Category: Adult Anxiety - GAD

PS9- #A1 - An Empirical Comparison of Theoretical Models of Worry

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

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

Generalized anxiety disorder (GAD) is a common disorder characterized by chronic, pervasive worry (APA, 2013). Although interventions for GAD have been proven to be efficacious, up to 40% fail to recover from the disorder following treatment (Hanrahan et al., 2013). Because conceptual models inform treatment, refining models has the potential to enhance treatment efficacy. A number of models explaining worry in GAD have been conceptualized and each informs interventions (Behar et al., 2009). These models include the 1) Avoidance Model of Worry (AMW; Borkovec et al., 1993), 2) Intolerance of Uncertainty Model (Dugas et al., 1998), 3) Metacognitive Model (MCM; Wells, 1995), 4) Emotion Dysregulation Model (EDM; Turk et al., 2005), and 5) Acceptance-Based Model (Roemer et al., 2002). While each model has empirical support, they have rarely been compared, which leaves little information about which model explains the most about worry. Information about which model performs best may suggest to clinicians which intervention is likely to have the largest impact. Therefore, the current study was designed to compare the amount of overall and unique variance in worry explained by each model in a mixed clinical sample, as well as in those with and without GAD. Based on previous literature indicating that negative beliefs about worry and meta-worry have strong associations with worry relative to other model components, it was hypothesized that the Metacognition Model would explain the most variance in worry compared to other models.


The sample included 205 (M = 33.06 years [SD = 15.31]; 55.1% female; 85.2% White) diagnostically heterogeneous patients from a partial hospital program. At pre-treatment participants completed the Penn State Worry Questionnaire, Cognitive Avoidance Questionnaire, Intolerance of Uncertainty Scale, Metacognitions Questionnaire-30, Negative Problem Orientation Questionnaire, Affective Control Scale, Difficulties Emotion Regulation Scale, Berkley Expressivity Questionnaire, and Acceptance and Action Questionnaire-II. Regression models were fit to examine the amount of variance explained by each of the models. Hierarchical regression analyses were then conducted to examine the unique contributions of relative models and again for individuals with and without GAD.


Results for the individual model regressions indicated that all five models accounted for a significant amount of variance in worry. The EDM accounted for the greatest amount of variability (44.6%) while the AMW accounted for the least (24.4%). Both the EDM and MCM accounted for a significant amount of variability (both 4.7%) above and beyond that explained by other models. In the group without GAD, no model explained a significant amount of variability beyond any other models. However, in the group with GAD, the EDM explained 12.5% more variability than competing models, F(8, 79) = 3.209, p = .003.


The present findings indicate that the EDM and MCM are the most effective at explaining worry when compared to other conceptualizations. These findings provide additional understanding of GAD and rationale for selection of specific protocols to maximize treatment effectiveness for GAD. Implications, future directions, and limitations will be discussed. 

Keith P. Klein

Graduate Student
Southern Illinois University
Carbondale, Illinois

Eva K. Harris

Graduate Assistant
Southern Illinois University - Carbondale

Jennifer Koran

Assistant Professor
Southern Illinois University - Carbondale

Thröstur Björgvinsson

Director, Behavioral Health Partial Program
McLean Hospital

Sarah J. Kertz

Assistant Professor
Southern Illinois University-Carbondale
Carbondale, Illinois