[Skip to Navigation]
Comment & Response
March 16, 2022

The Direct Effect of Cognitive Behavioral Therapy for Insomnia on Depression Prevention and the Mediation Effect via Insomnia Remission—Reply

Author Affiliations
  • 1Cousins Center for Psychoneuroimmunology, Jane and Terry Semel Institute for Neuroscience and Human Behavior, Department of Psychiatry and Biobehavioral Sciences, David Geffen School of Medicine, University of California, Los Angeles
JAMA Psychiatry. 2022;79(5):515. doi:10.1001/jamapsychiatry.2022.0152

In Reply In an estimated mediation model, Lin et al supported conclusions from our recent report1 that cognitive behavioral therapy for insomnia (CBT-I) vs an active comparator control, that is sleep education therapy (SET), reduced the risk of incident and recurrent depression by over 50% in older adults with insomnia. Further, Lin et al examined whether the survival benefit of CBT-I was mediated by insomnia remission. We found that CBT-I was more likely to result in sustained remission of insomnia compared with SET, and in a predefined secondary analysis, we reported that those who received CBT-I and had a sustained remission of insomnia showed an 83% reduction in the risk of depression compared with those who received SET without insomnia remission (Figure 4 in the main report1). In the total sample, sustained remission of insomnia was associated with a 64% reduction in the risk of depression (eTable 6 and eFigure 21). Moreover, in analyses that examined varying duration of insomnia remission, longer duration of insomnia remission was associated with greater reduction in risk of depression (eFigure 31). However, to extend these inferential results and examine whether insomnia remission mediates the survival benefit of CBT-I, a more robust sample is needed to achieve adequate statistical power. In response to Lin et al, we performed a mediation analysis using methods developed by Valeri and VanderWeele.2 The mediation model found a direct effect of CBT-I in preventing incident and recurrent depression (effect size, 1.83; 95% CI, 0.99-3.36; P = .05), whereas the indirect effect for sustained insomnia remission was not significant (effect size, 1.11; 95% CI, 0.94–1.32; P = .23). The proportion mediated by insomnia remission was 20%, similar to the estimate from Lin et al. Post hoc calculation of statistical power indicated that a 4-fold greater sample would be required to test this mediational model. Hence, additional studies with substantially larger sample sizes are needed to evaluate whether the pathway of insomnia remission mediates the efficacy of CBT-I in preventing depression in older adults with insomnia.

Add or change institution