Prescribing Eszopiclone (Lunesta)

Why Eszopiclone Occupies a Peculiar Niche

Eszopiclone is the Z-drug that wants to be a benzodiazepine. Among the three non-benzodiazepine hypnotics, it has the longest half-life (5 to 7 hours), it is the only one that reliably treats sleep maintenance insomnia, and, uniquely, it has an active metabolite with daytime GABAergic activity that gives it genuine, replicated augmentation data in depression and generalized anxiety disorder. No other Z-drug can say that.

It also arrived in 2004 wrapped in one of the more cynical marketing campaigns in modern psychopharmacology. Sepracor took the active S-enantiomer of zopiclone (a hypnotic sold everywhere else in the world for decades), patented it as a "new" drug, priced it at roughly 100 times the cost of generic temazepam despite an essentially identical half-life, and ran its next-day cognitive testing at 9.5 to 12 hours post-dose, a window chosen precisely because competitors were tested at the FDA-standard 8 hours, where eszopiclone would have looked worse. The Carlat verdict at launch was blunt: "Lunesta has no advantages over its competitors and has marked disadvantages."

The truth sits between the marketing and the cynicism. Eszopiclone is a real drug with a real, narrow set of advantages, and a real black-box warning, a signature metallic taste that drives a fifth to two-fifths of patients off it, and next-day impairment that its long half-life makes unavoidable. This guide's thesis: eszopiclone is a reasonable choice for a small, specific population: sleep-maintenance insomnia, or insomnia comorbid with depression or GAD where you want one drug to help both, used at the lowest effective dose, intermittently rather than nightly, in a patient without substance-use history, with the black-box parasomnia risk actively managed. Outside that niche, cheaper and safer agents usually win.

A grounding fact to keep in view throughout: across the entire hypnotic class, the drug-specific benefit is small. Pooled FDA-registration data (Huedo-Medina, BMJ 2012) show Z-hypnotics get patients to sleep about 22 minutes faster by polysomnography but only about 7 minutes faster by subjective report, and add only about 11 to 12 minutes of total sleep. Much of what patients experience as benefit is placebo plus drug-induced amnesia: they forget how poorly they slept. That does not make hypnotics useless. It does mean the risk-benefit math is tight, and every avoidable risk matters.

Part 1: Indications: Who Is Eszopiclone For?

FDA-Approved Use

• Insomnia (sleep onset and sleep maintenance).

Eszopiclone holds a notable regulatory distinction: it is one of the few hypnotics approved without a restriction on duration of use. Efficacy trials support benefit out to 6 months, with some data suggesting maintained effect at 12 months. This does not mean nightly indefinite use is wise (the tolerance, dependence, and mortality-signal concerns below argue strongly against it), but the label does not force you into a "short-term only" framing.

Critically, eszopiclone is one of the Z-drugs approved for sleep-maintenance insomnia (Wake After Sleep Onset), not just sleep onset. Its 5 to 7 hour half-life is what earns it that indication. Zaleplon (about 1 hour) and immediate-release zolpidem (about 2.5 hours) are not approved for maintenance because they clear too fast to carry a patient through the night.

The Evidence-Based Off-Label Uses: Where Eszopiclone Actually Shines

This is the part that separates eszopiclone from the other Z-drugs, and it traces to a single pharmacologic fact: eszopiclone metabolizes into a compound with benzodiazepine-like daytime activity. Zolpidem and zaleplon do not, which is why they failed the analogous trials.

Major depression with insomnia (augmentation to an antidepressant). This is eszopiclone's strongest off-label case.

• In a large RCT of eszopiclone 3 mg co-started with an SSRI in depressed patients with insomnia (Fava, Biol Psychiatry 2006), depressive symptoms improved faster and more substantially than with the antidepressant alone: NNT of about 11 for remission.
• Crucially, a follow-up analysis (McCall, J Clin Sleep Med 2010) showed the antidepressant benefit persisted even after removing the sleep items from the depression scale. The drug did more than just fix sleep and let the score fall.

Generalized anxiety disorder with insomnia (augmentation).

• Eszopiclone 3 mg added to escitalopram (Pollack, Arch Gen Psychiatry 2008) significantly improved anxiety, sleep, and daytime functioning versus escitalopram alone. Again, this is attributed to the daytime-active metabolite, an effect unique among Z-drugs.

PTSD-related insomnia: mixed, do not lead with it.

• One 2011 RCT (Pollack, J Clin Psychiatry) found improved insomnia and daytime PTSD functioning without tolerance or dependence.
• A 2020 RCT was negative, but it was small (n<50) with a 36% dropout rate, which by Cochrane conventions makes the result unreliable.
• Net: eszopiclone is an option in PTSD insomnia, not a first choice. Prazosin remains first-line for PTSD nightmares and sleep-related hyperarousal, and Z-drugs are not ideal when trauma-focused psychotherapy is ongoing (parasomnia and amnesia risk).

Who Is a Good Candidate

• Sleep-maintenance insomnia (falls asleep fine, cannot stay asleep): the longer half-life is the point.
• Depression or GAD with prominent insomnia, where you want a single agent to help both, especially at treatment initiation.
• A patient without active substance-use disorder, without a history of complex sleep behaviors, and who can commit to going to bed immediately after dosing and avoiding alcohol.

Who Is a Poor Candidate (Worst Uses)

• Any prior complex sleep behavior on a Z-drug: this is an absolute FDA contraindication (see Part 6).
• Active or recent substance-use disorder: Z-drugs are reinforcing and Schedule IV; prefer ramelteon or a DORA.
• Acutely agitated or manic patients: a hypnotic is not the tool; treat the underlying state.
• ADHD-related insomnia: both eszopiclone and zolpidem failed controlled trials here. The problem is circadian ("night owl") phase delay, not a GABA deficit. Melatonin (or an alpha-agonist like clonidine or guanfacine) is preferred.
• Untreated obstructive sleep apnea: Z-drugs are safer than benzodiazepines in apnea but still blunt respiratory drive; screen and treat the apnea first.
• Patients where cost is decisive: generic trazodone is roughly 100 times cheaper with comparable insomnia efficacy.

Part 2: Before You Start: Workup and Candidacy

Eszopiclone requires no baseline labs, no ECG, and no serum-level monitoring. That is genuinely different from lithium, and it is one of the drug's practical conveniences. The workup is entirely a clinical history and a risk conversation.

Before the first prescription, establish:

Part 3: How to Start and Dose

Formulation

Oral tablet only. Branded Lunesta; generic eszopiclone is now widely available and is what you should prescribe: the branded product's cost was never justified by any clinical advantage.

Dosing: Simple, and Deliberately Low

There is no titration ladder here the way there is with lithium. You pick a dose and use it.

• Start: 1 mg at bedtime. The FDA lowered the recommended starting dose to 1 mg (from the old 2 mg) specifically because higher blood levels the next morning impair driving and mental alertness. Start low on purpose.
• Usual effective dose: 2 to 3 mg at bedtime. Titrate up only if 1 mg is inadequate and next-morning function is intact.
• Maximum: 3 mg. Do not exceed it: 3 mg is the point at which next-day impairment and complex-sleep-behavior risk climb without added hypnotic benefit worth having.
• Lower the ceiling in specific groups: elderly, hepatic impairment, and patients on CYP3A4 inhibitors should generally not go above 2 mg, and often stay at 1 mg (see Parts 7 and 8).

The Timing Rules That Actually Matter

These are not fine print: they are the core of using this drug safely.

• Take it immediately before bed, after getting into bed and turning out the light. The gap between swallowing the pill and losing consciousness is when complex sleep behaviors happen. Close that gap.
• Allow a full night (at least 7 to 8 hours) of sleep opportunity before you need to be alert. With a 5 to 7 hour half-life, less than that guarantees a hangover.
• Do not take it with, or right after, a heavy or high-fat meal. Food delays absorption by roughly an hour. A delayed onset means the patient is awake, disinhibited, and amnestic longer, precisely the setup for a parasomnia. Advise no food within 30 minutes of the dose.
• No alcohol, full stop, on nights eszopiclone is taken.

Frequency: Intermittent, Not Nightly

Where the milligram count is simple, the frequency decision is the one that separates good prescribing from bad. Because roughly 60% of chronic Z-drug users develop tolerance, and because rebound insomnia can make post-discontinuation sleep worse than baseline, the durable strategy for most patients is intermittent use: a few nights per week, or only on anticipated bad nights, rather than every single night. This preserves efficacy, limits dependence, and keeps the eventual taper short.

Part 4: Monitoring

No serum levels, no labs, no ECG. Monitoring is entirely clinical, and it is focused on catching the two things that actually hurt patients: next-day impairment and parasomnias.

At each visit, assess:

• Next-day sedation and cognition. Ask directly about morning grogginess, "sleep drunkenness," and any near-misses driving. The 5 to 7 hour half-life makes this the highest-risk Z-drug for residual impairment; the manufacturer's reassuring data were gamed by testing at 9.5 to 12 hours instead of the FDA-standard 8.
• Complex sleep behaviors. Ask, and ask the bed partner. Patients are amnestic for these events by definition, so a negative self-report is not reassurance. Any confirmed episode, however mild (a bowl of cereal at 2 a.m. counts), mandates discontinuation.
• The metallic taste. Common enough (20 to 40%) to be a leading reason patients quit; ask whether it's tolerable.
• Efficacy and tolerance. Is it still working? Has the patient crept toward nightly use or asked for dose increases? Both are signals to reassess.
• Ongoing need. Reevaluate whether the drug is still indicated. Deprescribe in anyone over 65 on a Z-drug for insomnia, and in anyone under 65 who has been on it more than about 4 weeks once the underlying driver (depression, anxiety) is controlled.
• Signs of misuse in any patient: early refills, "lost" prescriptions, escalating doses.

Part 5: Side Effects and How to Manage Them

The governing principle mirrors the class: most eszopiclone side effects are dose-dependent and improve at lower doses, but two of them, the metallic taste and next-day impairment, are intrinsic to the molecule and its half-life, and they, not lack of efficacy, are the usual reasons patients stop.

Dysgeusia: the Metallic/Bitter Taste (the signature problem)

• Incidence: 20 to 40%. This is eszopiclone's calling card and its single most common reason for discontinuation. Patients describe a persistent metallic or bitter taste, often lingering into the next day.
• Mechanism: not well understood; likely related to the drug's own excretion into saliva rather than a metabolite quirk.
• Management: there is no evidence-based remedy. Sepracor's own detail reps suggested biting a lemon on waking, advice Carlat memorably dismissed. You can suggest it, along with mints or gum, but manage expectations. If the taste is intolerable, it usually forces a switch (to zolpidem, a DORA, low-dose doxepin, or trazodone). Do not fight a losing battle here; the taste rarely habituates.

Next-Day Cognitive and Psychomotor Impairment

• Driven by the 5 to 7 hour half-life, the longest of the Z-drugs. Expect residual sedation, slowed reaction time, and possible driving impairment the morning after.
• Eszopiclone's precursor, zopiclone, is well-documented to cause next-day impairment and increased car-accident risk, and eszopiclone shares the profile.

Management:

• Use the lowest effective dose (1 mg often suffices; the FDA lowered the starting dose for exactly this reason).
• Ensure a full 7 to 8 hour sleep window.
• Counsel explicitly against driving the morning after until the patient knows how they respond.
• If morning impairment persists at 1 mg, switch to a shorter-acting agent: zaleplon (about 1 hour) leaves essentially no morning residue.

Anterograde Amnesia

• A GABAergic class effect: patients may not remember events occurring after they take the dose. Worse with alcohol, other sedatives, and higher doses.
• Management: bed immediately after dosing; avoid alcohol and co-sedatives. Amnesia is the substrate of complex sleep behaviors; if amnestic events are occurring, take it seriously.

Complex Sleep Behaviors

Sleep-driving, sleep-eating, sleep-cooking, sleep-texting, sexual acts, and sleep-walking, all in an amnestic state. Incidence across Z-drugs is 3 to 15%. This is the most dangerous adverse effect and carries a boxed warning; it is covered in full in Part 6.

Other Effects

• Headache, dizziness, somnolence: common, generally mild and time-limited; lower the dose.
• Respiratory: eszopiclone blunts respiratory drive less than benzodiazepines, but caution in apnea and with opioids remains warranted.
• Rebound insomnia: expect a transient worsening of sleep on discontinuation, especially after nightly use (see Part 9).

Part 6: Toxicity and the Boxed Warning

The Black-Box Warning: Complex Sleep Behaviors

In 2019 the FDA elevated the complex-sleep-behavior warning to a boxed (black-box) warning for eszopiclone, zolpidem, and zaleplon. The trigger was a 26-year review of adverse events that found 66 cases of serious complex sleep behaviors: 46 serious injuries and 20 deaths. The mechanisms of harm are exactly what you'd fear from someone acting out, unconscious and amnestic:

• Motor vehicle accidents (patient driving while asleep)
• Falls, fractures, and burns (sleep-cooking)
• Near-drownings and drownings
• Carbon monoxide poisoning
• Hypothermia and cold exposure severe enough to cause limb loss
• Self-injury, including gunshot wounds and apparent suicide attempts

Acute Overdose

• There is no specific antidote in routine practice; management is supportive (airway, breathing, circulation, observation). Flumazenil can reverse Z-drug effects pharmacologically but is generally avoided outside specialist settings because of seizure risk, especially with mixed ingestions or sedative dependence.
• Eszopiclone alone is relatively safe in overdose. The danger is combination: alcohol, benzodiazepines, or opioids taken together can cause serious respiratory depression and death.
• Rare withdrawal seizures have been reported only at extreme chronic doses (reports of Z-drug misuse in the range of 160 to 2,000 mg/day, versus a therapeutic 1 to 3 mg).

Abuse Potential

• Schedule IV. Real but modest: roughly 4.5 abuse cases per 10,000 doses for Z-drugs versus 106.7 for benzodiazepines, about a 20-fold lower rate. Z-drugs are still reinforcing (animals self-administer them), so monitor for escalation, early refills, and diversion, particularly in anyone with a substance-use history.

The Mortality Signal

Observational studies have linked hypnotic use, even occasional use, to increased all-cause mortality (cancer, cardiac, and other causes). The signal is almost certainly confounded by the underlying illness that drives people to hypnotics (depression, anxiety, medical comorbidity), and aggressive statistical adjustment can eliminate it. But it has not been cleanly refuted either. The practical reading: this is one more reason to use the lowest dose, intermittently, for the shortest necessary duration, and to prefer non-drug approaches when they'll do.

Part 7: Drug Interactions

Two interaction axes matter: pharmacokinetic (CYP3A4) and pharmacodynamic (other CNS depressants). The second is the one that kills people.

Pharmacokinetic: CYP3A4

Eszopiclone is metabolized primarily by CYP3A4. Anything that inhibits CYP3A4 raises eszopiclone levels, prolongs its action, and increases next-day impairment and parasomnia risk.

Strong CYP3A4 inhibitors: reduce the eszopiclone dose (cap at 1 mg, or avoid):

• Nefazodone (relevant in psychiatry, an antidepressant)
• Azole antifungals: ketoconazole, itraconazole
• Macrolide antibiotics: clarithromycin, erythromycin
• HIV antiretrovirals (ritonavir-boosted regimens)

Weaker inhibitors: modest elevation, monitor:

• Verapamil, diltiazem, cimetidine, grapefruit juice

CYP3A4 inducers (carbamazepine, rifampin, St. John's wort) can reduce eszopiclone levels and blunt efficacy; consider this if a stable patient suddenly stops responding.

Pharmacodynamic: the Dangerous Axis

Additive CNS and respiratory depression, plus a sharply elevated complex-sleep-behavior risk, when eszopiclone is combined with other GABAergic or sedating agents:

• Alcohol: the single most important one. Counsel total avoidance on nights the drug is taken.
• Benzodiazepines, barbiturates
• Opioids: additive respiratory depression; the FDA warns specifically about sedative-opioid combinations.
• Valproate (GABAergic anticonvulsant)
• Herbal GABAergics: valerian, kava, skullcap
• Other sedating psychotropics (antipsychotics, sedating antihistamines) add sedation.

A useful exception: gabapentin does not share this GABAergic complex-sleep-behavior interaction despite its name, and is generally acceptable to combine.

Part 8: Special Populations

Elderly (65 and older)

This is where eszopiclone earns the most caution.

• Older patients absorb, distribute, and clear the drug differently and are more sensitive to CNS effects, with real risk of excessive next-day sedation, falls, cognitive impairment, and accumulation over nightly use.
• Cap the dose at 2 mg, and start at 1 mg.
• The Beers Criteria flag Z-hypnotics as potentially inappropriate in older adults. Eszopiclone falls into the "use with caution / prefer to avoid" tier.
• Deprescribing is actively recommended in any patient over 65 on a Z-drug for insomnia.
• Better first choices in the elderly: ramelteon, low-dose doxepin (Silenor 3 to 6 mg), or a DORA (suvorexant, lemborexant), all Beers-acceptable and far safer for falls and cognition. Doxepin is uniquely good for the last 2 to 3 hours of sleep (early-morning awakening). Avoid first-generation antihistamines like diphenhydramine (delirium risk) and long-acting benzodiazepines.

Pregnancy

• Human data are limited. There is no clean signal of major teratogenicity, but the evidence base is thin, and neonatal sedation/withdrawal is a theoretical concern with late-pregnancy use.
• CBT-I is first-line in pregnancy. If a pharmacologic agent is truly needed, that decision should be individualized in coordination with obstetrics, weighing the real harms of severe maternal sleep deprivation against limited fetal safety data. Do not use eszopiclone reflexively.

Lactation

Data specific to eszopiclone are sparse. As a small, lipophilic, CNS-active molecule it is expected to enter breast milk. If used, monitor the infant for sedation and poor feeding, and time dosing after a feed to minimize exposure. Agents with better-characterized minimal excretion may be preferable.

Hepatic Impairment

Because eszopiclone is cleared by the liver, severe hepatic impairment markedly reduces clearance. Reduce the dose, generally do not exceed 2 mg, and start at 1 mg. In severe impairment, weigh avoiding it altogether.

Renal Impairment

Elimination is not significantly renal-dependent, so ordinary renal impairment requires no specific dose adjustment. In advanced/end-stage renal disease, general CNS sensitivity rises; use the lower end of dosing and monitor.

Children and Adolescents

Not recommended. There is no established pediatric indication. For pediatric insomnia, behavioral/CBT-I approaches are first-line, and melatonin is the usual pharmacologic fallback. Notably, Z-drugs (both eszopiclone and zolpidem) failed controlled trials for ADHD-associated insomnia.

Part 9: Discontinuation and Deprescribing

Stopping eszopiclone is far less dramatic than stopping lithium or a benzodiazepine (there is no rebound-mania catastrophe and no 7-fold relapse cliff), but it is not nothing. Expect rebound insomnia, and in long-term nightly users, some tolerance and mild withdrawal.

When to Deprescribe

• Anyone over 65 on a Z-drug for insomnia.
• Anyone under 65 who has used it more than about 4 weeks for insomnia.
• Any patient whose underlying driver (depression, anxiety) is now well-controlled.

How to Taper

The general principle is fast at first, slow near the end (the mirror image of how you might intuit it) because the hardest, most rebound-prone stretch is the final approach to zero.

• Reduce by about 25% every 2 weeks while doses are higher.
• As you near the low end, slow to about 12.5% every 2 to 3 months.
• Monitor every 1 to 2 weeks during the taper and fine-tune the pace to the patient's tolerance.
• Go slower in the elderly because of accumulation.

Manage the Landing

• Warn the patient about rebound insomnia before you start tapering: sleep will get temporarily worse, then settle. Patients who aren't warned interpret rebound as "I clearly still need the drug" and abandon the taper.
• Substitute CBT-I, which is as effective as Z-drugs acutely and sustains benefit far better long-term. It is the single best tool for getting, and keeping, patients off hypnotics. Apps like CBT-i Coach lower the access barrier.
• There is no medication with strong evidence for preventing Z-drug withdrawal. Do not simply swap one sedative for another; switching to a different hypnotic to "cover" the taper usually just relocates the dependence.

Part 10: Eszopiclone vs the Alternatives

The honest summary: eszopiclone's advantages are narrow and its cost historically absurd. Know exactly when it beats the field and when it doesn't.

• vs Zolpidem (Ambien IR): Zolpidem clears faster (about 2.5 hours) and causes less next-day impairment but isn't approved for sleep maintenance. Eszopiclone's longer half-life earns the maintenance indication but at the price of morning residue and metallic taste. Claims that eszopiclone causes fewer parasomnias than zolpidem are anecdotal; no head-to-head RCT supports them.
• vs Zolpidem CR (Ambien CR): Both target sleep maintenance; the manufacturer conspicuously never ran a head-to-head trial. Roughly comparable in practice.
• vs Zaleplon (Sonata): Zaleplon's about 1-hour half-life means no morning hangover and it can even be taken in the middle of the night, but it does nothing for sleep maintenance. Choose zaleplon for pure sleep-onset problems and to sidestep CYP interactions; choose eszopiclone when the patient wakes at 3 a.m.
• vs Trazodone (generic): The pragmatic rival. Trazodone (50 to 100 mg) has a similar 7 to 8 hour half-life, comparable real-world insomnia efficacy, no metallic taste, no controlled-substance status, and costs roughly 1/100th as much. Its downsides are orthostatic hypotension (fall risk, especially in the elderly) and rare priapism. For a depressed patient with insomnia and cost concerns, trazodone is often the better opening move, though eszopiclone has the stronger antidepressant-augmentation data.
• vs Temazepam (Restoril): Nearly the same half-life as eszopiclone, about 100 times cheaper, but a benzodiazepine: more abuse liability and faster tolerance to the hypnotic effect. Eszopiclone's edge is a roughly 20-fold lower abuse rate.
• vs Ramelteon (Rozerem): Ramelteon is the safety champion: non-scheduled, no abuse potential, no complex sleep behaviors, no falls, no withdrawal. Its weakness is potency: no GABAergic "knockout punch," and no help for wake-after-sleep-onset. Prefer ramelteon in substance-use histories and the frail elderly; prefer eszopiclone when you genuinely need maintenance efficacy.
• vs DORAs (suvorexant, lemborexant, daridorexant): The orexin antagonists are the modern safety-forward choice: effective for both onset and maintenance, no complex sleep behaviors, no falls increase in the elderly (studied to age 93), no tolerance or withdrawal. Lemborexant even outperformed low-dose zolpidem CR on polysomnography. Their downsides are cost and insurance step-therapy hurdles. Where formulary allows, a DORA is often a better long-term maintenance agent than eszopiclone.
• vs low-dose Doxepin (Silenor 3 to 6 mg): Doxepin is Beers-acceptable, non-scheduled, and uniquely effective for the last 2 to 3 hours of sleep (terminal insomnia/early-morning awakening). For that specific complaint in an older patient, doxepin beats eszopiclone.

So where does eszopiclone actually win? Two places. First, sleep-maintenance insomnia where you want a GABAergic agent's reliable hypnotic effect and a Z-drug's lower abuse liability than a benzodiazepine. Second, depression or GAD with insomnia, where its daytime-active metabolite gives it augmentation data no other Z-drug can match. Outside those, cheaper (trazodone) or safer (ramelteon, DORAs, doxepin) agents usually deserve the first try.

Mechanism: The Short Version

Eszopiclone is the active S-enantiomer of zopiclone, a non-benzodiazepine ("Z-drug") that acts as a positive allosteric modulator at the GABA-A receptor. Like all Z-drugs it binds the benzodiazepine site more selectively than benzodiazepines do (which bind GABA-A subtypes non-specifically), which is why Z-drugs deliver a faster, more sleep-focused effect with somewhat less abuse potential and fewer daytime benzodiazepine effects.

Two pharmacologic facts drive everything clinically distinctive about this drug:

1. Half-life 5 to 7 hours: the longest of the Z-drugs (zaleplon about 1 hour; zolpidem about 2.5 hours), approaching the range of temazepam. This is what gives eszopiclone its sleep-maintenance indication and, unavoidably, its next-day impairment.
2. An active metabolite with benzodiazepine-like daytime activity: this is the leading explanation for why eszopiclone, alone among Z-drugs, shows genuine antidepressant- and anxiolytic-augmentation effects in RCTs. Zolpidem and zaleplon lack a comparable daytime-active metabolite and fail the analogous trials.

Metabolism is hepatic via CYP3A4 (hence the interaction profile); elimination is not meaningfully renal.

The Bedside Cheat Sheet