To Name :
To Email :
From Name :
From Email :
Comments :

Evidence-Based Reviews

When bipolar treatment fails: What’s your next step?

Be a troubleshooter: systematically eliminate whatever is perpetuating manic, depressive, or cycling symptoms.

Vol. 7, No. 1 / January 2008

All phases of bipolar disorder can be difficult to treat, and patients remain symptomatic on average about half the time.1 Not all bipolar patients who experience continued illness and disability are treatment-resistant (Box 1), but when symptoms persist you may ask yourself: Was treatment suboptimal or simply ineffective?

Patients with severe symptoms may be satisfied with a substantial decrease in symptoms, but any residual symptoms cause ongoing distress and lower the threshold for recurrences.2 Finding the right combination of therapies for your patient is key to achieving an enduring response.

Future studies may tell us which treatments to combine and in what sequence for complex bipolar disorder, but—since most published studies exclude complex and comorbid cases—for now we must rely on limited controlled data and clinical experience. Using those resources, we offer comprehensive, practical recommendations for trouble-shooting (Box 2)3-6 and getting better results when bipolar disorder does not respond to standard treatment.

Box 1

What is ‘treatment resistance’ in bipolar disorder?

Some studies define treatment resistance as failure to respond to lithium, and in other settings it is viewed as failure to respond to ≥2 treatment courses. Because euthymia and normal functioning are important for long-term prognosis, we define treatment-resistance as failure to achieve both symptomatic and functional remission following an adequate course of therapy.

Effective strategies for treating bipolar disorder depend on:

  • illness phase (later episodes are more difficult to treat than earlier ones)
  • symptom complexity (mixed symptoms probably reflect more complex pathophysiology and are more likely to require combination therapies)
  • predominant presentations (mania, depression, rapid and ultradian cycling)
  • whether symptoms are acute or chronic.

Unfortunately, the findings of and strategies used in clinical trials of refractory bipolar disorder are difficult to extrapolate to everyday practice. Most studies exclude patients with a history of treatment resistance, severe symptoms, and important comorbidities such as substance abuse. In addition, the usual primary endpoint is response (≥50% reduction of symptoms) rather than remission (minimal symptoms and no longer meeting criteria for the disorder). Very few studies address functional remission, which is necessary to reduce the risk of symptomatic recurrence.

In clinical practice, when initial treatment for bipolar disorder fails to produce remission, systematically addressing 5 questions (Box 2) can help direct your next step.


When a patient with mania does not respond as expected, the next step depends on which antimanic agent you prescribed:

Lithium can take a month to become fully effective for mania, which is why a benzodiazepine or antipsychotic is often added acutely to reduce agitation. Do not mistake neurotoxic interactions between lithium and antipsychotics for increased mania.

Although data vary on lithium’s optimal serum level, adjust to approximately 0.8 to 1 mEq/L, if tolerated, when lower levels are not effective. Children and young adolescents may need higher serum levels (such as 1.5 mEq/L) because the difference between serum and brain lithium levels is greater in younger patients than in adults.

Consider the dosing schedule. Because lithium’s elimination half-life with repeated dosing is 24 hours, most adults can take any formulation once daily—which improves adherence and reduces adverse effects. Children eliminate lithium more rapidly and need more frequent dosing.

Valproate. Empiric trials in bipolar disorder or epilepsy do not support the frequently reported “therapeutic range” of 50 to 125 μg/mL. Pooled data from three 21-day, double-blind studies of valproate in mania show a linear relationship between serum level and clinical response, with the most beneficial response at >94 μg/mL.7 Better results—but more side effects—are seen with levels >100 μg/mL.

High loading doses result in more rapid control of agitation, probably as a result of sedation. In our experience, however, rapidly sedating patients may interfere with long-term adherence.

Carbamazepine, other anticonvulsants. Because they less sedating, carbamazepine and other anticonvulsants might not appear to be rapidly effective for bipolar mania. If you wait up to a month, however, any antimanic effect will be obvious.

Antipsychotics are rapidly effective for mania. Higher doses work faster but produce more side effects. After an acute response, some patients can be maintained on a second-generation antipsychotic (SGA), but others do better on a standard mood stabilizer such as lithium or valproate.

Calcium channel blockers. Verapamil has been effective mostly for lithium-responsive mania in 27 of 30 studies. Nimodipine has been useful for more complex bipolar syndromes in a few studies using patients as their own controls.

To be effective for bipolar disorder, however, calcium channel blockers require frequent, high dosing (such as verapamil, 120 mg 4 times daily, or nimodipine, 60 to 120 mg 6 times daily), which makes adherence difficult.

Box 2

5 questions to consider when bipolar symptoms persist

1 Is the patient taking anything that is making symptoms worse?

Antidepressants can induce mania, hypomania, and cycle acceleration in bipolar disorder, even when mood stabilizers are co-prescribed.3 Stimulants also may destabilize bipolar mood disorders; consider this possibility when patients taking stimulants for apparent attention-deficit/hyperactivity disorder at first appear to improve and then deteriorate.

Alcohol and cocaine can induce mania and depression. Cocaine is a potent kindling stimulus that could contribute to enduring mood instability.

2 Is the patient taking the medication?

Treatment adherence by bipolar patients may be as low as 35%.4 Ask outpatients what kinds of problems they have encountered taking medications, not whether they have such problems. Talk with the patient about adherence after each dosage increase, and be readily available. Prescribe extended-release pills for patients who have trouble keeping track of medications.

3 Is treatment adequate?

Adjust mood-stabilizer dosing until the patient responds or cannot tolerate the medication; complex cases often require combination treatment. Give the medication sufficient time to work; most mood stabilizers take ≥1 month to become fully effective.

4 Is another condition interfering with treatment?

Up to 70% of patients with refractory mood disorders have subclinical hypothyroidism. Look for:

  • elevated thyroid stimulating hormone (TSH) with or without decreased thyroxine (T4)
  • elevated TSH response to thyrotrop-inreleasing hormone (TRH).5

Also consider hypercalcemia from chronic lithium therapy,6 anemia, sleep apnea, posttraumatic stress disorder, substance use disorders, and personality disorders.

5 Am I ignoring psychotherapy?

Address psychosocial issues that influence the course of illness. Attend to patients’ important relationships, loss, negative thinking, and biological and social rhythms.

Augment or switch? If mania does not respond to an adequate dose of an antimanic drug given for a sufficient time, the next question is whether to augment or switch treatments. No studies have compared augmenting vs switching in any bipolar disorder phase, but it seems reasonable to:

  • consider augmentation first when a patient has had a partial response to a given medication
  • switch when a patient cannot tolerate or shows no response to a therapeutic dose of a given medication.

Combinations. Benzodiazepines such as clonazepam, 2 to 6 mg/d, or lorazepam, 4 to 8 mg/d, are often used to control agitation and insomnia in mania, usually as adjuncts to mood stabilizers (although improved sleep by itself can ameliorate acute mania in some cases). Adding an SGA may help when mania responds partially to a mood stabilizer.8

Combinations of lithium and carbamazepine or valproate can be more effective than either drug alone, but therapeutic doses of each usually are needed. Carbamazepine has been used successfully to augment a partial response to nimodipine.9 In a small open-label trial, adding oxcarbazepine to lithium, valproate or antidepressants improved response in some patients with mild refractory mania.10

Switching among anticonvulsants can be useful because their actions and side effects differ. Clozapine in a wide range of doses can be very effective for refractory mania,11 but its use is difficult to monitor in highly agitated manic patients.

Other options. Electroconvulsive therapy (ECT) is the most effective treatment for mania, producing higher response rates than any antimanic drug.12 In a study of repetitive transcranial magnetic stimulation (rTMS), 8 of 9 patients with mania refractory to mood stabilizers had a sustained response after 1 month of right-sided rTMS treatment.13 Conversely, left-sided rTMS can aggravate mania.

Bipolar depression

Continuing controversy about the best way to treat bipolar depression makes it difficult to know if treatment has been suboptimal or a patient is treatment-resistant.

Antidepressants. No antidepressant is approved (or recommended) as monotherapy for bipolar depression, and most experts recommend against prescribing antidepressants without concomitant mood stabilizers. Even so:

  • Clinicians prescribing monotherapy for bipolar disorder choose antidepressants twice as often as mood stabilizers.
  • Antidepressants are prescribed more frequently in combination with mood stabilizers than as monotherapy, although empiric trials have shown most antidepressants are not effective for bipolar depression.14

A recent report from the Systematic Treatment Enhancement Program for Bipolar Disorder (STEP-BD) study15 found that adding bupropion or paroxetine to mood stabilizers was no more effective than adding placebo. Rates of mania induction also were no greater with antidepressants than with placebo, but the study lasted only 8 weeks. One interpretation of this finding is that when antidepressants do not induce mania and cycling, they also do not improve bipolar depression.

In many cases, an antidepressant seems to help at first and then induces a recurrence of depression, often mixed with dysphoric hypomanic symptoms. The recurrent episode improves when the clinician increases the antidepressant dose or changes to another antidepressant, only to be followed by another recurrence that may be interpreted as an incomplete antidepressant response.

Antipsychotics. Quetiapine16 and a combination of olanzapine and fluoxetine17 are approved for treating bipolar depression. The studies supporting this indication lasted only 8 weeks, however, and excluded patients with the kinds of complicated and comorbid mood disorders commonly seen in clinical practice.

Many patients dropped out before the studies were completed, and “screen fails” (patients with the diagnosis who were not enrolled in the study) were not reported. In addition, “remitted” patients remained symptomatic.

Therefore, FDA approval of this indication does not guarantee these medications’ long-term efficacy or safety for bipolar depression or that they are useful in patients with complex forms of bipolar depression.

Recommended approach. Treatment resistance of bipolar depression to multiple mood stabilizers—with or without an antidepressant—or to an antipsychotic may manifest as lack of response, partial response, or initial good response followed by relapse or recurrence. Sometimes depression improves but irritability or mood lability worsen.

No reliable controlled studies have addressed complex refractory bipolar depression, but clinical experience suggests 1 approach for all of these responses:

Reconsider possible hypothyroidism. A low-normal T4—especially if decreased over time—and a mid-range or high-normal TSH—especially if increased—may indicate that subclinical hypothyroidism is inhibiting a response to mood stabilizers and antidepressants.18

Stop the antidepressant. If your patient is taking an antidepressant, it may be ineffective, creating mixed dysphoric hypomania, and/or driving another recurrence of depression. This is especially likely if the patient shows an initial prompt antidepressant response, but depression returns with irritability, insomnia, restlessness, or other subtle symptoms of dysphoric hypomania.

Withdraw the antidepressant gradually; for example, you might reduce the dose by 10% every few weeks so that the agent is discontinued across several months. Discontinuing an antidepressant too rapidly—even if it does not seem to be having any effect—can cause rebound depression that creates the mistaken impression that the antidepressant is needed.

Treat mood lability and mixed hypomania first. Antidepressant therapy may be more likely to destabilize mood if hypomania and mood cycling are present when you start the antidepressant.19 Older studies suggest that lithium and carbamazepine can improve bipolar depression, and a few small studies suggest nimodipine may be useful when depression is prominent. In our experience, valproate is not particularly helpful for bipolar depression, although it may reduce the risk of depressive recurrence.

Did you miss this content?
Botulinum toxin for depression? An idea that’s raising some eyebrows