Emotional Processing Following Digital Cognitive Behavioral Therapy for Insomnia in People With Depressive Symptoms

Importance Cognitive behavioral therapy for insomnia (CBT-I) has been shown to reduce depressive symptoms, but the underlying mechanisms are not well understood and warrant further examination. Objective To investigate whether CBT-I modifies negative bias in the perception of emotional facial expressions and whether such changes mediate improvement in depressive symptoms. Design, Setting, and Participants A randomized clinical trial of digital CBT-I vs sleep hygiene education was conducted. Adults living in the UK who met diagnostic criteria for insomnia disorder and Patient Health Questionnaire-9 criteria (score ≥10) for depression were recruited online from the community and randomly assigned to either a 6-session digital CBT-I program or a sleep hygiene webpage. Participant recruitment took place between April 26, 2021, and January 24, 2022, and outcomes were assessed at 5 and 10 weeks post randomization. Data analysis was performed from December 1, 2022, to March 1, 2023. Main Outcomes and Measures Coprimary outcomes were recognition accuracy (percentage) of happy and sad facial expressions at 10 weeks assessed with the facial expression recognition task. Secondary outcomes were self-reported measures of insomnia, depressive symptoms, affect, emotional regulation difficulties, worry, perseverative thinking, midpoint of sleep, social jet lag, and the categorization of and recognition memory for emotional words. Intention-to-treat analysis was used. Results A total of 205 participants were randomly assigned to CBT-I (n = 101) or sleep hygiene education (n = 104). The sample had a mean (SD) age of 49.3 (10.1) years and was predominately female (165 [80.8%]). Retention was 85.7% (n = 175). At 10 weeks, the estimated adjusted mean difference for recognition accuracy was 3.01 (97.5% CI, −1.67 to 7.69; P = .15; Cohen d = 0.24) for happy facial expressions and −0.54 (97.5% CI, −3.92 to 2.84; P = .72; Cohen d = −0.05) for sad facial expressions. At 10 weeks, CBT-I compared with control decreased insomnia severity (adjusted difference, −4.27; 95% CI, −5.67 to −2.87), depressive symptoms (adjusted difference, −3.91; 95% CI, −5.20 to −2.62), negative affect (adjusted difference, −2.75; 95% CI, −4.58 to −0.92), emotional regulation difficulties (adjusted difference, −5.96; 95% CI, −10.61 to −1.31), worry (adjusted difference, −8.07; 95% CI, −11.81 to −4.33), and perseverative thinking (adjusted difference, −4.21; 95% CI, −7.03 to −1.39) and increased positive affect (adjusted difference, 4.99; 95% CI, 3.13-6.85). Improvement in negative affect, emotional regulation difficulties, and worry at week 5 mediated the effect of CBT-I on depression severity at 10 weeks (% mediated: 21.9% Emotion regulation difficulties; 24.4% Worry; and 29.7% Negative affect). No serious adverse events were reported to the trial team. Conclusions and Relevance This randomized clinical trial did not find evidence that CBT-I engenders change in the perception of facial expressions at post treatment, despite improvements in insomnia and depressive symptoms. Early change in negative affect, emotional regulation difficulties, and worry mediated lagged depression outcomes and deserve further empirical scrutiny. Trial Registration isrctn.org Identifier: ISRCTN17117237