Psychological Treatment of Comorbid Insomnia and Depression: A Double-Blind Randomized Placebo-Controlled Trial

Introduction: Insomnia and depression are highly prevalent disorders and commonly occur together. Cognitive behavioral therapy for insomnia, CBT-I, has been shown to be effective in treating insomnia and also comorbid depression. However, it is unclear whether effects of CBT-I on depression are specific or nonspecific. Also, depressive symptoms often remain too high after CBT-I, indicating a need for improved treatments. The objective was to determine whether combining CBT-I with CBT for depression, without increasing treatment length, reduces both insomnia and depression more than CBT for depression with a placebo insomnia intervention. Methods: A 12-week double-blind randomized controlled trial with a 6-month follow-up in a psychiatric setting using therapist-guided internet-delivered treatments was conducted. Patients (N = 126) were diagnosed with insomnia disorder and major depression by physicians. Primary outcome measures were as follows: self-rating scales Insomnia Severity Index (ISI) and Montgomery-angstrom sberg Depression Rating Scale (MADRS-S). Results: The combined treatment showed specific effects on insomnia severity over the control treatment (p = 0.007) but was not more effective in reducing depression severity. Within-group effects (Cohen's d) at post and at 6 months were as follows: ISI 1.40 and 1.42 (combined treatment), 0.95 and 1.00 (control); MADRS-S 0.97 and 1.12 (combined), 0.88 and 0.89 (control). Conclusions: CBT-I shows large specific effects on insomnia severity and is superior to control in this regard. Both treatments had similar effects on depression severity, i.e., combining CBT-I with CBT for depression did not enhance outcomes on depression compared to control. We suggest CBT-I should always be offered to patients with insomnia and depression comorbidity, possibly as the first-hand choice. Combining it with a psychological treatment for depression could be too burdening and may not be beneficial.