Early response to psychotherapy and long-term change in worry symptoms in older adults with generalized anxiety disorder

Abstract

Objectives: To determine the association of early and long-term reductions in worry symptoms after cognitive behavior therapy (CBT) for generalized anxiety disorder (GAD) in older adults.

Design: Substudy of larger randomized controlled trial.

Setting: Family medicine clinic and large multispecialty health organization in Houston, TX, between March 2004 and August 2006.

Participants: Patients (N = 76) aged 60 years or older with a principal or coprincipal diagnosis of GAD, excluding those with significant cognitive impairment, bipolar disorder, psychosis, or active substance abuse.

Intervention: CBT, up to 10 sessions for 12 weeks, or enhanced usual care (regular, brief telephone calls, and referrals to primary care provider as needed).

Measurements: Penn State Worry Questionnaire (PSWQ) administered by telephone at baseline, 1 month (mid treatment), 3 months (posttreatment), and at 3-month intervals through 15 months (1-year follow-up). The authors used binary logistic regression analysis to determine the association between early (1 month) response and treatment responder status (reduction of more than 8.5 points on the PSWQ) at 3 and 15 months. The authors also used hierarchical linear modeling to determine the relationship of early response to the trajectory of score change after posttreatment.

Results: Reduction in PSWQ scores after the first month predicted treatment response at posttreatment and follow-up, controlling for treatment arm and baseline PSWQ score. The magnitude of early reduction also predicted the slope of score change from posttreatment through the 15-month assessment.

Conclusion: Early symptom reduction is associated with long-term outcomes after psychotherapy in older adults with GAD.

Figure 1

Prediction of long-term change trajectories on the Penn State Worry Questionnaire (PSWQ) as a function of the 1-month PSWQ change score. Lines represent 25^th, 50^th, and 75^th percentiles of PSWQ 1-month difference scores, respectively. Baseline PSWQ score and treatment arm are held constant at their respective means.