Noon, Rest from Work by Vincent van Gogh, painted 1889-1890 while he was in a psychiatric hospital
Most sleep meds were tested on primary insomnia patients. Here’s a guide built for the secondary insomnia we typically see.
STUDY TYPE: Expert consensus algorithm
FUNDING: Independent
Background
The Harvard South Shore Psychopharmacology Algorithm Project reviewed the evidence for treating insomnia in psychiatric disorders and built a decision tree.
The Study
- A narrative review of controlled trials, meta-analyses, and prior Harvard algorithms for major depressive disorder, generalized anxiety disorder, PTSD, and bipolar mania.
- Authors synthesized findings into a six-node flowchart, prioritizing treatments by quality of evidence and tolerability.
- Primary insomnia, without a psychiatric diagnosis, was excluded.
Results
The algorithm begins with a biopsychosocial evaluation to identify the psychiatric driver of the insomnia. When more than one disorder is present, treat the one contributing most to the sleep problem first.
For major depressive disorder, eszopiclone has the strongest evidence base — at least seven studies — and improves both sleep and depression scores. Trazodone up to 100 mg is a reasonable second choice, though avoid it with fluoxetine or paroxetine due to a metabolite (m-CPP) that can cause anxiety and dysphoria. Benzodiazepines come last, given dependence risk and cognitive side effects. Zolpidem carries a new caution: a 2022 meta-analysis of over 340,000 patients found an 88% higher rate of suicide or suicide attempt with its use, though confounding by indication can’t be ruled out.
For generalized anxiety disorder, eszopiclone and zolpidem both have trial support. If an SSRI is causing the insomnia in GAD, consider switching to hydroxyzine or pregabalin, both of which treat anxiety and improve sleep.
For PTSD, prazosin is the standout choice. Ten placebo-controlled trials exist, seven positives. The effective dose is higher than many clinicians use: a mean of 16 mg at night and 5 mg in the morning for men, and 7 mg at night and 2 mg in the morning for women, after slow titration over a month.
For bipolar mania, a second-generation antipsychotic treats both the mania and the insomnia (Dr. Aiken’s tip: quetiapine has the best evidence to deepen sleep; lumateperone is also a good choice. However, for bipolar depression ramelteon is safer and has evidence to prevent bipolar depression). A short-term benzodiazepine can bridge the gap while the mood stabilizer takes effect. Avoid antidepressants.
For conditioned insomnia — the fear of sleeplessness that persists after the original cause is gone — cognitive behavioral therapy for insomnia (CBT-I) is the treatment of choice (consider the Sleep Coach app).
Practice Implications
- Find more algorithms from the Psychopharm Algorithm Project here.
- Orexin antagonists are missing here, but a new analysis makes a case for them in depression. They have a favorable profile, improving sleep quality and next-day cognition.
- Learn more about eszopiclone (Lunesta)’s unique effects in depression and anxiety in our Carlat Podcast.
—Chris Aiken, MD
Director, Psych Partners
Editor in Chief, Carlat Psychiatry Report








3 comments
Chien C.W
June 17, 2026 at 10:12 pm
I have several depressed patients who responded well to Dayvigo. As it is the only DORA available in my country, I think it’s still worth a try.😀
Sue Cummings
July 3, 2026 at 8:01 pm
Is prazosin effective for insomnia secondary to PTSD because it controls nightmares? Or does it work when there is insomnia secondary to PTSD without nightmares?
Chris Aiken, MD
July 4, 2026 at 5:15 am
It improves sleep in both cases, also reduces daytime symptoms of PTSD hyperarousal. More detail in PTSD algorithm:
https://psychopharm.mobi/algo_live/