New Data on Sleep-Disrupting Meds

A large real-world analysis names the worst offenders — including some that don’t warn you on the label

STUDY: Huang X et al, Medicine 2026;105(16):e48160

STUDY TYPE: Pharmacovigilance study (observational)

FUNDING: Independent

Background

This analysis mined the FDA Adverse Event Reporting System (FAERS) database for insomnia signals across drug classes.

The Study

• 179,697 adverse event reports flagging insomnia from the FAERS reporting system (2004–2024)
• Limited to cases where drug was primary suspect
• Applied four statistical algorithms, including disproportionality analysis, and required all four to flag a drug before calling it a signal

Results

The strongest signal by far was mefloquine, the antimalarial (ROR 22.22), consistent with its well-known neuropsychiatric toxicity. That was followed by:

1. Viloxazine, a non-stimulant for ADHD (ROR 12.29)
2. Flibanserin (ROR 12.02)
3. Finasteride (ROR 10.93) (and this one lacks a risk of insomnia on the label)
4. A hepatitis C antiviral combination (ROR 10.68)

Quetipine triggered a large number of reports, possibly due to withdrawal phenomena or akathisia, but this did not stand up to the four-level corrections. The psych meds that stood up to corrections are in the image below, led by

1. Viloxazine
2. Selegiline (Emsam)
3. Vilazodone
4. Ramelteon (I have seen paradoxical insomnia on it)
5. Varenicline (Chantix, a frequent cause of vivid dreams and nightmares)

Stimulants (methylphenidate, amphetamine) showed moderate signals. Non-stimulants like viloxazine and atomoxetine showed stronger ones, likely due to their norepinephrine reuptake inhibition (atomoxetine also ranked high for reports of somnolence in another study of this database, possibly due to high levels in patients with impaired CYP2D6 enzymes).

The risk with finasteride here is relatively new. This widely used med for prostate hypertrophy may cause insomnia through long-term impairment of neuroactive steroid synthesis via 5α-reductase inhibition, weakening GABAergic inhibition. Pimavanserin and fexofenadine also generated signals despite no insomnia label warnings.

Limitation: FAERS is a spontaneous reporting system. It can’t establish causality, and underreporting is substantial. Nearly 18% of reports lacked a recorded indication. These signals are hypothesis-generating, not proof.

Practice Implications

1. When patients complain of insomnia, ask when it began and whether it was related to these meds.
2. TIP: For nightmares on the smoking-cessation med varenicline (Chantix), consider temporary treatment with low-dose clonidine (which also has anti-smoking benefits).

— Chris Aiken, MD
Director, Psych Partners
Editor in Chief, Carlat Psychiatry Report

What’s Your Take? Share in Comments

1. What meds are you seeing insomnia on?