Cognitive-behavioral Therapy Versus Yoga for the Treatment of Worry in Anxious Older Adults: A Randomized Preference Trial

The primary aim of this study is to compare the effects of CBT and yoga on worry in older adults (as assessed by the PSWQ-A measured at post-intervention, Week 11). Secondary aims are to compare the effects of these treatments on anxiety and sleep (as assessed by the PROMIS anxiety scale and the ISI, respectively, measured at post-intervention, Week 11). Exploratory aims are to determine participant preference for CBT vs. yoga; examine participant preference effects on worry, anxiety, sleep, adherence to treatment, and attrition rates; and examine selection effects on worry, anxiety, sleep, adherence to treatment, and attrition rates. All analyses will be repeated for measures assessed at Week 37.

The treatment effect for the primary aim will be estimated by comparing mean PSWQ-A scores between CBT and yoga groups in the random group (N=250, 125 per group) using constrained mixed-model repeated measures analysis of covariance with an unstructured covariance matrix to account for the fact that the multiple measurements (at baseline-Week 0, mid-intervention-Week 6, post-intervention-Week 11) from participants are not independent. The model will contain terms for baseline psychotropic medication use, gender and race (both related to depression), and intervention effects that are specific to each follow-up time. Because this arm of the trial has been randomized, we will constrain the pre-randomization intervention-specific outcome means to be the same. A contrast will be used to test the primary hypothesis at the post-intervention (Week 11) time point using a two-sided 0.05 significance level. In the primary analysis, all randomized participants will be included in their original study group for analysis regardless of the final mode of intervention or the extent of compliance with the study protocol; that is, the primary analysis will follow an "intent to treat" philosophy.

As part of the secondary aims, the estimation of selection and preference effects will be performed with mixed models based on the complete sample using data collected in both preference and randomized arms of the trial. Therefore, these analyses will be based on a sample size of 500 individuals. The adjusted means and variance-covariance matrix needed to compute these effects and their standard error will be estimated from the fitted model. The standard error associated with the preference and selection effects will be derived using the delta-method and/or a bootstrapping approach, as needed.

Consistency of intervention effects will be explored within the following baseline subgroups: 1) depressive symptoms from PROMIS measure (none or mild vs. moderate or severe), 2) use of psychotropic meds (any vs none), 3) age (60-79 vs 80+), 4) gender (female/male), and 5) race (White vs. other races).