Prescribing Memantine

Why Memantine Matters

Memantine is the quiet one in the dementia cabinet. It doesn't reverse Alzheimer's, it doesn't halt it, and it doesn't produce the kind of visible "lift" that families hope for. It slows the rate of cognitive decline by a modest, statistically real margin in moderate-to-severe disease. The trials ran six months, and whether the benefit holds beyond that has not been established. In a disease where the therapeutic options are thin, a well-tolerated drug that slows decline, calms agitation, and lifts apathy earns its place.

Two things make memantine worth a full prescribing guide rather than a footnote. First, it is mechanistically distinct from everything else in dementia care. The cholinesterase inhibitors (donepezil, rivastigmine, galantamine) push acetylcholine up. Memantine works on a completely different system: it is a low-to-moderate-affinity NMDA receptor antagonist that dampens the pathologic glutamate excitotoxicity thought to drive neuronal injury. Because the mechanisms don't overlap, the two classes combine cleanly and additively, so "cholinesterase inhibitor first, add memantine as disease progresses" is standard practice.

Second, and increasingly relevant to general psychiatry, memantine has a real off-label track record outside dementia. The most striking result in the memantine literature comes from a 2023 randomized trial in trichotillomania and skin-picking, where memantine produced a 60.5% response versus 8.3% on placebo, an NNT of about 2. It has positive augmentation data in OCD (five of six small RCTs). The glutamatergic modulation that slows excitotoxicity in dementia appears to do something useful in the compulsive-spectrum disorders too.

Memantine (Namenda) was approved in 2003, the first new mechanism in dementia since the cholinesterase inhibitors, and for a long time the last. The extended-release form arrived in 2010, and the fixed combination with donepezil (Namzaric) in 2014. It has been off patent for years, is cheap, and, unlike much of what we prescribe, carries no boxed warning.

Part 1: Indications

FDA-Approved Use

• Moderate-to-severe Alzheimer's disease. That's the whole label. Not mild AD, not mild cognitive impairment.

The severity restriction reflects the evidence. A Cochrane review of 44 RCTs (about 10,000 patients) found no benefit over placebo in mild AD. Memantine's signal appears only once the disease is moderate to severe. Prescribing it for mild cognitive impairment or mild dementia is common in practice, but it is unsupported, and, paradoxically, side effects (confusion, somnolence) tend to be worse in milder disease. Don't reach for memantine early.

The Evidence-Based Clinical Picture

Moderate-to-severe Alzheimer's. Across six-month RCTs, memantine produces a small but statistically significant benefit versus placebo on cognition, activities of daily living, and behavior. Trials in this population measure cognition with the Severe Impairment Battery, a 100-point scale where higher is better, rather than the ADAS-Cog. Pooling up to 14 trials and about 3,700 patients, the Cochrane review puts the gain at 3.11 SIB points over placebo (95% CI 2.42 to 3.92), with 1.09 points on an activities-of-daily-living scale and 1.84 points on the Neuropsychiatric Inventory. Cochrane grades the effect small but clinically detectable, so it sits right at the threshold of what a clinician or family will perceive.

Behavioral and psychological symptoms. In practice, memantine often earns its keep here. It can reduce agitation, aggression, and psychotic behavior, and it lifts apathy, improving motivation, alertness, and initiative, sometimes even when the cognitive needle doesn't move. The behavioral benefit frequently precedes and exceeds the cognitive one. For a family exhausted by an agitated, apathetic patient, that can matter more than an ADAS-Cog point. (Caveat: the agitation data lean on pooled and retrospective analyses more than clean prospective trials, so it's a reasonable expectation, not a guarantee.)

Combination with a cholinesterase inhibitor. The standard and best-supported use. In the Tariot 2004 trial (JAMA; N=404, moderate-to-severe AD already on donepezil), adding memantine beat adding placebo on cognition, ADLs, and behavior. The mechanisms are complementary, so this is additive, not redundant.

Other dementias (off-label). Some evidence supports use in Parkinson's disease dementia, Lewy body dementia, and vascular dementia. None are FDA-indicated, but memantine is used off-label here under the pragmatic "my-grandmother rule": given how few tools exist, clinicians reasonably stray beyond the label when the alternative is nothing.

The Off-Label Psychiatric Niche

Memantine is becoming a glutamatergic augmentation tool in the compulsive spectrum. It is not a primary psychiatric drug the way an SSRI is, but it has earned a narrow place in a psychiatrist's toolkit as a cognitive and glutamatergic adjunct:

• Trichotillomania and excoriation (skin-picking) disorder. The standout evidence. Grant et al. (Am J Psychiatry 2023; N=100, 8 weeks, 20 mg/day): 60.5% response on memantine vs 8.3% on placebo, an NNT of roughly 2, an unusually large effect size for a psychiatric intervention. If you treat body-focused repetitive behaviors and have exhausted the usual approaches, this is a result to know about.
• OCD augmentation. Memantine improved OCD in five of six small RCTs, typically added to an SSRI, with benefit emerging within 8 weeks. Dosing mirrors the dementia range (10 to 20 mg/day).
• Occasionally explored as an adjunct in anxiety-spectrum and autism-related behavioral symptoms, though the evidence there is thinner and more preliminary.

Where Memantine Does NOT Help, or Harms

• Mild AD / mild cognitive impairment: no benefit vs placebo (Cochrane, 44 RCTs). Don't use it.
• Frontotemporal dementia: may worsen symptoms, the opposite of its Alzheimer's effect. Both memantine and cholinesterase inhibitors can be detrimental in FTD; prefer SSRIs/SNRIs for behavioral symptoms there.
• Cognitive deficits in schizophrenia: off-label trials showed little to no benefit over placebo.

Part 2: Workup and Candidacy

Memantine has none of lithium's monitoring apparatus. There are no drug levels, no mandated baseline labs, no required ECG. The workup here is the dementia workup, not a memantine workup.

The Relevant Baseline

• Confirm the diagnosis and stage. Memantine is for moderate-to-severe Alzheimer's. Before committing, run the standard dementia evaluation to exclude reversible causes and clarify the syndrome: MRI, CBC, electrolytes, thyroid panel (TSH), B12, and vitamin D. This is diagnostic due diligence, not memantine surveillance.
• Establish a cognitive/functional baseline you can track: an MMSE or MoCA, and ideally a note on ADLs and behavioral symptoms, so you have something to measure the drug against in 6 to 12 months.
• Ask about renal function. The one pharmacologic thing that changes your memantine prescribing is the kidney (see Part 8). A recent estimate of renal function (eGFR/creatinine) is worth having, especially in the elderly, because severe renal impairment mandates a dose reduction.

Who Is a Poor Candidate?

• Patients with mild AD or MCI (no benefit).
• Patients with frontotemporal dementia (may worsen).
• The label's only contraindication is known hypersensitivity to memantine or to any excipient in the formulation. Beyond that, caution (not prohibition) applies to severe renal impairment (dose-adjust), a history of seizures (theoretical), and baseline confusion that memantine might transiently worsen during titration.

Part 3: How to Start and Dose

Formulations

• Namenda IR (immediate-release): 5 mg and 10 mg tablets. The preferred starting choice: cheap, generic, and well tolerated.
• Namenda XR (extended-release): 7, 14, 21, 28 mg capsules; once-daily. Reserve for patients who don't tolerate IR or for whom once-daily dosing meaningfully improves adherence. Some patients have fewer GI complaints on XR.
• Namzaric (fixed combination): memantine XR plus donepezil, in 7 mg/10 mg, 14 mg/10 mg, 21 mg/10 mg and 28 mg/10 mg capsules. For patients already on both agents who struggle with pill burden. Note the equivalence: 14 mg XR is approximately 10 mg IR.

Starting Dose and Titration

You titrate memantine slowly to blunt the dizziness, somnolence, and transient confusion that come with too-fast escalation. The standard four-week schedule for IR:

• Maximum: 20 mg/day (10 mg BID) for IR, or 28 mg once daily for XR.
• The four-week titration is not optional: increment weekly by no more than 5 mg (IR) or 7 mg (XR). This is the single most effective thing you can do to keep the drug tolerable.
• XR: start 7 mg once daily, escalate by 7 mg weekly to a target of 28 mg once daily.

Off-Label Dosing (Psychiatric Uses)

• Trichotillomania / excoriation: 10 mg/day for week 1, then 20 mg/day, the regimen from the Grant 2023 trial (8 weeks). These are younger, cognitively intact patients (trial average age about 31), who generally tolerate 20 mg well.
• OCD augmentation: 10 to 20 mg/day, added to an SSRI; give it up to 8 weeks to declare benefit.

Restarting

If memantine was stopped and is being restarted after a brief gap, you can generally resume at the established target dose without re-titrating. After a longer interruption, re-titrate to be safe.

Part 4: Monitoring

This is the shortest monitoring section you will ever read for a psychiatric drug.

No routine labs, no drug levels, no ECG, no CBC or LFT surveillance mandated by the drug itself. What you monitor is clinical response and tolerability.

What to Watch

• Tolerability during titration: dizziness, somnolence, confusion. These are your rate-limiters; slow the titration or hold if they emerge.
• Efficacy over 6 to 12 months (a longer horizon than cholinesterase inhibitors). Track:
  • Cognition: MMSE/MoCA trajectory. The trials measured a slowing of decline over six months rather than a halt to it. You are tracking a slowed slope, not improvement.
  • Function (ADLs): often shows more benefit than pure cognition.
  • Behavior: agitation, aggression, apathy, initiative. Frequently the most visible gain.
  • Caregiver burden and quality of life: sometimes the outcome that matters most.

The Decision Point

If there is no clear slowing of decline and no behavioral benefit after roughly 6 to 9 (up to 12) months, reconsider: either try a combination approach (add/optimize a cholinesterase inhibitor) or trial the patient off (see Part 9). Because the benefit is modest and the disease progresses regardless, memantine is a drug you should periodically re-justify rather than continue on autopilot.

Part 5: Side Effects and How to Manage Them

Memantine is the mirror image of lithium here: well tolerated, with side effects that are mostly mild, dose-related, and front-loaded during titration, and that usually settle with continued dosing. It causes fewer GI problems than the cholinesterase inhibitors; its main liabilities are CNS, not gut.

CNS

• Confusion / somnolence: the most common adverse events. In the pivotal combination trial, confusion ran 7.9% on memantine vs 2.0% on placebo. Worse with fast titration and, paradoxically, in milder disease.
  • Management: slow the titration, split/lower the dose, and if significant sleepiness or confusion persists, reduce or discontinue. Reassure that transient confusion during escalation often resolves.
• Dizziness: common, dose-related, usually mild.
  • Management: slower titration, fall precautions (this is an elderly, often frail population; a fall is a bigger threat than the dizziness itself).
• Headache: one of the label's common reactions, 6% versus 3% on placebo; usually self-limited.

Gastrointestinal: Comparatively Mild

• Constipation: the most prominent GI effect. Manage with fiber, fluids, and standard bowel regimen.
• Nausea/vomiting: less common than with cholinesterase inhibitors; manage with timing and food if it occurs.

The Behavioral "Side Effects" That Are Benefits

These are the flip side of the ledger: memantine can decrease apathy and increase motivation, alertness, and initiative, and reduce agitation and aggression. In a dementia population they are often the most valued effects.

Rare / Serious

From the deprescribing and pharmacovigilance literature (uncommon): tremor, bradycardia, urinary problems, seizures, and rarely GI hemorrhage or rhabdomyolysis. Weight loss can occur, though it is more associated with cholinesterase inhibitors than with memantine. Psychotic symptoms are a theoretical concern but are more tied to the related agent amantadine; memantine carries lower risk.

Part 6: Overdose and Toxicity

• No boxed warning. Despite two decades of use in a frail elderly population, memantine carries none.
• Relatively safe in overdose. The label's overdose picture is largely an amplification of the known effects: agitation, confusion, dizziness, lethargy, restlessness, somnolence, unsteady gait and vomiting, with stupor, loss of consciousness, coma, psychosis and ECG changes at the severe end. In the largest documented ingestion, 2 grams, the patient had coma, diplopia and agitation, and recovered. Management is supportive.
• Wide margin in practice. In the off-label trials, the main tolerability-limiting event was dizziness (a couple of dropouts in the trichotillomania study), with no serious adverse events overall.

Compared with lithium's narrow window and dialysis-territory toxicity, memantine sits at the opposite end of the spectrum: this is a drug you can prescribe to a cognitively impaired patient without lying awake about an accidental double dose. That said, as with any drug in a confused patient, supervise administration and secure the supply.

Part 7: Drug Interactions

Memantine is one of the cleanest drugs in psychiatry for interactions, because it is largely not metabolized by the liver. Clearance is predominantly extrahepatic and renal, and it neither meaningfully inhibits nor induces the major CYP450 enzymes.

What This Means Clinically

• Minimal CYP-based interactions. You can combine memantine with hepatically metabolized drugs, antidepressants and most psychotropics, without the CYP juggling that donepezil and galantamine require. (Those two are CYP2D6/3A4 substrates, so SSRIs/duloxetine can raise their levels; memantine sidesteps this. Rivastigmine, also non-hepatic, shares memantine's clean profile.)
• Safe with cholinesterase inhibitors (donepezil, rivastigmine, galantamine), which is the whole basis of combination therapy. No washout needed when switching agents.
• Safe with SSRIs (escitalopram, sertraline, citalopram) and other psychiatric medications, which matters given the OCD/trichotillomania augmentation use.
• No problematic interaction with lecanemab (the anti-amyloid monoclonal antibody).

The One Interaction Mechanism to Remember

Because memantine is renally cleared and is a cation excreted partly by the kidney's tubular transport system, drugs that alkalinize the urine reduce its renal clearance and can raise memantine levels. Practically, this means agents like carbonic anhydrase inhibitors (e.g., acetazolamide) and sodium bicarbonate, and, more broadly, any state that markedly alkalinizes the urine, can increase memantine exposure. It's rarely a clinical crisis given the drug's wide margin, but if a patient on memantine becomes more confused or dizzy after such a drug is added, this is the mechanism to think of. Similarly, other renally cleared cationic drugs that share the same tubular transport can theoretically interact; monitor clinically.

Part 8: Special Populations

The Elderly

This is the population. Dementia trials enrolled patients in their 70s and 80s, and memantine is well tolerated among them.

• "Start low, go slow" applies, but finish the titration. The four-week escalation is still required; the mistake in the elderly is stopping at a subtherapeutic dose out of excess caution. Get to the target dose.
• Age alone does not mandate a dose reduction, but confusion, dizziness, and somnolence may be more pronounced, so watch closely and keep fall precautions front of mind.
• The apathy/motivation benefit is often especially valuable here.

Renal Impairment

Memantine is renally cleared, so this is the one special population that changes your dose.

• Normal-to-moderate renal function: use standard dosing (target 20 mg/day IR or 28 mg/day XR).
• Severe renal impairment: reduce the target dose. The commonly used target in severe renal impairment is 10 mg/day (IR) or the correspondingly reduced XR dose (around 14 mg/day). Because a large fraction of memantine is eliminated unchanged by the kidney, impaired clearance raises exposure and, with it, the risk of confusion and dizziness.
• In the elderly, age-related renal decline is exactly why a periodic check of renal function is worthwhile even though no drug level exists to monitor.

Hepatic Impairment

Minimal hepatic metabolism means little to no adjustment is needed in liver disease, one of the few psychotropics you can use comfortably here (a profile it shares with rivastigmine).

Pregnancy and Lactation

Solid human safety data are lacking. Given that memantine's population is overwhelmingly elderly, this rarely arises, but for the off-label psychiatric uses in younger patients (trichotillomania, OCD), it matters. There is no established safety in pregnancy or lactation, so it is generally avoided in these settings absent a compelling indication and a careful risk/benefit discussion. Don't reach for memantine as an OCD augmenter in a patient who is pregnant or planning pregnancy without that conversation.

Pediatrics

Not indicated for dementia (pediatric dementia is all but nonexistent). Off-label psychiatric use in children is minimal and not standard; the trichotillomania evidence is in adults.

Part 9: Discontinuation

Unlike lithium, memantine has no meaningful withdrawal syndrome, no dependence, and no rebound. You are not going to precipitate a crisis by stopping it.

Switching Between Dementia Agents

• You can switch from donepezil to memantine safely, either abruptly or by cross-titration, with no mandatory washout (Waldemar et al., 2008).

Deprescribing

Because the benefit is modest and the disease progresses regardless, memantine is a legitimate candidate for a trial off when:

• there's no clear benefit after an adequate trial,
• the patient declined during treatment anyway, or
• the patient has reached end-stage dementia where the drug no longer serves a purpose.

How to stop: although abrupt cessation is not dangerous, deprescribing guidelines suggest a gradual taper, reducing by 25 to 50% every 1 to 2 weeks, partly to watch for any decompensation in cognition or behavior that would argue for restarting, and to make attribution clear. Optimize non-pharmacologic supports (music therapy, behavioral and environmental approaches) around any discontinuation.

Part 10: Memantine vs the Alternatives

vs Cholinesterase Inhibitors (donepezil, rivastigmine, galantamine)

These two classes are not rivals; they complement each other. They are the two pillars of symptomatic dementia treatment, and the standard approach uses them in sequence and then together.

• Mechanism: cholinesterase inhibitors raise synaptic acetylcholine; memantine blocks pathologic NMDA/glutamate activity. Different systems, so the benefits add up.
• Indication span: cholinesterase inhibitors are approved across mild-to-moderate (and donepezil into severe) disease; memantine is moderate-to-severe only.
• Efficacy: both slow decline, neither reverses it. Memantine's pooled benefit in moderate-to-severe disease is 3.11 SIB points over placebo at six months. The two classes were tested on different scales in different stages of disease, so their effect sizes do not line up side by side.
• Tolerability, the practical differentiator: cholinesterase inhibitors cause more GI upset (nausea, vomiting, diarrhea), vivid dreams, and bradycardia (occasionally symptomatic). Memantine causes more dizziness, somnolence, and confusion but spares the gut. Memantine has fewer dropouts for side effects.
• The standard sequence: start a cholinesterase inhibitor (donepezil is the usual first choice, approved across all stages); add memantine once the disease reaches moderate-to-severe. The Tariot 2004 combination trial (N=404) and the pivotal Namzaric trial (Grossberg 2013, N=677) both show the combination beats a cholinesterase inhibitor alone on cognition, with the honest caveat that ADL differences were not always significant.

vs High-Dose Donepezil (Aricept 23 mg)

The 23 mg donepezil dose beat 10 mg on only one of two required outcome scales (a 2.2-point difference on a 100-point scale) while causing substantially more nausea and vomiting; its FDA approval was controversial (the review team recommended against it). On risk-benefit, memantine looks equal or better than pushing donepezil to 23 mg.

vs Lecanemab (Anti-Amyloid Monoclonal Antibody)

Lecanemab reduces amyloid and slows decline about 27% in early AD, a different disease stage and a different ambition, but at roughly $26,500/year plus imaging, with a 12.6% rate of amyloid-related imaging abnormalities (cerebral edema/hemorrhage) requiring APOE testing and serial MRIs. Memantine offers a modest benefit at trivial cost with a benign safety profile and no monitoring. For most moderate-to-severe patients, the traditional cholinesterase-inhibitor-plus-memantine approach remains safer, cheaper, and reasonable.

vs Amantadine and Ketamine

Clinicians conflate memantine with the other NMDA-active drugs, so it helps to set them side by side:

• Amantadine is a related adamantane that is more dopaminergic and a stronger NMDA blocker; it carries higher risk of psychotic symptoms. Memantine is the more selective, lower-affinity NMDA antagonist.
• Ketamine is the instructive contrast. Both hit the NMDA receptor, but ketamine is a high-affinity, use-dependent channel blocker producing profound, rapid, dissociative and antidepressant effects. Memantine has no rapid antidepressant effect and no dissociative punch, precisely because its low-to-moderate affinity lets it blunt pathologic excitotoxic glutamate signaling without blocking normal synaptic transmission. That selectivity is the whole design: it is why memantine is neuroprotective-by-dampening rather than psychoactive, and why you can give it daily to a demented 82-year-old without inducing a ketamine-like state.

Where It Fits Outside Dementia

Outside dementia, memantine's competitors are the SSRIs, N-acetylcysteine, and habit-reversal therapy for body-focused repetitive behaviors, and SSRI/antipsychotic augmentation for OCD. Its glutamatergic mechanism gives it a different lever, and in trichotillomania/excoriation, the effect size (NNT approximately 2) is larger than most of the alternatives. It belongs on the shortlist for treatment-resistant compulsive-spectrum presentations.

Mechanism

Memantine is a low-to-moderate-affinity, uncompetitive, use-dependent NMDA receptor antagonist. In Alzheimer's disease, chronically elevated glutamate produces a low-level, tonic overactivation of NMDA receptors, a "noisy" excitotoxic signal that both damages neurons and drowns out the phasic glutamate bursts that carry real synaptic information. Memantine's trick is its affinity and kinetics: it binds loosely enough, and comes off the channel fast enough, that it blocks the pathologic tonic overactivation while permitting the normal phasic transmission needed for learning and memory. That is the pharmacologic reason it protects without the dissociation and cognitive blockade of high-affinity NMDA antagonists like ketamine.

Downstream, this is thought to reduce glutamate-mediated excitotoxicity, with secondary effects that may include AMPA-receptor-mediated signaling and modulation of dopamine systems (plausibly underlying the improvements in apathy, alertness, and initiative). Speculative work borrowed from the ketamine literature invokes BDNF and reversal of stress-related changes, but that remains hypothesis, not established fact. The clinically important point is settled: memantine slows the rate of decline; it does not halt or reverse the disease.

Pharmacokinetically, the features that define everyday prescribing are: a long half-life (on the order of 60 to 80 hours), which supports once-daily XR dosing and smooth levels; minimal hepatic metabolism (hence the clean CYP interaction profile); and predominantly renal elimination of unchanged drug (hence the renal dose adjustment and the urinary-pH interaction).

The Bedside Cheat Sheet