Which panic disorder patients benefit from which treatment: cognitive therapy or antidepressants?
Dusseldorp E, Spinhoven P, Bakker A, van Dyck R, van Balkom AJ;
Commented by , 30 Apr 2007
Aim of the study
To determine whether patient beliefs about the controllability of panic disorder (locus of control or LOC) predicts improvement following Cognitive Behavioural Therapy or antidepressant treatment.
Participants with panic disorder (DSM-III-R) were randomly assigned to antidepressant treatment (paroxetine or clomipramine: n = 64), CBT (n=35) or pill placebo (n=32). Another group of patients received CBT by preference (n= 31). As preliminary analyses found this group did not differ from the randomised group, the two CBT groups were amalgamated for analysis. < /p> < p> Treatment duration was 12 weeks. LOC of panic was assessed using the Multidimensional Anxiety Locus of Control Scale, which has four subscales (internal, chance, medication and therapist). The primary outcome variable was the post-treatment Hamilton Rating Scale for Anxiety (HAMA) score.
Using the regression trunk approach, a significant interaction between CBT and moderate locus of control (> 3.25 and <=4.42 on the internal locus of control scale)was found, indicating that the moderate locus of control group treated with CBT had lower posttest HAM scores compared with the pill-placebo condition. The interaction between low (<= 3.25) and high (>4.42) LOC was not significant.
Beliefs concerning the role of chance, medication and the therapist in controllability of panic did not predict changes in HAMA in people treated with CBT. Similar results were found when other outcome measures were used, including the CGI severity scale, anxiety subscale and agoraphobia subscale of the Marks-Sheehan Phobia Scale.
Dr Hood's and Prof Nutt's comments
This finding may be somewhat counterintuitive. People who have a high internal locus of control (a belief that they have control over their panic symptoms) do not fair better than placebo following CBT, although people with a moderate internal locus of control do.
Although replication of the study with a larger sample size might reveal a significant effect of high locus of control also, it may be that those with a moderate locus actually have more to gain from CBT. Belief in medication was not required for efficacy in the medication arms - reinforcing a commonly espoused belief.
In the one-size-fits-all approach to treatment efficacy, individual differences are often seen as a source of error variance, an annoyance or something to be controlled. Understanding the source of this variance, as was attempted in the study described here, may lead to better matching of patients to treatments, with a consequent reduction in time spent unwell and financial strain on the health system.
The challenge will be to establish reliable predictors of treatment response/non response to available treatments that pragmatic within a standard clinical context. Within a research context, understanding why certain therapies do not work in all subjects should lead to a better understanding of how they do work in others - which may open up new avenues for treatment.
There are a few methodological concerns that are not answered by this study. The first is that a psychological-placebo as well as a pill-placebo group would have been welcome. Although good examples of methodologically sound CBT / antidepressant comparator trials are hard to find, there is at least one shining example (ref. 1, ref. 2).
Secondly, the authors reference prior publications in this area, leading to the suspicion that that current analysis is a secondary analysis of an established dataset. We would prefer that authors "come clean" about such practices, as although it wouldn't significantly detract from the findings it would explain the failure to recruit a psychological-placebo for example.
Thirdly, there is a numerically significant increase in drop-out rates between antidepressant treated subjects (lower) than psychologically-treated subjects; however this result is not discussed. If statistically significant, one might consider the psychologically-treated subjects to be a healthy survivor sample. Finally, it would also have been useful to know if subjects completed their pre-test ratings before or after randomisation (or choice in one arm) to a particular treatment group.
What impact does active choice of therapy have? Interestingly, CBT-by-preference subjects appear to have similar dropout rates to those randomised to CBT – and higher than both medication groups. Patients claim to prefer talking therapies over medications (ref. 3), but how much does this preference impact upon therapeutic outcomes?
Quitkin started the ball rolling re. identifying the patterns of response to psychiatric therapy ("pattern analysis") some years back now (ref. 4) however there appears to be little recent research interest in this area. Genetic profiling services (eg: (ref. 5)) are starting to gain some traction in this area, however it seems prudent to supplement these data with psychological and other profiling methods as is partially the intent of the study reported here.
Unpacking one of our most powerful, cost-effective and oft-maligned therapies in psychiatry – the placebo effect – into its component parts should be a fruitful pursuit.
1. Heimberg RG, Liebowitz MR, Hope DA, Schneier FR, Holt CS, Welkowitz LA, et al. Cognitive behavioral group therapy vs phenelzine therapy for social phobia: 12-week outcome. Archives of General Psychiatry 1998; 55 (12); 1133-1141. Free full text article
2. Liebowitz MR, Heimberg RG, Schneier FR, Hope DA, Davies S, Holt CS, et al. Cognitive-behavioral group therapy versus phenelzine in social phobia: long-term outcome. Depression and Anxiety 1999; 10 (3); 89-98
3. Dwight-Johnson M, Sherbourne CD, Liao D, Wells KB. Treatment preferences among depressed primary care patients. Journal of General Internal Medicine 1999; 2000; 15 (8); 527-534
4. Quitkin FM. Placebos, drug effects, and study design: a clinician's guide. American Journal of Psychiatry 1999; 156 (6); 829-836. Free full text article
5. Pharmacogenetics - Personalizing Medicine today!. Health & DNA