Sleep Therapy Beats Sleep Meds

Therapy boosts sleep meds, but sleep meds don’t add to the therapy

STUDY: Buysse DJ et al, J Clin Sleep Med 2026;22:58

STUDY TYPE: Systematic review, meta-analysis, and GRADE assessment

FUNDING: American Academy of Sleep Medicine

Background

CBT-I is the most effective treatment for chronic insomnia — better than medications in head-to-head trials. Yet clinicians routinely combine the two. Until now, no clinical practice guideline had directly addressed whether combining them actually helps.

The Study

The American Academy of Sleep Medicine commissioned this systematic review and meta-analysis to look at whether outcomes are improved by combining cognitive behavioral therapy for insomnia (CBT-I) with sleep medication. Among the 15 trials, most used benzodiazepines or z-hypnotics (like Ambien) alongside in-person CBT-I over 6–16 weeks in adults with chronic insomnia.

Adding CBT-I to medication produced clinically meaningful improvements in global insomnia severity (Insomnia Severity Index [ISI], Sleep Quality Index [PSQI]) and sleep continuity — patients fell asleep about 8 minutes faster and spent 14 fewer minutes awake during the night — compared to medication alone.

However, adding medication to CBT-I produced no clinically meaningful improvements in insomnia severity, sleep continuity, or daytime symptoms compared to CBT-I alone. The combination did add about 14 minutes of total sleep time compared to CBT-I alone, which the task force rated as a meaningful but not critical benefit. Daytime symptoms — mood, fatigue, worry — trended worse in the combination group than in CBT-I alone, though the difference didn’t reach the clinical meaningfulness threshold. The overall evidence was rated low to moderate certainty.

Practice Implications

1. CBT-I is already ranked top-line for insomnia by most medical societies, and this analysis adds more reason. Learn how to use it, and guide patients with the free Sleep Coach app.
2. It raises the possibility that sleep meds slightly worsen next-day functioning, and indeed most studies of benzos and z-hypnotics show they do not accomplish what sleep is supposed to. Next-day functioning is a mix of better and worse on them, with no difference overall.
3. The analysis did not include orexin-antagonists, which do improve next-day functioning. Ramelteon was also left out.