Changes in affect during treatment for depression and anxiety

Abstract

We tested the hypothesis that the tripartite model [Clark, L. A., & Watson, D. (1991). Tripartite model of anxiety and depression: Psychometric evidence and psychometric implications. Journal of Abnormal Psychology, 100, 316–336] can be extended to account for change during treatment for anxiety and depression. Forty-one patients treated naturalistically in private practice with cognitive behavior therapy completed weekly measures of depression, anxiety, negative affect (NA), positive affect (PA), and anxious arousal (AA). Consistent with the model, NA was associated with anxiety and depression during treatment, PA was more strongly related to depression than to anxiety, and AA was more strongly related to anxiety than to depression. As predicted, symptoms of depression and anxiety and NA all decreased during treatment. As predicted, AA also decreased, particularly for patients with panic disorder. PA increased during treatment, but only for patients who showed a significant decline in depression and only over an extended period of treatment. Nearly two-thirds of the variance in anxiety change was accounted for by changes in depression and NA, and just over three-fourths of the variance in depression change was accounted for by changes in anxiety and NA, indicating that much of the change in anxiety and depression across the course of treatment is shared in common.

Introduction

Depression and anxiety are among the most common psychological disorders, with lifetime prevalence rates for depression at just over 16% (Kessler, Berglund, Demler, & Jin, 2003) and for the anxiety disorders at nearly 29% (Kessler et al., 2005). Not only are these conditions common, they are also highly comorbid. Nearly two-thirds of individuals with depression meet diagnostic criteria for an anxiety disorder (e.g., Mineka, Watson, & Clark, 1998), and as many as 50% of individuals who meet criteria for an anxiety disorder are depressed (e.g., Brown, Campbell, & Lehman, 2001).

Given the substantial overlap of anxiety and depression, researchers and clinicians alike have long been interested in understanding the relationship between them. One of the most influential theories about the relationship between anxiety and depression is the tripartite model proposed by Clark and Watson (1991; Watson, Clark et al., 1995; Watson, Weber et al., 1995; Watson, Weise, Vaidya, and Tellegen, 1999). In this model, a general distress factor characterized by high levels of negative affect (NA) is common to both anxiety and depression. A positive affect (PA)/anhedonia factor¹ that is characterized by low levels of PA or pleasurable engagement with the environment is specific to depression, and a third factor, variously referred to as anxious arousal (AA) or somatic arousal, is specific to anxiety. A revision to the model was later proposed to better account for the heterogeneity among the anxiety disorders (Mineka et al., 1998). The revised model, termed the integrative hierarchical model, followed from additional data suggesting that high levels of AA were more characteristic of panic disorder rather than all of the anxiety disorders (Brown, Chorpita, & Barlow, 1998; Zinbarg & Barlow, 1996).

The tripartite model has achieved broad empirical support across children, younger and older adults, college students and psychiatric patients (e.g., Beck et al., 2001; Brown et al., 1998; Cook, Orvaschel, Simco, Hersen, & Joiner, 2004; Joiner, Catanzaro, & Laurent, 1996; Lonigan, Phillips, & Hooe, 2003; Marshall, Shebourne, Meredith, Camp, & Hays, 2003; Watson, Clark, & Carey, 1988; Watson, Clark et al., 1995; Watson, Weber et al., 1995). However, some studies have found more limited support for the model, perhaps reflecting differences in data analytic techniques (e.g., Burns & Eidelson, 1998; Wetherell, Gatz, & Pedersen, 2001), less clear applicability in elderly samples (e.g., Beck et al., 2003; Shapiro, Roberts, & Beck, 1999; Wetherell et al., 2001; but see Cook et al., 2004), or differences in measures of the three factors (e.g., Burns & Eidelson, 1998; Mineka et al., 1998; Wetherell et al., 2001). Taken together, there is a good deal of support for the central tenets of the tripartite and integrative hierarchical models, namely that NA characterizes both anxiety and depression, low PA is more characteristic of depression than anxiety, and AA is more characteristic of anxiety, especially panic disorder, than depression. Most support for the tripartite model comes from cross-sectional studies, and few studies have been carried out in patients receiving treatment.

In the study reported here, we tested the hypothesis that the tripartite model can be extended to the pattern of changes in depression and anxiety over time—and, in fact, during treatment. We predicted that over the course of treatment, patients would show reductions in symptoms of anxiety, symptoms of depression, and NA, and increases in PA. We predicted that reductions in anxiety would be associated with reductions in AA, especially for patients with panic disorder. Tests of these hypotheses have both theoretical and clinical importance. They have theoretical importance because they extend the boundaries of the explanatory power of the tripartite model, and they have clinical importance because increased understanding of the relationship between anxiety and depression over time and during treatment has the potential to inform clinicians’ work with anxious depressed patients and even to lead to new and improved treatments for these patients.

To our knowledge, only three studies have assessed change in NA, PA, or AA over time and during treatment, and all these studies were of carefully selected patients with depression. Mohr et al. (2005) showed that over the course of a 16-week telephone-administered psychotherapy, depressed multiple sclerosis patients who were randomly assigned to either cognitive behavioral (T-CBT) or supportive emotion focused therapy (T-SEFT) showed decreases in depression symptoms and increases in PA (NA and AA were not assessed). The patients who received T-CBT showed a greater decrease in most measures of depression and a greater increase in PA than patients who received T-SEFT.

Tomarken and colleagues (Tomarken, Dichter, Freid, Addington, & Shelton, 2004) measured change in depression, anxiety, NA, PA, and AA in outpatients with depression across 12 weeks of medication treatment (bupropion [wellbutrin] SR). During the first 6 weeks of treatment (phase one), patients were randomly assigned to receive medication or placebo. During the last 6 weeks of treatment (phase two), all patients knowingly received medication. NA, PA, and AA were assessed before and during treatment using the Mood and Anxiety Symptoms Questionnaire (MASQ), a measure developed by Watson and colleagues to test the tripartite model (Watson, Clark et al., 1995; Watson, Weber et al., 1995). During phase one, depressed patients who received medication showed a significantly larger decrease in depression and NA and a significantly larger increase in PA than patients who received placebo. No change in AA or anxiety occurred in either group during phase one. In phase two (medication for all), all patients showed a decrease in depression, anxiety, NA and AA. Only the group who received medication in both phases showed a significant increase in PA. There was no significant group difference in the rate of decline in depression symptoms during phase two, although the patients who received medication in both phases had a very low level of symptoms at the end of treatment.

Finally, a third study reported preliminary findings on changes in PA, NA, and AA across 16 weeks of treatment for depression with cognitive behavior therapy or paroxetine (Paxil) (Schmid, Freid, Hollon, & DeRubeis, 2002). PA, NA, and AA were assessed at pre-treatment, mid-treatment (week 8), and post-treatment. Regardless of treatment type, all patients experienced a significant decrease in NA and AA, and a significant increase in PA over the course of treatment. Changes in NA and PA were more rapid for patients receiving medication, but by the end of treatment there were no differences between the treatment groups. Data on comorbid anxiety disorders were presented for half the sample, and there were no differences in PA, NA, or AA at pre-treatment for depressed patients with and without a comorbid anxiety disorder. Unfortunately, data regarding change among individuals with comorbid anxiety were not presented.

These studies support the notion that the tripartite model's predictions about depression can be extended to account for changes over the course of treatment. As predicted by the (extended) model, the studies showed that treatment was associated with decreased depression, decreased NA, and increased PA. None of these studies, however, tested whether the tripartite model's predictions about anxiety can be extended to account for changes over the course of treatment. In the present study, we collected weekly measures of symptoms and affect in order to address whether the tripartite model's predictions about depression and anxiety can be extended to account for changes during treatment in patients who have symptoms of depression and anxiety.

The study reported here also sought to extend the tripartite model by assessing the degree to which change during treatment reflects change in common or distinct features of anxiety and depression. According to the tripartite model, general emotional distress, or NA, is common to both anxiety and depression. An intriguing and as yet unanswered question regarding the relationship between measures of anxiety, depression, and NA is how much of the variance in change during treatment is accounted for by common aspects of anxiety and depression, including NA, versus how much is accounted for by distinct aspects of anxiety and depression. The answer to this question is theoretically important in that it can further extend the tripartite model by illuminating the nature of the change that occurs during treatment. In addition, understanding the amount of variance in change that is common and specific can guide treatment development by informing us about what actually changes during treatment.

Thus, the present study examined whether the tripartite model can characterize change in anxiety and depression across the course of treatment in a naturalistic and highly comorbid sample, and whether change during treatment is best captured by common or specific aspects of anxiety and depression. We tested several hypotheses based on the tripartite model. First, we hypothesized that during treatment, NA would be related to both anxiety and depression, PA would be more strongly related to depression than anxiety, and AA would be more strongly related to anxiety than depression. Second, we tested the hypothesis that the tripartite model can be extended to account for change during treatment for anxiety and depression. We tested the hypotheses that as depression decreased, NA would be significantly reduced and PA would be significantly increased. Consistent with the integrative hierarchical model, we further hypothesized that as anxiety decreased, NA would be significantly reduced, and AA would also be significantly reduced, but primarily for those patients with panic disorder. Finally, we assessed the proportion of the variance in change during treatment that is attributable to common versus specific aspects of the anxiety and depression measures.