patient I treated with short-term CBT finally became depression-free in her tenth year of treatment.

Unanswered Questions in the

Treatment of Depression

           I wish therapy researchers would stop spending money on placebo controls.  The superiority of CBT over placebos for depression is well established.  It is also well established that ADM is superior to placebos while patients are taking their medication, but ADM has no benefit once it is discontinued.  In real life, no one takes a placebo, so other than for obtaining FDA approval of new drugs, there is no reason to include placebo groups in depression outcome studies.

In real life, many patients are treated with CBT plus ADM over months or even years, and it is pretty well established that during the usual four months of research treatment, CBT, ADM, and CBT plus ADM do about equally well.  It is also well established that after treatment, ADM-only patients revert to their pre-medication levels of depression, while psychotherapy patients maintain their gains. The key question facing clinicians now is “Does medication given concurrently with CBT impair the effectiveness of the CBT once patients stop taking their medication?”  David Barlow and previous NJ-ACT presenter Michael Otto have published research that concurrent medication impairs the effectiveness of CBT for panic disorder.  But is this also true for depression?

Finally, future research should study treatment that is longer than 16 weeks.  In the NIMH study, only 28% of the CBT patients were depression-free after 16 weeks of treatment and remained depression-free 18 months later.  In the Hollon study, only 25% of the CBT patients were depression-free after 16 weeks of treatment and remained depression-free 2 years later.  If you were depressed, would you undertake a treatment that was only 25% effective?

©2002 NJ-ACT. All rights reserved.
Reproduction of any documents, related graphics, or any other material from this World Wide Web site is strictly prohibited without permission from the organization. For permission, click here to email us.