Lunesta FAERS Safety Signals: What Post-Market Surveillance Reveals About Eszopiclone

How FAERS Works and Why Eszopiclone Matters

FAERS is the FDA's primary post-market safety surveillance database, collecting voluntary adverse event reports from healthcare professionals, patients, and manufacturers. Reports do not prove causation. They generate statistical signals that the FDA evaluates for regulatory action.

Eszopiclone entered the FAERS pipeline following its December 2004 approval as the first non-benzodiazepine sedative-hypnotic granted approval without a restriction on duration of use. The original key trial by Krystal et al. (Sleep, 2003) demonstrated sustained efficacy over 6 months in 308 patients with primary insomnia, but the study period was too short and the sample too small to capture rare safety events that surface only with widespread prescribing. Between 2005 and 2012, eszopiclone prescriptions exceeded 30 million in the United States, per IMS Health data reported by the FDA. That prescription volume supplied the denominator against which FAERS signals gained statistical weight.

What makes the eszopiclone FAERS record clinically important is not any single adverse event report. It is the pattern: the same types of serious events (parasomnias, next-day cognitive deficits, taste disturbance) appeared repeatedly across independent reporters, age groups, and dose levels. That pattern drove two distinct FDA regulatory actions within five years.

The 2014 Dose Reduction: Next-Morning Impairment

On May 15, 2014, the FDA issued a Drug Safety Communication lowering the recommended starting dose of eszopiclone from 2 mg to 1 mg. The agency cited evidence that blood levels of eszopiclone at 7.5 hours post-dose could remain high enough to impair driving, memory, and coordination.

The FDA reviewed a driving simulation study in which patients taking 3 mg eszopiclone showed impairment equivalent to a blood alcohol concentration above 0.05% at 7.5 hours post-dose. Patients on the 2 mg dose also demonstrated measurable impairment. The FDA safety communication stated: "Healthcare professionals should prescribe the lowest effective dose of Lunesta. The recommended starting dose of Lunesta is being lowered from 2 mg to 1 mg." The agency specifically warned that patients should not drive or engage in activities requiring full alertness the morning after taking eszopiclone.

This action followed a parallel January 2013 dose reduction for zolpidem (Ambien), where women were found to be at particular risk for next-morning impairment due to slower drug clearance. The eszopiclone label change applied the same concern to the broader insomnia drug class. FAERS reports of motor vehicle accidents, falls, and workplace injuries among eszopiclone users provided supportive real-world evidence alongside the controlled pharmacokinetic data.

The 2019 Boxed Warning: Complex Sleep Behaviors

On April 30, 2019, the FDA mandated a boxed warning on all three Z-drug sedative-hypnotics (eszopiclone, zolpidem, and zaleplon) for complex sleep behaviors. This is the most serious warning the FDA can require on a prescription drug label.

The trigger was a systematic FAERS review identifying 66 cases of serious injuries or deaths associated with complex sleep behaviors across the Z-drug class. These included sleepwalking episodes resulting in falls, burns, near-drowning, hypothermia from cold exposure, self-inflicted gunshot wounds, apparent suicide attempts, and motor vehicle collisions during sleep-driving. Twenty of those cases resulted in death per the FDA review.

The FDA's Acting Commissioner at the time, Dr. Ned Sharpless, stated: "While these incidents are rare, they are serious and it is important that patients and health care professionals are aware of the risk." The new labeling added a contraindication: eszopiclone is now contraindicated in patients who have previously experienced a complex sleep behavior episode with any sedative-hypnotic.

Complex sleep behaviors include sleepwalking, sleep-driving, and engaging in activities such as making phone calls, preparing food, or having sexual intercourse while not fully awake. Patients typically have no memory of these episodes. The FAERS data showed that these events occurred even at the lowest recommended dose, after only a single dose, and in patients with no prior history of parasomnias.

Signal Breakdown: What FAERS Reports Show by Category

FAERS reports for eszopiclone cluster into several distinct signal categories. Each category carries different clinical weight.

Complex sleep behaviors represent the highest-severity signal. Reports describe patients found walking outside their homes at night, driving vehicles with no recall the next day, and preparing meals while asleep. The 2019 boxed warning specifically addressed this category. A systematic review published in the Journal of Clinical Sleep Medicine documented parasomnias across all non-benzodiazepine receptor agonists and noted that the mechanism may involve selective suppression of cortical arousal while preserving subcortical motor activity.

Next-morning functional impairment accounts for a large proportion of FAERS reports. This category includes reports of motor vehicle accidents, occupational injuries, and falls, particularly among elderly patients. The pharmacokinetic basis is well established: eszopiclone has an elimination half-life of approximately 6 hours, but active metabolites and individual variation in hepatic CYP3A4 activity can extend functional impairment beyond 8 hours in some patients per the FDA-approved prescribing information.

Dysgeusia (metallic or unpleasant taste) is the most frequently reported adverse event in FAERS and in clinical trials. In the key Krystal et al. trial, unpleasant taste occurred in 34% of patients receiving eszopiclone 3 mg compared with 3% in the placebo group. This is not a safety signal per se but contributes to discontinuation and medication non-adherence.

Amnesia and cognitive effects form a clinically relevant signal cluster. Reports describe anterograde amnesia (inability to form new memories after taking the drug), confusion on awakening, and hallucinations, particularly visual hallucinations in the period between taking the medication and falling asleep. The American Geriatrics Society Beers Criteria lists all non-benzodiazepine hypnotics, including eszopiclone, as potentially inappropriate for adults aged 65 and older, citing risks of cognitive impairment, delirium, falls, fractures, and motor vehicle crashes.

Dependency and withdrawal signals appear in FAERS as well, though less prominently than for benzodiazepines. Eszopiclone is a Schedule IV controlled substance. Reports include tolerance development (needing higher doses for the same effect), rebound insomnia upon discontinuation, and withdrawal symptoms including anxiety, tremor, and nausea. The prescribing information notes that abrupt discontinuation may produce withdrawal signs.

FAERS Limitations: What the Data Cannot Tell You

FAERS is a hypothesis-generating system. It has significant limitations that affect interpretation of eszopiclone signals.

Reporting is voluntary. The FDA estimates that FAERS captures only 1% to 10% of actual adverse events for most drugs, a phenomenon known as underreporting bias. This means the absolute number of FAERS reports substantially underestimates the true incidence of any adverse event. Conversely, media attention (such as widespread coverage of Ambien sleepwalking episodes) can produce a reporting spike that inflates signal strength through stimulated reporting.

FAERS reports do not include a denominator. Without knowing how many patients took eszopiclone during a reporting period, you cannot calculate incidence rates from FAERS data alone. The FDA addresses this gap by cross-referencing FAERS signals with controlled trial data, prescription volume data from sources like IQVIA, and active surveillance systems such as the Sentinel Initiative, which analyzes claims data from over 100 million patients.

Confounding is a persistent problem. Many eszopiclone users take other medications, particularly other CNS depressants, opioids, or alcohol. An adverse event report listing eszopiclone does not prove that eszopiclone caused the event. A Sentinel analysis of sedative-hypnotic users found that co-prescription of benzodiazepines and Z-drugs occurred in nearly 10% of patients, complicating attribution of adverse events to a single agent.

Duplicate reports can inflate counts. The same event may be reported by both the prescribing physician and the patient, or by a pharmacist and the manufacturer, generating multiple entries for a single clinical event.

How the FDA Translates FAERS Signals Into Action

The FDA uses a structured pipeline to move from raw FAERS data to regulatory decisions. For eszopiclone, this pipeline operated across two phases.

Phase 1: Signal detection. FDA pharmacovigilance staff run disproportionality analyses on FAERS quarterly data extracts. The most common method is the Empirical Bayesian Geometric Mean (EBGM), which compares the observed reporting frequency of a drug-event pair against the expected frequency based on all drugs and all events in the database. An EBGM score above 2.0 generally triggers further evaluation. Complex sleep behaviors and next-morning impairment for eszopiclone exceeded this threshold by the late 2000s.

Phase 2: Signal evaluation. Once flagged, the FDA's Division of Drug Information assembles a comprehensive review incorporating FAERS case series, published literature, clinical pharmacology data, and, where available, active surveillance results. For the 2014 dose reduction, the agency commissioned a dedicated driving simulation study that quantified next-morning impairment at each dose level. For the 2019 boxed warning, the agency reviewed 66 individual FAERS case narratives in detail, assessing causality, excluding alternative explanations, and grading severity.

This two-phase process explains why the delay between early FAERS signals and regulatory action can span years. The 2019 boxed warning drew on FAERS reports accumulated over the entire 15-year post-marketing period.

Current Label Requirements and Prescribing Implications

The current eszopiclone prescribing information reflects all FAERS-driven changes as of the 2019 revision.

The label now carries a boxed warning for complex sleep behaviors, a starting dose of 1 mg (with allowance to increase to 2 mg or 3 mg if clinically needed), a contraindication in patients with prior complex sleep behavior episodes on any sedative-hypnotic, and specific warnings about CNS-depressant interactions and next-morning impairment. The elderly dosing recommendation is 1 mg, with a maximum of 2 mg, because of decreased clearance in older adults.

Prescribers should document informed consent discussions about parasomnia risk. Patients should be counseled to take eszopiclone only when they can dedicate at least 7 to 8 hours to sleep. Co-administration with other CNS depressants, including alcohol, opioids, and benzodiazepines, increases the risk of every FAERS signal category described above.

Comparative FAERS Context: Eszopiclone vs. Other Z-Drugs

FAERS data for the Z-drug class shows overlapping but non-identical safety profiles. Zolpidem (Ambien) generates the highest absolute volume of complex sleep behavior reports, partly because it has the largest market share and has received the most media coverage. A comparative analysis in Pharmacoepidemiology and Drug Safety found that all three Z-drugs (zolpidem, eszopiclone, zaleplon) shared statistically elevated FAERS signals for parasomnias relative to non-Z-drug sleep aids.

Zaleplon (Sonata) has a shorter half-life (approximately 1 hour) and generates fewer next-morning impairment reports, but its parasomnia signal remains present. Eszopiclone occupies the middle ground: a longer half-life than zaleplon (roughly 6 hours) but lower parasomnia report volume than zolpidem. The 2019 boxed warning applied identically to all three drugs, reflecting the FDA's conclusion that the class effect outweighed individual agent differences.

For clinicians evaluating sedative-hypnotic options, the FAERS record does not support the claim that any single Z-drug is meaningfully safer than the others for parasomnia risk. The American Academy of Sleep Medicine's 2017 Clinical Practice Guideline recommends cognitive behavioral therapy for insomnia (CBT-I) as first-line treatment, with pharmacotherapy reserved for patients who do not respond to or cannot access behavioral treatment. When pharmacotherapy is used, the guideline suggests eszopiclone as one of several options for sleep-onset and sleep-maintenance insomnia, with the caveat that all sedative-hypnotics carry risks that should be weighed against the morbidity of untreated insomnia.

Patients currently prescribed eszopiclone at doses above 1 mg should discuss dose optimization with their prescriber, particularly if they drive within 8 hours of dosing or have experienced any episodes of confusion, amnesia, or unusual nighttime behavior.