Two mechanisms have been proposed to explain the effectiveness of CBT:
To date, there has been only one demonstration that change in a cognitive variable precedes, rather than follows, clinical improvement (Bouchard et al. 1994). However, their study examined only change in dysfunctional beliefs. The contribution of self-efficacy must still be establish, as well as its relationship with change in dysfunctional beliefs.
AimTo assess the relative importance of dysfunctional beliefs and self-efficacy in the treatment of panic disorder.
MethodSubjects had to receive the diagnosis of panic disorder with agoraphobia.
All subjects were diagnosed with the S.C.I.D. interview.
All subjects had to report more than 4 panic attacks in the last month during the S.C.I.D. interview.
To be recruited, all subjects had to suffer from panic disorder for more than 6 months, be aged between 18 et 60 and show no signs of psychotic features, organic disorder or substance-related disorder.
If on medication, it had to be stable and ineffective (i.e., still meeting selection criteria).
Random assignment to either:
Sample22 participants were recruited and treated.
In order to study treatment mechanism, only the 15 participants who were panic-free at post-treatment were selected for our analyses. Among them, self-monitoring was unreliable for 3 cases.
92% were female.
Sequence of events1- Pre-treatment meeting to discuss:
3- Beginning of 15 sessions of group treatment within 18 weeks:
Information of the Cognitive Restructuring conditionTreatment was based on Clark's model of panic. This model relies on a cognitive explanation of panic attacks.
Cognitive techniques were used.
A limited number of behavioral experiments were used.
If participants asked about exposure, its importance was dismissed.
Information on the Exposure conditionTreatment was based on Barlow's model of panic. This model relies on a conditioning explanation of panic attacks.
Exposure techniques were used for exposure to interoceptive and agoraphobic cues.
If participants asked about beliefs, they were presented as by-products.
Treatment fidelityDifferentiation between the conditions was maximized with:
Limitations of standard analyses
in the assessment of treatment processWith regressions it is impossible to show temporal precedence of change in one variable over change in another variable.
When multiple measurements are made, non independence of the observations (serial dependency) violates a very important assumption of ANOVAs and regressions.
Aggregating the data of all subjects together to conduct the analyses may seriously blur individual differences.
Advantages of multivariate time series analysis.It is based on many observations (> 100) and controls for serial dependency.
It is applied independently for each subject.
It allows to test the presence of temporal precedence.
Results: Example with subject no. 3Female.
She suffered from the disorder for 8 years.
It was her third psychotherapy.
She had an history of use of antidepressant and benzodiazepines.
Her main belief was: "I will loose consciousness during a panic attack".
She was in the cognitive restructuring condition.
Multivariate time series analysisMultivariate time series analysis provides a plausible description of the data by developing a model which takes into account the serial dependency, the relationship between the variables, and an error term. This process is illustrated in the Figure below.
Illustration of "causality testing"In performing causality analysis (Boudjellaba et al., 1992), the researcher is testing the Weiner-Granger description of causality which can be formulated as: is past information on the strength of the main belief essential to adequately explains the current level of panic apprehension?
Subject no. 3: Plot of the seriesBefore we present the summary of the results for all subjects, the following two Figures will illustrate Subject 3's results. Plots for the other subjects are similar.
Path diagram for subject no. 3Results of the causality tests for Subject 3 are presented in the form of a path analysis. Significant relationships across time are illustrated by an arrow. Horizontal arrows represent the serial dependency components, and diagonal arrows represent the relationship between the variables. Results for all subjects are summarized in the following Table.
Results of time series analysis
ConclusionTreatment mechanism is a complex process that varies between individuals.
Change in cognitive variables does precede change in panic apprehension in all panic-free subjects.
Change in belief and in self-efficacy are both important variables implicated in the treatment mechanism.
There may be a need to subtype patients in order to tailor more effective treatment.
Multivariate time series analysis is a promising technique.
ReferencesBouchard, S, Gauthier, J., Laberge B., Plamondon, D., French, D., & Pelletier, M. H. (1994). In vivo exposure versus cognitive therapy in the treatment of panic disorder with agoraphobia. 28th convention of the Association for the Advancement of Behavior Therapy (AABT), San Diego, November 1994.
Boudjellaba, H., Dufour, J-M. & Roy, R. (1992). Testing causality between two vectors in multivariate autoregressive moving average models. Journal of the American Statistical Association, 87, 1082-1090.