رابطه زمانی بین باورهای ناکارآمد، خودکارآمدی و دلهره و وحشت در درمان اختلال پانیک با موقعیت هراس

The aim of this study is to assess if changes in dysfunctional beliefs and self-efficacy precede changes in panic apprehension in the treatment of panic disorder with agoraphobia. Subjects participated in a larger study comparing the effectiveness of cognitive restructuring and exposure. Four variables were measured: (a) the strength of each subject's main belief toward the consequence of a panic attack; (b) perceived self-efficacy to control a panic attack in the presence of panicogenic body sensations; (c) perceived self-efficacy to control a panic attack in the presence of panicogenic thoughts; and (d) the level of panic apprehension of a panic attack. Variables were recorded daily on a “0” to “100” scale using category partitioning. Multivariate time series analysis and “causality testing” showed that, for all participants, cognitive changes preceded changes in the level of panic apprehension. Important individual differences were observed in the contribution of each variable to the prediction of change in panic apprehension. Changes in apprehension were preceded by changes in belief in three cases, by changes in self-efficacy in six cases, and by changes in both belief and self-efficacy in the remaining three cases. This pattern was observed in participants in the exposure condition as well as those in the cognitive restructuring condition. These results provide more empirical support to the hypothesis that cognitive changes precede improvement. They also underlie the importance of individual differences in the process of change. Finally, this study does not support the hypothesis that exposure and cognitive restructuring operate through different mechanisms, namely a behavioral one and a cognitive one.

Two theories are usually proposed to explain the treatment mechanism of panic disorder with agoraphobia. The most popular theory has been proposed by Clark (1986) and Beck (1988). According to them, therapeutic improvement of people suffering from panic disorder with agoraphobia operates through changes in the strength of subjects’ dysfunctional beliefs toward bodily sensations. There are many evidences suggesting that this theory is valid (see Rapee, 1993 for a review). For example, Clark, Salkovskis, Hackman, Middleton, Anastasiades, and Gelder (1994) and Margraf and Schneider (1991) used change scores from pre- to post treatment and found that changes in dysfunctional beliefs were the most powerful predictors of treatment outcome. Booth and Rachman (1992) and Shafran, Booth, and Rachman (1993) have also assessed the strength of subject's dysfunctional beliefs and investigated the extent to which they change during treatment. These researchers found that the treatment process and the relapses were related to the decrease in the number of specific negative cognitions and to the degree of belief in these cognitions. They also found some subjects for whom improvement did not appear to be related to changes in cognitions. It should be noted, however, that this study was conducted with patients suffering from claustrophobia, not panic disorder. None of these studies could shed light on the direction of the relationship between changes in dysfunctional beliefs and outcome. Casey, Oei, and Newcombe (2005) also showed that a reduction in dysfunctional beliefs predicts subsequent changes in panic severity. Self-efficacy theory is also put forward to explain the treatment mechanism of panic disorder with agoraphobia. According to Williams (Williams & Laberge, 1994) and Casey, Oei, and Newcombe (2004), therapeutic improvement in the treatment of panic disorder also operates through an increase in perceived self-efficacy to control a panic attack or the bodily sensations occurring during a panic attack. Barlow's (1988) model of panic disorder also stresses low perceptions of control as a key variable in the development and maintenance of panic disorder. Sanderson, Rapee, and Barlow (1989) clearly demonstrated that an illusion of control can reduce the likelihood of panic attacks induced by the inhalation of 5% of CO2. In order to assess treatment mechanism Borden, Clum, and Salmon (1991) used cross-lagged panel analysis and found that changes in self-efficacy precede changes in catastrophic thoughts and sometimes precede changes in panic symptoms. When trying to predict agoraphobic fear, Hoffart (1996) and Zane and Williams (1993) found that self-efficacy is a better predictor than dysfunctional beliefs. Using a series of hierarchical regressions, Casey, Oei, Newcombe, and Kenardy (2004) showed that self-efficacy predicts panic severity. They also showed that dysfunctional beliefs predicted panic severity and that both dysfunctional beliefs and panic self-efficacy contributed independently to the prediction. Casey et al. (2005) further revealed that changes in self-efficacy and changes in dysfunctional beliefs predicted changes in panic severity at mid-treatment. Changes in panic severity from pre- to post-treatment were significantly predicted only by changes in self-efficacy. In order to address the question of causality in outcome research, it may be necessary to solve many important limitations related to current methodology and the studies cited above. The first methodological limitation has to do with the number of measurements. To show that temporal precedence occurs, very frequent assessments of the key variables must be performed. To measure treatment mechanisms, day-to-day observations may be needed to understand change processes. Unfortunately, frequent measurements of the same variable is subject to violation of a central assumption in standard statistical analyses (such as regression, cross-lagged panel analysis, ANOVA, t-tests, and their non-parametric counterparts), namely, the independence of observations ( Kirk, 1982). It has been clearly demonstrated that violation of this assumption could lead to severe problems in statistical inference and invalidate the conventional tests ( Nurius, 1983; Scheffe, 1959). A second methodological limitation is the use of aggregated data which may blur individual differences (Hilliard, 1993). For example, subjects may progress at a different rate during therapy. Moreover, the treatment mechanism may be different for some subjects. In fact, the presence of individual differences has been found in the perception of bodily sensations (Steptoe & Vogele, 1992), the physiological functioning of panic disorder patients (Anastasiades et al., 1990) and in the cognitive-behavioral treatment of agoraphobia (Michelson, 1986) or fear (Vallis & Butcher, 1986). A third methodological problem frequently encountered is the inclusion of treatment nonresponders in the analysis of treatment mechanism. As put by Hoffart “a mediational analysis of therapeutic change presupposes the existence of changes to be explained” (Hoffart, 1996, p. 316). Therefore, it would be more beneficial to assess treatment mechanism among treatment responders. The last but not least methodological problem is that the diversity of treatment strategies used in treatment protocols may involve different mechanism. For example, Beck and Zebb (1994) and Hoffart (1993) suggested that cognitive restructuring and exposure operate through different mechanisms. If this is case, research on treatment mechanism should control for the type of strategy delivered to the patients. Obviously, delineating treatment mechanism and the day-to-day temporal relationship between variables involved represents a major challenge for clinical research. At the present time, one of the most powerful and most appropriate methodology to study closely the process of change over time is the multivariate time series analysis of longitudinal data (Mills, 1990; Shumway, 1988; Wei, 1990). Time series analysis (Box & Jenkins, 1970; Glass, Wilson, & Gottman, 1975; Shumway, 1988; Wei, 1990) refers to a set of statistical procedures for which the primary objective “is to develop mathematical models that provide plausible descriptions for sample data [equally recorded over a given period of time]” (Shumway, 1988, p. 10). The multivariate (vector) form of time series analysis offers a direct test to assess if there is a lagged relationship between variables, it controls for the nonindependence of the data across time and is not performed on data aggregated across subjects. The aim of the present study is to test on a day-to-day basis if there is a temporal precedence (lagged effect) in the relationship between the strength of subjects’ main dysfunctional belief, self-efficacy to control a panic attack and the level of panic apprehension over the course of treatment.