ReviewA meta-(re)analysis of the effects of cognitive therapy versus ‘other therapies’ for depression
Introduction
Over the years, many randomized clinical trials involving cognitive therapy for depression have been conducted. To aggregate the results of the numerous studies, meta-analyses have been conducted. These meta-analyses have consistently shown that the outcomes of cognitive therapy (CT) are approximately equal to the outcomes of behavior therapy and are superior to the outcomes of placebo or no treatment controls (Dobson, 1989, Robinson et al., 1990, Gaffan et al., 1995, Gloaguen et al., 1998). An important comparison, however, which has yielded inconsistent results in these meta-analyses, is the comparison of CT and ‘other therapies’ or ‘verbal therapies’. Although ‘other therapies’ and ‘verbal therapies’ have been ill defined, comparisons involving these treatments have been used to establish the specificity of cognitive therapy:
Is CT superior to other psychotherapies (behaviour therapy excluded)? If it were true, it would mean that cognitive therapy is a specific psychological treatment for depression. (Gloaguen et al., 1998, p. 60).
In a recently conducted meta-analysis that was distinguished by its sophistication of method, Gloaguen et al. (1998) found that CT was superior to ‘other therapies,’ which were defined as psychotherapies without distinct cognitive or behavioral components. However, they also found that the effects produced by the CT/other therapies comparisons were heterogeneous, indicating that some aspect of these comparisons was creating unexplained variance in the outcomes. The purpose of this meta-analysis is to investigate the source of heterogeneity in the CT/other therapies comparisons.
A perspicuous source of the heterogeneity produced by ‘other therapies’ may be related to whether or not these treatments were intended to be therapeutic. Wampold and co-workers (Wampold, 1997, Wampold et al., 1997a, Wampold et al., 1997b) have argued that treatments that are not intended to be therapeutic lack the common factors that have been shown to be therapeutic, such as a therapeutic relationship, belief in the treatment by the therapist, or a cogent rationale. Moreover, any therapy not containing components based on legitimate psychological principles is essentially a placebo (Grünbaum, 1981). The purpose of these treatments typically is to serve as a control rather than as a ‘bona fide’ treatment for the disorder. Wampold et al. (1997b) meta-analyzed comparisons between bona fide treatments and found that bona fide treatments were uniformly efficacious. On the other hand, comparisons between bona fide treatments and placebos have shown relatively large effects (Lambert and Bergin, 1994). Consequently, the heterogeneity produced by the CT/other therapies comparisons in the Gloaguen et al. (1998) meta-analysis may have been created by aggregating within the category ‘other therapies’ those that were bona fide with those that were not intended to be therapeutic.
The present meta-analysis reanalyzed the Gloaguen et al. (1998) data by segregating the bona fide and non-bona fide treatments that were contained within the ‘other therapies’ category. The following hypotheses were tested:
- 1.
For bona fide (non-cognitive and non-behavioral) treatments, the effects produced by the comparison of bona fide ‘other therapies’ with CT would be zero. Moreover, these comparisons would be homogeneous.
- 2.
For non-bona fide (non-cognitive and non-behavioral) treatments, the effects produced by the comparison of non-bona fide ‘other therapies’ with CT would show that CT was superior to these treatments. Because non-bona fide therapies range in the degree to which they contain the common factors of psychotherapy, it is expected that these comparisons will be heterogeneous.
Section snippets
Gloaguen et al. (1998) meta-analysis
Data for this meta-analysis were obtained from J. Cottraux, co-author of the Gloaguen et al. (1998) meta-analysis, which is described briefly here. The studies included in the meta-analysis were randomized trials comparing CT with a comparison group and included subjects who met criteria for major depression or dysthymic disorder but not the criteria for psychotic depression or bipolar disorder. Gloaguen et al. retrieved 22 studies that compared CT to ‘other therapies.’
For each study, effect
Results
The results of the analyses are summarized in Table 2. The hypothesis that the comparisons of CT to bona fide ‘other therapies’ would yield zero effect sizes was corroborated, as the 95% confidence interval for the true effect size contained zero (i.e., the null hypothesis was not rejected). It should be noted that the estimate of the effect size for these comparisons (viz., d+=0.16) is classified as a small effect by Cohen (1988). The 10 effect sizes were heterogeneous, contrary to the
Discussion
Segregating ‘other therapies’ into those that were intended to be therapeutic for depression (i.e., bona fide treatments) and those that were not (i.e., non-bona fide treatments) clarified the results obtained by Gloaguen et al. (1998). When CT was compared to bona fide psychotherapies without behavioral or cognitive components, the null hypothesis that the effects of these comparisons were zero could not be rejected. Moreover, when the outlier in this data set, which was produced by a flawed
Conclusion
The following limitations to the conclusions from this meta-analysis should considered. First, this review included relatively few studies. Second, the conclusions are dependent on the validity of the exclusionary criteria for classifying treatments as bona-fide. Finally, the quality of the studies included were variable. It also should be noted that the original meta-analysis as well as the present meta-analysis relied solely on the BDI as an outcome measure, which favors CT because the BDI
Acknowledgements
Appreciation is extended to J. Cottraux for graciously providing the data from Gloaguen et al. (1998).
References (51)
- et al.
A meta-analysis of the effects of cognitive therapy in depressed patients
J. Affect. Disord.
(1998) The placebo concept
Behav. Res. Ther.
(1981)- et al.
Brief therapy of the stress response syndrome
Psychiatr. Clin. North Am.
(1979) Comparison of effectiveness of group interventions for depression in women
Arch. Psychiatr. Nurs.
(1993)- et al.
The role of homework and skill acquisition in the outcome of group cognitive therapy for depression
Behav. Ther.
(1990) - et al.
- et al.
An inventory for measuring depression
Arch. Gen. Psychiatry
(1961) - et al.
- et al.
Predictors of differential response to cognitive, experiential, and self-directed psychotherapeutic procedures
J. Consult. Clin. Psychol.
(1991) Treatment of depressed in-patients: cognitive therapy plus medication, relaxation plus medication, and medication alone
Br. J. Psychiatry
(1990)