Prescribing Lithium

Why Lithium Still Matters

Lithium is the oldest drug in modern psychiatry and, by almost every measure that counts, still the best mood stabilizer we have. It is the only psychiatric medication with robust, replicated evidence for preventing suicide. It outperforms valproate, lamotrigine, olanzapine, and quetiapine as monotherapy in bipolar I. It lowers all-cause mortality. It may protect against dementia and myocardial infarction. And it costs pennies a day.

Yet it is steadily falling out of use, not because the evidence turned against it, but because it has no marketing budget, requires blood draws, and carries a reputation for toxicity that dates to the 1940s, when lithium chloride was sold as a salt substitute and killed cardiac patients who used it unmonitored. None of that reputation reflects how lithium behaves when prescribed thoughtfully.

A note on John Cade, who discovered it: in 1949 the Australian psychiatrist was using lithium urate merely as a soluble vehicle to inject uric acid into guinea pigs, suspecting a uric-acid factor in mania. The lithium itself calmed the animals. He tried it in agitated manic patients and watched them settle. Serendipity launched the lithium era, and seventy-five years of accumulated data have only deepened the case for it.

Part 1: Indications: Who Is Lithium For?

FDA-Approved Uses

• Acute mania in bipolar I disorder (monotherapy)
• Maintenance/prophylaxis in bipolar I disorder

Lithium is not FDA-approved as monotherapy for bipolar depression, though it is used off-label and has reasonable supporting data (8 of 9 placebo-controlled studies in a 1993 review were positive).

Bipolar I: The Home Turf

This is where lithium shines, particularly in the classic Kraepelinian presentation: euphoric (not dysphoric) mania, followed by depression, with genuinely well euthymic intervals between episodes. About 27–35% of bipolar patients are "excellent responders" who have no further episodes once stabilized on lithium and adherent. Roughly a third of the bipolar population can essentially have their illness arrested by this drug, a response rate no other agent approaches.

Bipolar II

Less evidence, but used clinically; some data suggest a useful depression-prevention effect. Lamotrigine is often preferred first-line here, but lithium remains a strong option, especially when suicide risk is present.

Treatment-Resistant Depression (Augmentation)

A major, underused indication. Meta-analysis of 69 RCTs found lithium augmentation gave a response RR of 1.25 vs placebo, comparable to aripiprazole (1.57) and quetiapine (1.34). In a network meta-analysis, lithium augmentation in TRD performed similarly to IV ketamine and esketamine, with fewer dropouts. Lithium belongs in your TRD algorithm, not as a last resort.

Suicide Prevention: Lithium's Unique Property

Nothing else in psychiatry does this:

• Meta-analysis of RCTs (2,400 patients): lithium reduces completed suicide by ~60% vs placebo.
• A case-control study of 50,000 patients confirmed the effect is unique to lithium, not shared by other mood stabilizers.
• Over 110,000 person-years of data support it.
• The effect spans both unipolar and bipolar disorder.
• Critically, the anti-suicide effect is independent of mood stabilization: it doesn't track with whether lithium is preventing episodes. This is a distinct pharmacologic property.

Geriatric Depression

Counterintuitively, patients ≥65 with TRD respond better to lithium augmentation than younger patients (68.2% vs 46.9%). The likely explanation: an aging blood-brain barrier becomes more permeable, so adequate brain levels are achieved at lower serum concentrations, and lithium's neuroprotection matters more when neurodegeneration contributes to late-life depression.

Emerging: Dementia Prevention

Three of four epidemiologic studies found lower dementia incidence in mood-disorder patients on lithium. The Forlenza 2019 trial showed patients with mild cognitive impairment maintained at a low serum level (~0.4 mEq/L) preserved cognition vs decline on placebo. Ecological data link trace lithium in drinking water to lower dementia and suicide rates. The mechanism (GSK-3β inhibition, neuroprotection, telomere preservation) is plausible. Not yet practice-changing, but watch this space.

Predictors of Response

Part 2: Before You Start: Workup

Get these labs before starting, but don't delay treatment more than a week waiting for them.

Who Is a Poor Candidate?

• Patients unable or unwilling to do periodic blood draws
• Advanced CKD (eGFR <30)
• Significant cardiac arrhythmia / sino-atrial disease
• Uncontrolled hyperthyroidism
• Conditions causing recurrent severe dehydration
• Psoriasis (may flare; relative contraindication)

Note: eGFR 30–60 is not an automatic disqualifier; see Special Populations.

Part 3: How to Start and Dose

Formulation Choice

Start with generic immediate-release (IR) lithium carbonate. It's cheap, and IR produces a single daily peak rather than constant exposure, which may be gentler on the kidney over time. Extended-release products (Lithobid, Eskalith CR) exist and reduce GI upset for some patients but cost more. Lithium citrate liquid (300 mg carbonate ≈ 5 mL citrate) is invaluable for fine titration and detail-oriented patients.

Starting and Titration

• Start: 300 mg at bedtime (QHS). Aggressive initial dosing produces side effects that drive non-adherence and sour patients on lithium permanently.
• Titrate 300 mg every 3–7 days. A clean schedule: 300 mg × 5 days → 600 mg × 5 days → 900 mg → check level ≥5 days later.
• Rule of thumb: each 300 mg increment raises serum level by ~0.2 mEq/L in a non-geriatric adult. This lets you predict where you'll land.
• Most patients reach a therapeutic level within ~4 weeks.
• Geriatric: start lower (often 150 mg), titrate slower, target ~30% lower levels.

Why Dose at Night

Dose the entire daily amount at bedtime when feasible. Peak levels then occur 1–2 hours later while the patient sleeps, blunting daytime tremor and cognitive dulling. Single nightly dosing is also easier on the kidneys than split dosing. Splitting BID/TID is reserved for GI intolerance.

Target Serum Levels

How to Draw a Level: The Part Patients Get Wrong

• Trough only: 12 hours after the last dose.
• No sooner than 5 days after any dose change (steady state isn't reached before then).
• A non-trough level is uninterpretable. Blood is drawn in the morning, 12 hours after the bedtime dose, before taking that day's dose.

Part 4: Monitoring Schedule

Monitor more frequently if: age >65, clinically unstable, new interacting drug started, or renal function declining.

Action Thresholds

• eGFR <60: consider nephrology referral, but do not reflexively stop lithium. Patients with eGFR 40–60 can often continue on a lower dose with closer monitoring.
• TSH >5–6 mIU/L: start levothyroxine. Don't stop lithium for treatable hypothyroidism.
• Elevated calcium: check PTH. Hyperparathyroidism (~4% of users) is usually asymptomatic but causes osteoporosis and renal stones.
• New palpitations / syncope / dizziness: get an ECG.

Part 5: Side Effects and Management

The governing principle: most lithium side effects are dose- and level-dependent, most improve with time, and almost all are manageable. But you must manage them aggressively and early: side effects, not lack of efficacy, are the usual reason patients quit.

Gastrointestinal (Nausea, Diarrhea)

Common early, and a frequent dealbreaker if left unaddressed.

• Take with or after meals.
• Split the dose BID/TID if needed.
• For nausea: switch to slow-release formulation or lithium citrate liquid.
• For diarrhea: the opposite: switch to IR (extended-release worsens diarrhea).
• Antiemetics: ondansetron 4 mg PRN (preferred, no tardive dyskinesia risk). Alternatives: ginger 1,000–2,000 mg q12h PRN; lorazepam 0.5–1 mg q8h PRN.
• Timing trick: give the antiemetic 30–90 minutes before dinner, then lithium after dinner.

Polyuria / Polydipsia (Nephrogenic Diabetes Insipidus)

Very common. Increased urine output does not mean the kidneys are being damaged. At the extreme, lithium causes nephrogenic DI: the kidney stops responding to ADH, risking dehydration and hypernatremia.

• Educate first: this is expected and not, in itself, harmful.
• Dose entirely at night.
• Stay well-hydrated with water (not sugary drinks).
• Amiloride 5 mg/day is first-line for problematic NDI.
• N-acetylcysteine may offer renal protection.
• HCTZ 12.5 mg/day works but paradoxically raises lithium levels: if used, halve the lithium dose and recheck within a week.

Tremor

A fine, high-frequency tremor (not coarse/Parkinsonian). Management in order:

1. Lower the dose: often resolves it.
2. Propranolol 20 mg BID–TID, or Inderal LA 60 mg QAM (up to ~240 mg/day).
3. Reduce caffeine.
4. Vitamin B6 as pyridoxal-5'-phosphate (P5P), not plain pyridoxine, 150–300 mg BID.
5. Benzodiazepine (clonazepam, alprazolam) if needed.
6. Refractory: primidone 12.5–25 mg QHS titrated to 50–250 mg; topiramate 25–400 mg/day; gabapentin 1,200–3,600 mg/day (least effective).

Cognitive Dulling

Real, dose-dependent, and a common quiet reason for non-adherence.

• Check the TSH: hypothyroidism mimics this exactly.
• Dose QHS only.
• Lower the dose if significant.
• Reassure: long-term, lithium preserves cognition better than antipsychotics and may prevent dementia. The acute dulling and the long-term neuroprotection are different phenomena.

Endocrine

• Hypothyroidism: 10–20% of long-term users; ~1/40 per year. Treat with levothyroxine: keep lithium, treat the thyroid.
• Hyperparathyroidism: ~4%, mostly asymptomatic; screen calcium annually.

Weight

Weight gain in ~25% of patients (range 4.5–12 kg long-term), via fluid retention, appetite, subclinical hypothyroidism, and thirst-driven caloric beverages. Among mood stabilizers, lithium has among the lowest weight-gain risk (second only to lamotrigine).

• If on an antipsychotic, trim the antipsychotic first: it's usually the bigger culprit.
• Water, not sugared drinks.
• Pretreatment: metformin ER 500–1,000 mg/day; melatonin 3–5 mg QHS.

Less Common

• Dermatologic: acne, psoriasis flare (may force discontinuation), hair thinning/coarsening.
• Cardiac: rarely sinus node dysfunction. Lithium does not prolong QT. ECG only for cardiac history or symptoms.
• Neurologic, rare: taste changes, muscle weakness, pseudotumor cerebri (headache, papilledema, rare but serious), sleepwalking.

Part 6: Toxicity

Clinical Progression

1. Early: nausea, vomiting, diarrhea, coarsening tremor.
2. Progressive: confusion, ataxia, slurred speech.
3. Severe: seizures, encephalopathy, renal failure, coma.

What Precipitates Toxicity

Dehydration, NSAIDs, thiazides, ACE inhibitors/ARBs, acute renal injury, low-sodium states, and dose errors. Sodium and water are lithium's chemical cousins: anything that contracts volume or depletes sodium makes the kidney hoard lithium.

The Renal Question: What the Data Actually Show

The old fear was that lithium inevitably destroys kidneys. Modern data reframe this completely. A Swedish cohort of ~11,000 patients comparing lithium to valproate found:

• No significant difference in CKD progression (~3.5% in each group).
• No difference in non-CKD eGFR decline or albuminuria.
• AKI risk was actually 3.2% lower with lithium than valproate.
• But: at levels >1.0 mmol/L, CKD risk nearly tripled; at >0.8 mmol/L, AKI risk rose significantly (HR 2.56).

The bottom line: keep levels ≤0.8 mEq/L, dose once nightly, avoid toxic spikes, keep patients hydrated, and lithium's renal risk is modest, comparable to valproate.

Lithium's Medical Benefits (Often Forgotten)

• Mortality: In 25,787 hospitalized bipolar patients, lithium showed the most profound mortality reduction of any mood stabilizer, aHR ~0.37–0.39 for all-cause, suicide, and natural-cause death.
• Cardiac: Associated with lower MI risk and lower cardiovascular death.
• Stroke: No significant effect, whereas carbamazepine increases overall stroke and valproate increases hemorrhagic stroke.

Part 7: Drug Interactions

Because the therapeutic window is narrow, interactions matter more than with almost any other psychotropic.

Part 8: Special Populations

Pregnancy

The reputation is worse than the reality. The classic fear is Ebstein's anomaly (a tricuspid valve defect):

• Relative risk ~10-fold over baseline.
• Absolute risk only ~1 in 1,000–2,000 (0.05–0.1%), vs ~1 in 20,000 baseline.
• This is far safer than valproate (~5% neural tube defect risk).

Against this sits the danger of untreated illness: relapse rates in pregnancy reach ~71%, and stopping lithium roughly doubles relapse risk.

Breastfeeding

Most authorities discourage it. If a woman chooses to breastfeed, monitor the infant (lithium level, TSH, BUN, creatinine, electrolytes) immediately postpartum, at 4–6 weeks, then every 8–12 weeks, in coordination with pediatrics.

Geriatric (>65)

• Start lower (often 150 mg), titrate slower, target ~30% lower serum levels.
• Monitor more frequently: renal and thyroid decline with age, and polypharmacy raises interaction risk.
• But remember the upside: older TRD patients respond better (68.2% vs 46.9%), and medical mortality is lower on lithium.

Renal Impairment

• eGFR >60: use normally; monitor annually.
• eGFR 30–60: don't reflexively stop. Many patients hold stable eGFR for years at a lower dose. Weigh psychiatric severity; co-manage with nephrology.
• eGFR <30: generally contraindicated.

Part 9: Discontinuation: Handle With Extreme Care

How to Taper

• Minimum taper: 30 days. A reasonable pace is 300 mg per month.
• Long-term users: taper over 6–12 months, slowing further near the end.
• Monitor closely throughout. Have a plan to resume promptly if symptoms emerge.

The message to give patients: "It's far easier to prevent a relapse than treat an episode once it starts, so we taper slowly, and we never stop on our own."

A meta-analysis of 22 RCTs found 6-month relapse of 32.3% on maintenance vs 52.7% on discontinuation (RR ~0.61 for any episode; mania most protected, RR 0.45). Patients are roughly twice as likely to stay well if they keep taking it.

Part 10: Lithium vs the Alternatives

In the Finnish nationwide cohort (18,000 patients, 1987–2012), lithium had the lowest risk of all-cause rehospitalization (HR 0.71) and psychiatric rehospitalization (HR 0.67): the single most effective treatment studied. Benzodiazepines actually increased rehospitalization risk.

Mechanism: The Short Version

Lithium's full mechanism remains incompletely understood, but key actions include: inhibition of the inositol monophosphate/IP3 signaling pathway; modulation of G-protein-coupled second-messenger cascades; inhibition of GSK-3β (central to its neuroprotective and possibly anti-dementia effects); and effects across serotonin, dopamine, and GABA systems. It is neuroprotective (telomere preservation, anti-apoptotic), anti-inflammatory, and has documented antiviral activity against herpes-family viruses.

The fact that its anti-suicide effect is independent of mood stabilization strongly suggests at least two distinct mechanisms operating in parallel, which fits its efficacy across both unipolar and bipolar illness.

Bedside Cheat Sheet