Canadian Network for Mood and Anxiety Treatments (<scp>CANMAT</scp>) and International Society for Bipolar Disorders (<scp>ISBD</scp>) 2018 guidelines for the management of patients with bipolar disorder

This is not a test of a single intervention. It is a systematic review and consensus-based guideline that evaluated the entire published literature on pharmacological and psychological treatments for bipolar disorder. Specifically, the guideline authors:

Reviewed all randomised controlled trials (RCTs), meta-analyses, and systematic reviews published up to 2017 on treatments for bipolar I disorder (acute mania, acute depression, maintenance) and bipolar II disorder.

Ranked treatments into first-line, second-line, and third-line categories based on:

- **Level of evidence for efficacy** (e.g., multiple double-blind RCTs vs. single open-label study)

- **Clinical support** (expert consensus on real-world usefulness)

- **Safety and tolerability** (side effect profiles, dropout rates in trials)

- **Treatment-emergent switch risk** (the risk that a drug used for depression will trigger a manic episode)

Created a novel **hierarchical ranking** within first- and second-line categories for bipolar I disorder, considering each treatment's impact across all phases of illness (mania, depression, maintenance) rather than just one phase.

The comparators were placebo, active drug comparators (e.g., lithium vs. quetiapine), and combination therapies (e.g., lurasidone plus lithium or divalproex). Outcome measures included standardised rating scales for mania (Young Mania Rating Scale, YMRS) and depression (Montgomery-Åsberg Depression Rating Scale, MADRS; Hamilton Depression Rating Scale, HAM-D), as well as relapse rates during maintenance.

This guideline synthesises data from hundreds of clinical trials involving tens of thousands of patients. The populations studied across the included trials were:

**Adults (18–65 years)** diagnosed with bipolar I disorder (most common) or bipolar II disorder (less data available)

**Specific subpopulations** addressed separately: women at various stages of the reproductive cycle (pregnancy, postpartum, breastfeeding), children and adolescents, and older adults (>65 years)

**Patients with comorbidities**: substance use disorders, anxiety disorders, and metabolic disorders (obesity, diabetes, dyslipidaemia)

**Exclusion criteria typical of the underlying trials**: patients with rapid cycling (≥4 episodes/year) were often excluded from some trials; patients with acute suicidality or unstable medical conditions were typically excluded

The guideline does not report a single sample size because it is a synthesis. However, the key trials cited individually ranged from ~100 to ~500 patients per arm, and meta-analyses included 10–30+ trials each.

The guideline authors did not conduct new measurements. Instead, they evaluated existing studies that used standardised, validated instruments:

**For acute mania**: Young Mania Rating Scale (YMRS, 0–60 scale, higher = more severe mania). Response typically defined as ≥50% reduction from baseline. Remission typically defined as YMRS ≤12.

**For acute depression**: Montgomery-Åsberg Depression Rating Scale (MADRS, 0–60 scale, higher = more severe depression) or Hamilton Depression Rating Scale (HAM-D, 17-item version, 0–52 scale). Response = ≥50% reduction. Remission = MADRS ≤10 or HAM-D ≤7.

**For maintenance**: Time to relapse (any mood episode), time to manic relapse, time to depressive relapse, and proportion of patients remaining euthymic (stable) at 6, 12, and 24 months.

**For safety and tolerability**: Discontinuation rates due to adverse events, weight gain (kg), metabolic parameters (glucose, lipids), thyroid function, renal function, and incidence of treatment-emergent affective switch (transition from depression to mania/hypomania).

**For bipolar II disorder**: Same depression scales, but mania/hypomania assessed with YMRS or clinical interview.

The guideline also considered "clinical support" ratings, which were based on expert consensus surveys (not a single instrument) where clinicians rated how useful a drug was in real-world practice, beyond what RCTs show.

### Study design

This is a **clinical practice guideline** developed using a formal consensus process. It is not a primary study (RCT, cohort, etc.) but rather a systematic synthesis of existing evidence combined with expert opinion.

### How the evidence was graded

The authors used a standardised evidence-grading system:

**Level 1**: Meta-analysis or ≥2 double-blind RCTs showing efficacy vs. placebo, with adequate sample sizes

**Level 2**: ≥1 double-blind RCT showing efficacy vs. placebo, or ≥1 RCT showing non-inferiority to an established active comparator

**Level 3**: Open-label trials, case series, or post-hoc analyses

**Level 4**: Expert opinion, clinical experience, or extrapolation from other populations

Treatments were then assigned to **first-line** (Level 1 evidence + good safety/tolerability + low switch risk), **second-line** (Level 2 evidence or Level 1 with significant tolerability issues), or **third-line** (Level 3 evidence or Level 2 with poor tolerability).

### The novel hierarchy

New to the 2018 guidelines, the authors created a **hierarchical ranking** within first- and second-line treatments for bipolar I disorder. This was done by considering each drug's efficacy across all three phases (mania, depression, maintenance) simultaneously. For example, a drug that works well for mania but poorly for depression (or increases depression risk) would be ranked lower than a drug that works well for all phases. This was a qualitative consensus process, not a quantitative meta-analytic ranking.

### Duration of evidence

The included trials varied widely:

**Acute mania trials**: Typically 3–4 weeks

**Acute depression trials**: Typically 6–8 weeks

**Maintenance trials**: Typically 6–24 months (most common: 12 months)

### What this design can and cannot prove

**What it can prove:**

It provides a comprehensive, evidence-based summary of which treatments have the strongest empirical support for each phase of bipolar disorder.

It identifies gaps in the evidence (e.g., limited data for bipolar II disorder, children, older adults).

It offers a practical clinical framework that balances efficacy with safety and tolerability.

**What it cannot prove:**

It cannot prove that one treatment is definitively "better" than another in an individual patient. The hierarchy is population-level, not personalised.

It cannot account for individual genetic, metabolic, or psychological differences that might make a second-line drug work better for a specific person than a first-line drug.

The consensus process introduces subjectivity. Expert opinion may be influenced by personal experience, publication bias, or cultural prescribing patterns.

The guideline is static (published in 2018) and does not incorporate newer evidence published after the literature search cutoff.

### Major methodological weaknesses

**Industry funding**: Many of the RCTs cited were funded by pharmaceutical companies. The guideline authors attempted to mitigate this by including only published, peer-reviewed data, but publication bias (positive results are more likely to be published) remains a concern.

**Lack of patient-reported outcomes**: The guideline focuses on clinician-rated scales and relapse rates. It does not systematically include quality of life, functional outcomes, or patient satisfaction.

**Limited data on long-term side effects**: Many maintenance trials are only 12 months long. Long-term metabolic, renal, and cognitive effects of newer drugs (e.g., asenapine, cariprazine) are not well characterised.

**Bipolar II disorder is understudied**: The recommendations for bipolar II are largely extrapolated from bipolar I data or based on small trials. The evidence base is much weaker.

### Acute mania (bipolar I disorder)

**First-line treatments (hierarchical order):**

1. **Lithium** – Multiple RCTs show superiority to placebo. Response rates ~50–60% at 3–4 weeks. Number needed to treat (NNT) for response ≈ 5–7.

2. **Quetiapine** – Monotherapy effective. Response rates ~55–65%. NNT ≈ 4–6.

3. **Divalproex (valproate)** – Effective, but response rates slightly lower than lithium/quetiapine (~45–55%). NNT ≈ 6–8.

4. **Asenapine** – Effective. Response rates ~45–55%. NNT ≈ 6–8.

5. **Aripiprazole** – Effective. Response rates ~45–55%. NNT ≈ 6–8.

6. **Paliperidone** – Effective. Response rates ~45–55%. NNT ≈ 6–8.

7. **Risperidone** – Effective. Response rates ~50–60%. NNT ≈ 5–7.

8. **Cariprazine** – Newer agent. Response rates ~45–55%. NNT ≈ 6–8.

9. **Combination therapy** (e.g., lithium/divalproex + antipsychotic) – More effective than monotherapy in some trials. Response rates ~60–70%. NNT ≈ 4–5.

**Second-line treatments:** Carbamazepine, haloperidol, chlorpromazine, olanzapine (olanzapine is first-line in some guidelines but ranked second-line here due to metabolic side effects).

### Acute depression (bipolar I disorder)

**First-line treatments (hierarchical order):**

1. **Quetiapine** – Best evidence. Two large RCTs (BOLDER I and II) showed quetiapine 300 mg/day and 600 mg/day both superior to placebo. Response rates ~55–65% vs. ~35–40% for placebo. NNT ≈ 5–6. Effect size (Cohen's d) ≈ 0.4–0.5.

2. **Lurasidone + lithium or divalproex** – One large RCT (PREVAIL) showed lurasidone 20–120 mg/day added to mood stabiliser superior to placebo. Response rates ~57% vs. ~42% for placebo. NNT ≈ 7.

3. **Lithium monotherapy** – Older data, but effective. Response rates ~40–50%. NNT ≈ 8–10.

4. **Lamotrigine** – Effective for acute depression, but slow titration (6–8 weeks to reach therapeutic dose) limits its utility. Response rates ~45–55%. NNT ≈ 7–9.

5. **Lurasidone monotherapy** – One RCT showed efficacy. Response rates ~53% vs. ~30% for placebo. NNT ≈ 5.

6. **Adjunctive lamotrigine** – Added to lithium or divalproex. Response rates ~50–60%. NNT ≈ 6–8.

**Second-line treatments:** Olanzapine + fluoxetine (Symbyax), valproate, carbamazepine, ECT (electroconvulsive therapy).

**Important note:** Antidepressant monotherapy (e.g., SSRIs, SNRIs) is **not recommended** for bipolar I depression due to risk of switching to mania. The switch rate with antidepressant monotherapy is ~10–20% vs. ~3–5% with placebo.

### Maintenance therapy (bipolar I disorder)

**First-line treatments (hierarchical order):**

1. **Lithium** – Gold standard. Reduces risk of manic relapse by ~40–50% and depressive relapse by ~30–40% compared to placebo. NNT to prevent any relapse ≈ 4–5.

2. **Quetiapine** – Effective for preventing both manic and depressive relapses. Risk reduction ~40–50% for mania, ~30–40% for depression. NNT ≈ 4–6.

3. **Divalproex** – Effective for preventing manic relapse (risk reduction ~30–40%), less robust for depressive relapse. NNT ≈ 6–8.

4. **Lamotrigine** – Uniquely effective for preventing depressive relapse (risk reduction ~30–40%), less effective for mania. NNT ≈ 6–8.

5. **Asenapine** – Effective for preventing manic relapse. NNT ≈ 7–9.

6. **Aripiprazole** – Effective for preventing manic relapse. NNT ≈ 7–9.

7. **Combination therapy** (e.g., lithium + quetiapine, lithium + lamotrigine) – More effective than monotherapy for patients with frequent relapses. NNT ≈ 3–5.

**Key principle:** Medications that worked for acute episodes should generally be continued for maintenance. Exception: antidepressants should be tapered after remission due to risk of cycle acceleration and switch.

### Bipolar II disorder

**Acute depression**: Quetiapine is first-line (one RCT showed response rates ~60% vs. ~40% for placebo). Lamotrigine and lithium are second-line.

**Maintenance**: Lithium and lamotrigine are first-line. Quetiapine is second-line.

**Note**: The evidence base is much smaller. Most recommendations are based on extrapolation from bipolar I data or small trials.

### Safety and tolerability

**Lithium**: Narrow therapeutic window. Requires regular monitoring of serum levels (target 0.6–1.2 mmol/L), thyroid function, and renal function. Common side effects: tremor, polydipsia, polyuria, weight gain, cognitive dulling.

**Quetiapine**: Sedation (dose-dependent), weight gain (mean 2–4 kg in 8 weeks), metabolic syndrome risk (increased glucose, lipids). Higher doses (600 mg) cause more sedation.

**Divalproex**: Weight gain (mean 2–5 kg), tremor, hair loss, polycystic ovary syndrome risk in women. Requires liver function monitoring.

**Lamotrigine**: Well-tolerated. Risk of Stevens-Johnson syndrome (rare, ~0.1%) if titrated too quickly. No weight gain, no metabolic issues.

**Atypical antipsychotics (asenapine, aripiprazole, paliperidone, risperidone, cariprazine)**: Variable weight gain (aripiprazole and cariprazine are weight-neutral; risperidone and paliperidone cause moderate weight gain; asenapine causes mild weight gain). All carry risk of extrapyramidal symptoms (tremor, rigidity) and metabolic syndrome.

The effect sizes in bipolar disorder treatment are moderate compared to many other medical interventions:

**For acute mania**: First-line drugs reduce YMRS scores by ~10–15 points more than placebo over 3–4 weeks. This is roughly equivalent to going from severe mania (YMRS 30) to mild mania (YMRS 15–20). About 50–60% of patients achieve a ≥50% reduction in symptoms.

**For acute depression**: Quetiapine and lurasidone reduce MADRS scores by ~6–10 points more than placebo over 6–8 weeks. This is roughly equivalent to going from moderate depression (MADRS 25) to mild depression (MADRS 15–18). About 55–65% of patients respond vs. 35–40% on placebo.

**For maintenance**: Lithium reduces the absolute risk of any relapse by ~20–30% over 12–24 months. This means if 50% of placebo patients relapse, ~20–30% of lithium patients relapse. The number needed to treat (NNT) to prevent one relapse is 4–5, which is considered good for psychiatric treatment.

**For context**: The effect size of lithium for preventing manic relapse (Cohen's d ≈ 0.6–