A recent pharmacovigilance analysis published in Medicine (April 2026) examined FDA adverse event reporting data and identified a range of medications disproportionately associated with insomnia. That kind of signal detection is useful—but incomplete. Disproportionality metrics tell us what gets reported, not necessarily what drugs cause under controlled conditions, nor how those effects evolve over time.
To get closer to a clinically meaningful estimate of insomnia risk, I pulled together three complementary layers of evidence: placebo-controlled randomized trials (focusing on the excess insomnia attributable to the drug over 9–12 weeks), longer-term continuation data (to understand whether those effects persist over months of treatment), and the pharmacovigilance signals highlighted in the original analysis. Integrating these sources allows for a more nuanced view—one that separates true pharmacologic liability from reporting bias, and transient effects from those that meaningfully persist in practice.
Insomnia Data from Multiple Sources
| Drug | RCT Δ Insomnia (%) | ROR (95% CI) | IC (IC025) | Timing | Concordance across Data |
| Viloxazine (adult) | +16 | 6–8 | ≈2.5 | Early/adaptive | Concordant |
| Amphetamine | +4–15 | 5.1 (4.8–5.4) | ≈2.2 | Persistent subset | Concordant |
| Bupropion SR 400 | +10 | 5.7 (5.3–6.1) | ≈2.3 | Early/adaptive | Concordant |
| Atomoxetine | +7 | 6.1 (5.7–6.6) | ≈2.4 | Persistent subset | Concordant |
| Duloxetine | +4 | 8.5 (7.9–9.1) | ≈3.0 | Persistent subset | Concordant |
| Vilazodone | +4–5 | 9.2 (8.3–10.2) | ≈3.2 | Mixed | Concordant |
| Paroxetine | +3–7 | 5.2 (4.9–5.6) | ≈2.1 | Mixed | Partially concordant |
| Armodafinil | +4 | 5.8 (5.3–6.3) | ≈2.3 | Persistent subset | Concordant |
| Pitolisant | +4 | 6.3 (5.6–7.1) | ≈2.5 | Persistent subset | Concordant |
| Varenicline | +2–5 | 7.7 (7.2–8.2) | ≈2.8 | Confounded | Discordant |
| Lurasidone | +2–3 | 5.9 (5.5–6.3) | ≈2.3 | Mild | Partially concordant |
| Ramelteon | +1 | 8.3 (7.5–9.2) | ≈3.0 | Discordant | Discordant |
| Vortioxetine | ~0 | 5.6 (5.0–6.2) | ≈2.1 | Neutral | Discordant |
| Quetiapine | Low/NR | Low | Low | Protective/mixed | Discordant |
Notes: ROR = Reporting Odds Ratio; IC = Information Component. Values approximate based on pharmacovigilance literature. RCT Δ reflects placebo-subtracted insomnia rates.
Signal concordance reflects agreement between placebo-controlled RCT effect size, pharmacovigilance disproportionality metrics, and pharmacologic mechanism. Discordant signals often reflect indication bias, reporting bias, or non-causal associations.
Concordance definitions (for reference)
- Concordant = RCT Δ positive + PV signal positive + mechanism consistent
- Partially concordant = mixed or modest alignment across domains
- Discordant = strong PV signal but weak/absent RCT signal or clear confounding
Drug-by-drug notes
Viloxazine (adult) — Concordant
- RCT: Large Δ (+16%) → strongest causal signal in table
- PV: Elevated ROR/IC
- Mechanism: Noradrenergic modulation → activating
- Temporal: Likely early, sometimes adaptive
→ Clean alignment across all domains
Amphetamine — Concordant
- RCT: Moderate–large Δ (+4–15%)
- PV: Strong signal (ROR ~5)
- Mechanism: Direct stimulant (dopamine/norepinephrine)
- Temporal: Persistent in subset, dose/timing dependent
→ Prototypical insomnia-inducing agent
Bupropion (SR 400) — Concordant
- RCT: Clear dose-dependent Δ (+10%)
- PV: Strong signal
- Mechanism: Dopaminergic/noradrenergic activation
- Temporal: Often attenuates but present early
→ True signal, though less persistent than stimulants
Atomoxetine — Concordant
- RCT: Moderate Δ (+7%)
- PV: Strong signal
- Mechanism: Selective norepinephrine reuptake inhibition
- Temporal: Persistent in subset
→ Consistent across all domains
Duloxetine — Concordant
- RCT: Modest Δ (+4%)
- PV: Very strong signal (high ROR/IC)
- Mechanism: SNRI activation
- Temporal: May persist in subset
→ Strong alignment despite modest Δ
Vilazodone — Concordant
- RCT: Moderate Δ (+4–5%)
- PV: Very strong signal
- Mechanism: SSRI + 5-HT1A partial agonism (activating)
- Temporal: Mixed; some adaptation
→ Robust but not universally persistent
Paroxetine — Partially concordant
- RCT: Moderate Δ (+3–7%)
- PV: Strong signal
- Mechanism: SSRI, but relatively sedating vs others
- Temporal: Mixed (insomnia vs somnolence)
→ Internal inconsistency (bidirectional sleep effects)
Armodafinil — Concordant
- RCT: +4%
- PV: Strong signal
- Mechanism: Wake-promoting (orexin/histamine systems)
- Temporal: Persistent during exposure
→ Mechanistically and empirically aligned
Pitolisant — Concordant
- RCT: +4%
- PV: Strong signal
- Mechanism: H3 inverse agonist → histaminergic activation
- Temporal: Persistent in subset
→ Clean mechanistic + empirical alignment
Varenicline — Discordant
- RCT: Small Δ (+2–5%)
- PV: Strong signal
- Mechanism: Partial nicotinic agonist (not strongly activating)
- Confounds:
- nicotine withdrawal
- vivid dreams/sleep fragmentation
- Temporal: tied to treatment phase
→ PV signal likely inflated by confounding
Lurasidone — Partially concordant
- RCT: Small Δ (+2–3%)
- PV: Positive signal
- Mechanism: Antipsychotic; insomnia often secondary to akathisia
- Temporal: Mild/persistent subset
→ Real but indirect (not primary pharmacologic insomnia)
Ramelteon — Discordant
- RCT: Minimal Δ (+1%)
- PV: Strong signal
- Mechanism: Melatonin agonist (sleep-promoting)
- Confound:
- indication paradox (treated insomnia reported as AE)
→ Classic disproportionality artifact
- indication paradox (treated insomnia reported as AE)
Vortioxetine — Discordant
- RCT: ~0%
- PV: Modest signal
- Mechanism: multimodal serotonin modulation; generally neutral
- Temporal: neutral
→ Likely reporting noise or mild subgroup effect
Quetiapine — Discordant
- RCT: No meaningful insomnia signal
- PV: Weak or inconsistent
- Mechanism: Strong H1 antagonism → sedating
- Clinical reality: often used to treat insomnia
→ Negative control / protective profile
Key meta-observation
The concordance ratings expose a critical pattern:
High-confidence insomnia drugs
→ stimulants, SNRIs, wake-promoting agents
(agreement across all domains)
Moderate / mixed
→ SSRIs, atypical antipsychotics
(bidirectional or indirect effects)
False-positive pharmacovigilance signals
→ ramelteon, varenicline
(driven by indication + reporting bias)
Figure 1 Top 20 drugs associated with reports of insomnia
References
Core pharmacovigilance reference
- Huang X, Chen Y, Li S, Wen Y, Liu H, Cao G.
Drug-associated insomnia: A pharmacovigilance study based on FDA adverse event reporting system.
Medicine (Baltimore). 2026;105(16):e48160. doi:10.1097/MD.0000000000048160. PMID: 41995510; PMCID: PMC13095331.
FDA label–derived RCT adverse event sources
- Qelbree (viloxazine) prescribing information.
U.S. Food and Drug Administration. Updated 2025. - Goodell K, Erlich D.
Viloxazine (Qelbree) for ADHD.
Am Fam Physician. 2023;107(3):309–310. - Bupropion hydrochloride prescribing information.
U.S. Food and Drug Administration. Updated 2024. - Strattera (atomoxetine) prescribing information.
U.S. Food and Drug Administration. Updated 2025–2026. - Chantix (varenicline) prescribing information.
U.S. Food and Drug Administration. Updated 2026. - Viibryd (vilazodone) prescribing information.
U.S. Food and Drug Administration. Updated 2023. - Duloxetine prescribing information.
U.S. Food and Drug Administration. Updated 2025. - Armodafinil prescribing information.
U.S. Food and Drug Administration. Updated 2023. - Wakix (pitolisant) prescribing information.
U.S. Food and Drug Administration. Updated 2026. - Mixed amphetamine salts prescribing information.
U.S. Food and Drug Administration. Updated 2025. - Latuda (lurasidone) prescribing information.
U.S. Food and Drug Administration. Updated 2025. - Rozerem (ramelteon) prescribing information.
U.S. Food and Drug Administration. Updated 2026. - Amantadine hydrochloride prescribing information.
U.S. Food and Drug Administration. Updated 2025–2026. - Ropinirole prescribing information.
U.S. Food and Drug Administration. Updated 2025. - Phenelzine sulfate prescribing information.
U.S. Food and Drug Administration. Updated 2026.
Randomized trials / meta-analyses supporting effect estimates
- Punja S, Shamseer L, Hartling L, et al.
Amphetamines for ADHD in children and adolescents.
Cochrane Database Syst Rev. 2016;CD009996. - Oliva HNP, Prudente TP, Mayerson TF, et al.
Safety of stimulants across patient populations: a meta-analysis.
JAMA Netw Open. 2025;8(5):e259492. - Thomas KH, Martin RM, Knipe DW, Higgins JPT, Gunnell D.
Risk of neuropsychiatric adverse events associated with varenicline.
BMJ. 2015;350:h1109. - Zhou S, Li P, Lv X, et al.
Adverse effects of antidepressants on sleep: dose-effect network meta-analysis.
Sleep. 2023;46(10):zsad177. - Alberti S, Chiesa A, Andrisano C, Serretti A.
Insomnia and somnolence with second-generation antidepressants.
J Clin Psychopharmacol. 2015;35(3):296–303.
Additional mechanistic / clinical context
- Jha MK, Qamar A, Vaduganathan M, Charney DS, Murrough JW.
Screening and management of depression in cardiovascular disease.
J Am Coll Cardiol. 2019;73(14):1827–1845. - Remick RA, Froese C, Keller FD.
Common side effects associated with monoamine oxidase inhibitors.
Prog Neuropsychopharmacol Biol Psychiatry. 1989;13:497–504. - Jarrett RB, Schaffer M, McIntire D, et al.
Phenelzine vs placebo in atypical depression.
Arch Gen Psychiatry. 1999;56(5):431–437.
Statistics sources
- Bate A, Evans SJW.
Quantitative signal detection using spontaneous ADR reporting.
Drug Saf. 2009;32(4):313–325. - Szarfman A, Machado SG, O’Neill RT.
Use of screening algorithms and computer systems to detect adverse drug reactions.
Drug Saf. 2002;25(6):381–392.
Related Content
