Category: Adult Anxiety - GAD

PS9- #A2 - Validity and Specificity of the Worry Behavior Construct in GAD

Saturday, Nov 18
9:45 AM – 10:45 AM
Location: Indigo Ballroom CDGH, Level 2, Indigo Level

Keywords: Anxiety | Worry | GAD (Generalized Anxiety Disorder)

Generalized anxiety disorder (GAD) is characterized by persistent and excessive worry about multiple domains of daily life. Research suggests that GAD is also manifested by worry behaviors such as avoidance, over-preparation, procrastination, and reassurance seeking (Mahoney et al., 2016). However, there is currently insufficient evidence to include worry behaviors in the DSM-5 definition of GAD. This study evaluated the validity of the worry behavior construct by examining the latent structure of the 4 worry behaviors proposed for inclusion in DSM-5, and by examining the differential relationships of this construct with diagnostic and self-report measures of GAD and neighboring disorders (social anxiety disorder, SAD; unipolar depression, DEP; obsessive-compulsive disorder, OCD). It was hypothesized that a robust latent dimension of worry behaviors would emerge that: (a) was more strongly related to GAD than other emotional disorders; and (b) would uniquely predict GAD beyond the extant, core features of GAD (excessive worry, associated symptoms).

The sample was 702 outpatients with a range of anxiety and mood disorders (307 met criteria for GAD). Individuals were assessed using the Anxiety and Related Disorders Interview Schedule for DSM-5 (ADIS-5). During the GAD section of the ADIS-5, clinicians rated each of the 4 worry behaviors on a scale of 0 (never) to 8 (always/very severe), along with the excessiveness of 8 worry areas, and the severity of 6 associated symptoms. Patients also completed questionnaires that represented dimensional indices of GAD (GAD-7), SAD (SIAS), OCD (OCI-R), and DEP severity (DASS-DEP).

After inclusion of an error covariance for the avoidance and procrastination behaviors, a unidimensional confirmatory factor analytic model fit the data well; χ2 (1) = 0.70, p = .39. Although the factor loadings for each behavior were salient and statistically significant (range = .44 to .61), the scale reliability of the Worry Behaviors construct was somewhat modest (ρ=.55, estimated using the CFA parameter estimates). As predicted, at the diagnostic level, the Worry Behavior latent variable was more strongly correlated with GAD (.715) than with other disorders (rs = .01, .24, and .24 for OCD, DEP, and SAD, respectively). Logistic regression results supported the incremental validity of the worry behavior construct by showing that the Worry Behaviors factor contributed significantly to the prediction of GAD diagnostic status (ΔR2 = .052) even after holding the key dimensional features of GAD constant (excessive worry, associated symptoms). However, concurrent validity was less favorable using dimensional, self-report indicators of GAD and other disorders. In a CFA where these self-report measures were brought into the analysis as covariates, the Worry Behaviors factor was correlated as strongly with DEP as with GAD (both rs = .34); rs with SAD and OCD were .28 and .26, respectively. The Worry Behaviors factor also did not explain significant unique variance in the dimensional GAD-7 indicator when holding GAD excessive worry and associated symptoms constant. In addition to presenting other results, the findings will be discussed with respect to the robustness and contribution of the worry behaviors construct in GAD.

Svetlana Goncharenko

Research Technician
Boston University

Jeannette K. Correa

Doctoral Student
Boston University
boston, Massachusetts

Esther S. Tung

Boston University

Anthony J. Rosellini

Boston University

Timothy A. Brown

Boston University