4.1 Lithium Maintenance: The Gold Standard

Lithium remains the foundational maintenance agent against which all others are measured. It is the only agent universally recommended as first-line maintenance therapy across every major international guideline.

The Geddes meta-analysis. The landmark systematic review by Geddes and colleagues, analyzing data from five RCTs comprising 770 participants, provided definitive evidence of lithium's prophylactic superiority over placebo (Geddes et al., 2004, American Journal of Psychiatry). Lithium significantly reduced the risk of all relapses (RR = 0.65; 95% CI 0.50--0.84), with robust efficacy against manic relapse (RR = 0.62; 95% CI 0.40--0.95) and a smaller, statistically borderline effect against depressive relapse (RR = 0.72; 95% CI 0.49--1.07). Over one year, the average risk of any mood relapse was reduced from 61% on placebo to 40% on lithium, an absolute risk reduction of approximately 21% and an NNT of 5 to prevent one relapse over one year (Geddes et al., 2004, American Journal of Psychiatry).

The anti-suicidal effect. Beyond mood stabilization, lithium possesses a unique and clinically critical anti-suicidal property that is practically absent among other mood stabilizers and antipsychotics. Given that bipolar disorder carries a suicide mortality rate estimated at 20 to 30 times higher than the general population (Plans et al., 2019, Acta Psychiatrica Scandinavica), this property directly influences treatment selection.

The most influential data comes from Cipriani and colleagues' systematic review and meta-analysis of 48 RCTs comprising 6,674 participants (Cipriani et al., 2013, BMJ). Lithium was significantly more effective than placebo in reducing completed suicides, with an odds ratio (OR) of 0.13 (95% CI 0.03--0.66) based on four RCTs (n = 485). An OR below 1.0 means lithium reduces the risk, and an OR of 0.13 indicates a very large protective effect. Lithium also reduced deaths from any cause (OR = 0.42; 95% CI 0.21--0.87). An earlier meta-analysis by the same group found an OR of 0.21 (95% CI 0.08--0.50) for the composite of suicide plus deliberate self-harm (Cipriani et al., 2005, American Journal of Psychiatry). When corrected for exposure time, a separate meta-analytic review found lithium was associated with a 1.65-fold lower risk of suicidal behaviors compared to placebo or other active treatments (Baldessarini & Tondo, 2009, Bipolar Disorders).

It is important to note that subsequent meta-analyses with larger samples have yielded ORs in the same protective direction but with wider confidence intervals that did not reach statistical significance. Riblet and colleagues (2017) found OR = 0.23 (95% CI 0.05--1.02), and a 2022 meta-analysis of seven RCTs (N = 1,138) reported OR = 0.30 (95% CI 0.09--1.02) (Noel-Maynak et al., 2022, BJPsych Open). These findings reflect the fundamental difficulty of studying completed suicide as an endpoint in randomized trials rather than a reversal of the protective signal. The overall direction of evidence favors a protective effect, but the magnitude of that effect remains less certain than the Cipriani 2013 OR of 0.13 suggests.

The clinical implication is direct: for any patient with bipolar disorder who has a history of suicide attempts, recurrent suicidal ideation, or a high-risk profile, the decision to use an agent other than lithium for maintenance requires careful justification against this evidence.

4.2 Valproate Maintenance: The BALANCE Trial and the Monotherapy Disappointment

Valproate was long considered a first-line maintenance agent based on its proven efficacy in acute mania. However, rigorous long-term evaluation has revealed a significant gap between its acute and prophylactic evidence.

The foundational Bowden et al. (2000) trial, a 52-week RCT of 372 patients, produced an unexpected result: divalproex monotherapy did not differ significantly from placebo in the primary outcome of time to recurrence of any mood episode (Bowden et al., 2000, Archives of General Psychiatry).

The BALANCE trial provided the definitive head-to-head comparison (Geddes et al., 2010, The Lancet). This 24-month, randomized, open-label, investigator-sponsored pragmatic trial of 330 patients with bipolar I compared lithium monotherapy, valproate monotherapy, and the combination. The open-label design (clinicians and patients knew which treatment was assigned) is a limitation, as it may introduce performance and detection bias; however, the pragmatic design and investigator (rather than manufacturer) sponsorship strengthen its ecological validity. Over two years, the primary outcome (new intervention for a mood episode or hospitalization) occurred in 54% of the combination group, 59% of the lithium group, and 69% of the valproate group. The key comparisons, expressed as hazard ratios (HR, where values below 1.0 indicate the first treatment is superior, and lower values mean greater superiority):

The combination was significantly superior to valproate monotherapy (HR = 0.59; 95% CI 0.42--0.83; p = 0.0023), representing a 41% risk reduction. Lithium monotherapy was significantly superior to valproate monotherapy (HR = 0.71; 95% CI 0.51--1.00; p = 0.047). The combination was not significantly superior to lithium monotherapy (HR = 0.82; 95% CI 0.58--1.17; p = 0.27) (Geddes et al., 2010, The Lancet).

In practical terms: lithium alone outperformed valproate alone, and adding valproate to lithium provided only a modest further benefit that did not reach statistical significance. Valproate's contribution to the combination was real but smaller than lithium's.

Real-world registry data has confirmed these findings: psychiatric hospitalization rates are significantly higher for patients maintained on valproate compared to lithium (HR = 1.33; 95% CI 1.18--1.48) (Joas et al., 2017, British Journal of Psychiatry).

The clinical takeaway: valproate monotherapy is inferior to lithium for maintenance. Patients who are stable on valproate alone but experiencing frequent relapses should be augmented with lithium or transitioned to lithium monotherapy. Valproate retains a role in combination regimens and may be preferred over lithium in specific situations (rapid cycling presentations, mixed features), but as a standalone prophylactic agent, its evidence is weaker than previously assumed. Its use is further constrained by severe teratogenicity (the capacity to cause birth defects when taken during pregnancy; see Module 6) and the 2024--2025 NICE/MHRA restrictions against its use in women and girls of childbearing potential.

4.3 Lamotrigine Maintenance: Filling the Depressive Gap

Lamotrigine's value in bipolar disorder is primarily a maintenance story. Its acute antidepressant effects are modest (see Section 2.4), but its prophylactic efficacy against depressive relapse is robust and specific.

Two pivotal 18-month RCTs published in 2003 established the evidence. The Calabrese et al. trial, evaluating patients with an index depressive episode, showed that both lamotrigine and lithium significantly prolonged time to intervention compared to placebo, with median survival times of 200 days (lamotrigine), 170 days (lithium), and 93 days (placebo). Lamotrigine was statistically superior to placebo specifically for delaying depressive episodes (p = 0.047). At one year, 57% of lamotrigine patients remained intervention-free for depression versus 46% for lithium and 45% for placebo (Calabrese et al., 2003, Journal of Clinical Psychiatry).

The companion trial by Bowden et al. confirmed the polarity-specific pattern: lithium was superior for preventing manic episodes (p = 0.006), while lamotrigine was superior for preventing depressive episodes (p = 0.02) (Bowden et al., 2003, Archives of General Psychiatry).

Subsequent meta-analyses confirmed this complementary relationship. The Miura et al. (2014) network meta-analysis of 33 RCTs and 6,846 participants established that lamotrigine yields an NNT of 8.3 for preventing any mood episode, driven primarily by its antidepressant prophylaxis. Lamotrigine's adverse event profile was largely indistinguishable from placebo (Miura et al., 2014, The Lancet Psychiatry).

Note on industry funding: The pivotal lamotrigine maintenance trials were sponsored by GlaxoSmithKline.

For patients whose bipolar illness is predominantly characterized by frequent, disabling depressive episodes, lamotrigine maintenance is a first-line strategy. Its weight neutrality, absence of sedation, and minimal cognitive effects make it one of the most tolerable long-term agents available.

A practical note on titration: many patients find the 5-6 week titration period frustrating because they see no therapeutic benefit during this period while taking a medication that has been presented as helpful. Clinicians should set expectations clearly at initiation: "This medication takes 4-6 weeks before you will notice any mood benefit. That does not mean it is not working; it means we are building up the dose slowly to keep you safe. The benefit comes once we reach the therapeutic dose, and the long-term evidence is strong." Titration dropout is a real-world barrier to lamotrigine's effectiveness; proactive expectation management reduces it.

Clinical Pearl: Lithium's Role in Treatment Selection

Lithium is often underused relative to its evidence base, in part because of concerns about monitoring burden (renal function, thyroid, serum levels) and narrow therapeutic index. These concerns are legitimate but should be weighed against the data: lithium is the only mood stabilizer with a consistent protective signal against completed suicide in bipolar disorder (clozapine has similar evidence in schizophrenia, but lithium’s anti-suicide data in mood disorders is unique), it has the strongest overall maintenance evidence of any mood stabilizer, and its NNT of 5 for relapse prevention over one year is among the most favorable in psychiatry. The monitoring requirements, while real, follow a standardized schedule that becomes routine in clinical practice. For a detailed lithium management guide (initiation, therapeutic levels, monitoring protocol, drug interactions, toxicity recognition), see Module 3.

PsychHQ

Maintenance Pharmacotherapy: Lithium, Valproate, Lamotrigine