Focus Points

• Many of the recommendations in the American Psychiatric Association’s “Practice Guideline for the Treatment of Patients with Bipolar Disorder” are supported by a limited evidence base.

• Guideline recommendations may change rapidly as new evidence is generated.

• Some treatments, including antipsychotics, benzodiazepines, and arguably newer antidepressants, could be considered mood stabilizers.

• The frequent recommendation in the guideline of adding medications to nonresponsive patients does not consider the possibility that switching medications is also reasonable.

• Olanzapine and lamotrigine have new evidence of efficacy beyond that which is described in the guideline.

• Directional response is important in choosing medications for bipolar patients.


The American Psychiatric Association’s “Practice Guideline for the Treatment of Patients with Bipolar Disorder” represents a major synthesis of the available research and expert clinical opinion on bipolar disorder. The guideline extends beyond the clinical research evidence base and bridges the highly-controlled conditions of clinical trials and the broader range of clinical situations encountered in practice. However, any practice guideline is inevitably outdated by the time it is published, due to the time required in the process of its creation and publication. This article summarizes the recommendations in the APA bipolar practice guideline and reviews new relevant research available since its publication. The concept of a “mood stabilizer,” which is not incorporated in the guideline, is reviewed. In addition, the numerous clinical assumptions underlying the recommendations in the guideline are described, and the strength of the research evidence supporting them is reviewed. These include the assumptions that lithium and anticonvulsants are preferred treatments while antipsychotics and benzodiazepines are not, that nonresponding patients should always have additional medication added rather than switching to a new medication, that all antidepressants have a significant liability for causing switching into mania, and that all bipolar patients should receive long-term maintenance. New research with lamotrigine, olanzapine, and other atypical antipsychotics in bipolar disorder is reviewed, as well as the implications of this research on the recommendations in the guideline. The concept of directional efficacy of mood stabilizers, only briefly mentioned in the guideline, is discussed and expanded. Finally, the importance of understanding the research basis of the guideline before applying its recommendations to clinical practice is described.


The current American Psychiatric Association (APA) “Practice Guideline for the Treatment of Patients with Bipolar Disorder”1 was published in 2002 and is a revision of an earlier version published in 1994. The latest revision represents a summary of a much larger evidence base than the previous version and is a distillation of both available research and expert opinion.

Unfortunately, the guideline has several problems: first, such an endeavor, by nature, is outdated by the time it is published. Second, while the guideline contains descriptors that indicate the level of author confidence in the evidence supporting its recommendations, many of the recommendations are based essentially on opinion and are not supported by well-designed trials. Recommendations primarily based on expert opinion or clinical consensus with either no evidence basis or an uncontrolled case series may be erroneous. Such untested or unproven treatments that are popular among clinicians may find their way into the practice guideline simply on the basis of their popularity, which can lead to the widespread perpetuation of clinical myth and folklore. For example, topiramate and gabapentin are common treatments for acute mania that have failed to demonstrate efficacy in placebo-controlled trials. Their popularity developed from numerous case series suggesting efficacy in mania. Despite the negative trial data, clinicians have already become familiar with using these medications; currently, ancillary uses include prescribing topiramate for drug-induced weight gain and gabapentin for insomnia or anxiety. Unfortunately, neither the practice guideline nor Food and Drug Administration product labels are designed to discuss such ancillary uses. Should such a treatment or practice be included in the guideline simply because it is popular?

A third problem with the practice guideline is that it is indirectly impacted by commercial influences. While any obvious influence can be reduced by mechanisms such as those utilized in the development of the APA guideline, such as independent review and full disclosure of potential conflicts, both subtle and conspicuous influences from the industry still make their way into the guideline. For example, clinical trial data from pharmaceutically-funded projects is often considered proprietary; theoretically, the industry sponsor could conduct multiple controlled trials and suppress any negative results. Recent proposals to create an inclusive registry of all clinical trials may reduce this problem. Even the recommendations arising from scientifically high-caliber, placebo-controlled, double-blind studies are based overwhelmingly on industry-sponsored trials, which are subject to bias introduced through both study design and selective reporting of results. Thus, the myriad of choices made in designing any clinical trial are often driven by potential drug sales rather than science. Furthermore, the use of leading academics as authors seldom reflects the independent control of the design, the statistical analysis, or the reported results of the study, as is perhaps implied.

Clearly, limiting recommendations exclusively to medications derived from high-quality, placebo-controlled clinical trials is impractical. Unfortunately, solely using clinical practice as a guide can also be risky, since the majority of clinical practice involves patients who do not meet the inclusion criteria of highly-controlled clinical trials, and the majority of clinical decisions are beyond the tightly-defined conditions of clinical trials. The practice guideline, though, despite its many problems, represents an important effort to bridge the gap between clinical trials and clinical practice.

This review summarizes the APA guideline’s strengths and weaknesses (Table 1), and introduces new research completed since its publication. In addition, it examines many of the guideline’s recommendations that are not evidence based. It discusses alternative rationales that warrant consideration and presents several key questions, which have not as yet been studied but are important for the development of future standards of treatment for bipolar disorder.

What is a Mood Stabilizer?

Despite its widespread use, the term “mood stabilizer” was intentionally omitted from the APA practice guideline due to a lack of consensus on a definition. Simplistically speaking, the term is most often used to describe a first-line treatment for bipolar disorder that would not destabilize the disorder. Beyond this, though, there is disagreement about whether both acute antimanic and antidepressant (for bipolar depression) efficacy are required or whether both acute and maintenance efficacy are required. While the term is technically based on use of treatments in bipolar disorder, medications identified as mood stabilizers are also often used for lability, impulsivity, and aggression in bipolar patients and in patients with other psychiatric disorders. Clinicians can benefit from a clarification of this term.

If the term mood stabilizer is a proxy for a “first-line treatment for any phase of bipolar disorder,” as is often the intent with clinicians who use it, then both uni-directional antimanic treatments, as well as uni-directional bipolar depression treatments, could be considered mood stabilizers regardless of whether bi-directional or maintenance efficacy has been convincingly demonstrated. This is, of course, assuming that the treatment does not worsen the course of the illness by causing a switch to the other pole or by exacerbating rapid cycling.

Lithium, divalproex, valproate, and carbamazepine have been shown to have efficacy in mania without exacerbating depression and are generally considered mood stabilizers (Figure 1).2 Likewise, lamotrigine has been shown to have efficacy in bipolar depression without exacerbating mania.3 Paroxetine has shown some evidence of efficacy in bipolar depression4 with little liability of inducing switching. Buproprion may also have efficacy in bipolar depression with a low liability for switching,5 although its efficacy has never been tested in a placebo-controlled trial.

Switching is the most relevant problem for many treatments of bipolar depression. One review that attempted to examine switching in more detail compiled data from placebo-controlled trials of sertraline, paroxetine, tricyclic antidepressants (TCAs), and placebo in bipolar depression. This review included a post-hoc analysis of 242 cases of bipolar depression treated under controlled conditions with selective serotonin reuptake inhibitors (SSRIs) and found an associated switch rate into mania of 3.7%, which was no different from the switch rate on placebo. The 125 bipolar depressed patients treated with TCAs had a switch rate of 11.2%, which was clearly greater than the switch rate on placebo.6 This report supports the well-documented and clearly demonstrated risk of switching with TCAs (primarily imipramine). Monoamine oxidase inhibitors were also found in controlled trials to have significant risks of switching. Thus, the logical conclusion is that the risk of switching caused by SSRIs is comparatively low.

It is important to separate evidence from controlled trials and anecdotal reports or case series, especially in the case of induction of mania, which occurs naturally in bipolar depressed patients anyway. In fact, to address this issue, it is critical to estimate the expected switch rate on placebo. While there is no single large set of data that allows for a direct comparison of switch rates on antidepressants versus placebo, switch rates on placebo from smaller sets of data are as high as 5%.3

Whether drugs traditionally considered tranquilizers (specifically typical antipsychotics, benzodiazepines, and atypical antipsychotics) should also be considered safe first-line treatments for bipolar disorder is a controversial issue. There is extensive evidence of the efficacy of typical antipsychotics in acute mania,7 and indeed these drugs are used for acute and maintenance treatment of bipolar disorder. While typical antipsychotics are often described as causing depression, the evidence for this is unconvincing and derived largely from two early reports on flupenthixol decanoate. Both reports had design problems and did not statistically test for this side effect.8,9 Furthermore, there are case reports of bipolar patients who became more depressed when a typical antipsychotic was discontinued. The benzodiazepines clonazepam and lorazepam have also been shown to have efficacy for acute mania10-15 with no suggestion of any aggravation of depressive symptoms. Thus, typical antipsychotics, clonazepam, and lorazepam could arguably be considered mood stabilizers.

Typically, though, these medications are not regarded as mood stabilizers because of tolerability issues. The risk of tardive dyskinesia from typical antipsychotic use and substance dependence with benzodiazepines may relegate these drugs to second-line choices. It is notable that reports of an increased risk of tardive dyskinesia in bipolar patients on typical antipsychotics compared to schizophrenic patients were based on methodologically limited studies,16 and the question of how likely dependence on benzodiazepines occurs in bipolar patients under medical supervision, even for long periods of time, has never been studied. Furthermore, it is clear that many bipolar patients do use benzodiazepines chronically with apparently good efficacy and without any dependence problems.

More interesting is the question of whether atypical antipsychotics are mood stabilizers. Olanzapine has demonstrated efficacy in acute mania (Figure 2),17 bipolar depression, and bipolar maintenance.17-20 Quetiapine, risperidone,21 ziprasidone, and aripiprazole (Figure 3)22 also have been shown to be effective in treating acute mania, either as monotherapy or in combination with lithium or divalproex (Figures 4 and 5).21,23-26 The liability of such atypical antipsychotics for inducing depression or tardive dyskinesia seems to be low. These drugs are commonly used for bipolar disorder but only olanzapine is listed in the guideline as a first-line treatment for mania. Generally, the atypical antipsychotics as a class are not recommended as first-line treatments because of limited available research regarding long-term use, although such data is emerging rapidly. They are recommended for patients with persistent psychosis or for those who relapse in spite of other treatments.

The industry-supported research on atypical antipsychotics for bipolar disorder has to some degree avoided this question by designing studies that combine atypical antipsychotics with an accepted mood stabilizer, specifically lithium or divalproex. The implication is that the other medication (ie, lithium or divalproex) is the primary treatment and the atypical antipsychotic is the added tranquilizer. Unfortunately, most of these trials have underdosed lithium or divalproex, which further leads to difficulty in interpreting the results. Trials of atypical antipsychotics in mania, bipolar depression, and maintenance treatment should consider monotherapy designs to avoid the need to interpret the effects of the other drug.

The question remains: should atypical antipsychotics be considered first-line treatments for bipolar disorder? Their potential to be useful in this role is clear, but more research is needed. On the other hand, anticonvulsants such as divalproex and lamotrigine have, in some ways, been considered mood stabilizers despite only slight or no evidence of bi-directional effects.

Other mood-stabilizing treatments that are considered by the practice guideline include electroconvulsive therapy (ECT) and psychotherapy. ECT may be effective for both mania and bipolar depression, and psychotherapies are generally accepted as helpful in bipolar depression, although controlled data for this is lacking. A comprehensive index of possible mood stabilizers (ie, treatments that do not exacerbate bipolar disorder) is listed in Table 2 and Table 3.1

Combination Therapy Versus Sequential Monotherapy

Important to consider when using mood stabilizers is the concept of adding rather than switching treatments, which is one of the key underpinnings of the guideline. This practice has been studied only minimally due to the many logistical problems associated with attempting to research this topic. At issue is whether multiple medications should be used early or if they should be utilized only after sequential monotherapy has failed.

The answer to this question relates to some degree to culture: the tradition of psychopharmacology in the United States has for decades emphasized monotherapy. Monotherapy was largely driven by the FDA approval process, which has principally required placebo-controlled monotherapy studies due to their clear interpretability. Clinicians in many other countries, however, routinely prescribe multiple medications and can be highly respected for their skills at safely combining medications. US practices have shifted toward polypharmacy, as reflected in the bipolar practice guideline. Combination therapy or polypharmacy can hold theoretical benefits of enhanced efficacy and reduced side effects with lower doses of multiple medications, but is fraught with the risks of additive side effects and, most importantly, subtherapeutic dosing of medications.

There appears to be a growing clinical impression reflected in the practice guideline that response rates to multiple medications are greater than response rates to a single medication. This is based on case reports, uncontrolled case series, and very few actual clinical trials. This impression has led to widespread recommendations for combination therapy and polypharmacy, with the implication that combinations work together synergistically. However, from a scientific viewpoint, such conclusions may be premature since two medications used at adequate doses in sequence could theoretically produce comparable response rates. Consider a scenario in which two treatments, Drug A and Drug B, have both been shown to be effective as monotherapy for some specific disorder. Each achieved a response rate of 70% based on some predetermined definition of response, such as a reduction in symptoms. This means that 30% of subjects given each drug did not achieve the predefined response. However, the 30% of exposed subjects who are nonresponsive to each of the two medications differs in each study. In fact, while Drug A and Drug B could each produce efficacy in 70% of subjects, these are not the same subjects. In other words, while there may have still been a 70% response rate had the subjects in the study of Drug A instead been in the study of Drug B, it would have been different subjects who responded.

Another study starts similar subjects on a combination of Drugs A and B and achieves a response rate of >70%. This is often interpreted as either additive or synergistic efficacy of combination treatment. However, even without any synergistic or additive efficacy, a higher rate of response would be expected from two efficacious treatments because some of the nonresponders to Drug A are responders to Drug B and vice versa. Had the study started with Drug A and then given Drug B to the Drug A nonresponders, a total response rate might have equaled that of combination treatment. Thus, higher response rates in combination treatment could be explained simply by individual differences in drug responsiveness, rather than by additive or synergistic efficacy.

A small number of reports on synergistic effects and differences in individual drug responsiveness have led to widespread recommendations for combination treatments. The effort to achieve higher efficacy rates is clearly desirable, and the side effects may not necessarily be worse with more medications, especially if they are all underdosed. On the other hand, if clinicians plan to combine medications either at the start of treatment or shortly thereafter, there is a strong likelihood that the dosing of each medication will be lower, which leads to an inadequate trial of the first treatment in terms of both dose and duration. Such an approach may also minimize clinical efficacy of each medication and is clearly a divergence from the largely FDA-driven American practice of using a single drug at a high enough dose to better determine either drug responsivity or drug refractoriness.

Combining or switching medications is difficult to address in clinical trials, but as a routine clinical practice is legitimate material for a consensus-based guideline. However, the add versus switch question is complex and may not only reflect scientific issues, but also philosophical and cultural ones beyond the scope of such a guide. Essentially, adding additional medications for nonresponders makes sense, but switching to another medication is also a reasonable option that has not been considered. While combination therapy is common in other specialties such as oncology and infectious disease, such treatments are recommended based on clinical trial evidence. Combination treatments for psychiatric disorders work, but their superiority over sequential use of monotherapies has not been clearly demonstrated by clinical trials. In the absence of better-designed studies on the efficacy of combination therapies that address the questions of proper dosing of each medication and adding versus switching, it is likely that use of combination therapy for bipolar disorder will become arbitrary and without a credible evidence base.

Treatment of Manic or Mixed Episode

The bipolar practice guideline recommends lithium, divalproex, and olanzapine as first-line treatments for mild to moderate mania or mixed episode of bipolar disorder. For severe mania or mixed episode, a combination of either lithium or divalproex with an antipsychotic is recommended. For mixed episode, divalproex is preferred over lithium (Figure 6).2 Alternatives to lithium and divalproex include carbamazepine and oxcarbazepine. In addition, short-term use of benzodiazepines is recommended. Tapering of antidepressants and psychosocial treatments is also recommended.

There are many assumptions regarding treatment of manic or mixed episode that underlie these recommendations (Table 4).1 While these assumptions may reflect standard clinical care, the rationale behind them is based on limited research. The first assumption is that lithium and divalproex are appropriate for both acute and maintenance treatment, while antipsychotics and benzodiazepines are not. Supporting this is the fact that substantially more clinical and research trials with lithium and divalproex in bipolar maintenance have been performed. Although the guideline avoids calling such medications mood stabilizers, the authors clearly consider lithium and divalproex mood stabilizers, but do not consider other antipsychotics or benzodiazepines as such (Table 5).

This conceptual dichotomy between anticonvulsants and antipsychotics arises out of the long history of lithium as the primary treatment for mania, and the assumption that anticonvulsants should be used like lithium while antipsychotics should not. The dichotomy developed in the era of typical antipsychotics and was largely driven by concerns about acute extrapyramidal side effects, tardive dyskinesia, and the unproven perception that typical antipsychotics could make patients depressed. This dichotomy seemed sensible since the development, FDA labeling, and early use of anticonvulsants and antipsychotics had previously been for completely different disorders, specifically epilepsy and schizophrenia. Today, the dichotomy makes less sense, as atypical antipsychotics clearly have a reduced risk of both acute extrapyramidal side effects and tardive dyskinesia. Also, as more clinical trials of atypical antipsychotics in bipolar disorder are made available, the potential use of atypicals for all phases of bipolar disorder will become more apparent. The point is that a dichotomy separating anticonvulsants and atypical antipsychotics is not fundamentally driven by any research that has been done in bipolar disorder. Recent clinical trials, with a few exceptions,18,19,26 have done little to reduce the perpetuation of this now-outdated conceptual dichotomy.

Furthermore, the question of whether lithium and divalproex are better tolerated than atypical antipsychotics in long-term treatment has never actually been addressed by a clinical trial. Since many medications are effective for mania, suitability for long-term treatment is a key issue in drug selection. Antipsychotics may have long-term side effects, but the long-term effects of lithium and divalproex are also considerable. Lithium is associated with tremor, weight gain, acne, cognitive impairment, hypothyroidism, diabetes insipidus, nephrotoxicity, gastrointestinal problems, alopecia, and edema. Divalproex is associated with tremor, weight gain, alopecia, hepatotoxicity, leukopenia, thrombocytopenia, pancreatitis, and possibly polycystic ovarian syndrome. Thus, the preference of lithium and divalproex over atypical antipsychotics for maintenance treatment is partly based on an assumption of better tolerability that has never been tested and is questionable at best.

A second assumption underlying treatment of a manic or mixed episode maintains that if a patient is nonresponsive to a single treatment, other treatments should be added rather than substituted. Adding versus switching has not been well addressed by clinical trials, and the few studies that have addressed it primarily studied a combination of lithium and carbamazepine or divalproex and had small numbers or methodological limitations.27-32 One small study found a lithium-divalproex combination superior to lithium alone for maintenance treatment,33 but there were only five subjects on the combination. Systematic attempts to address adding or substituting other drugs have failed to consistently show that combination treatment produces superior efficacy compared to a single efficacious treatment, especially if the treatment is an adequately dosed antipsychotic for mania.34-37 One large well-designed multicenter acute mania study38 found that the addition of valproate to a flexible-dose typical antipsychotic allowed lower antipsychotic dosing and superior efficacy, but the antipsychotic in the study was dosed openly. The finding of superior efficacy in the combined valproate plus antipsychotic group is one of the rare examples of a demonstrated advantage to combination over antipsychotic alone.

Adding another medication for an unresponsive or partially-responsive patient may be a clinically reasonable approach, and the practice guideline has clearly endorsed this approach as a routine recommendation. However, the research evidence supporting the idea that combination treatments produce greater efficacy than optimal dosing of both drugs separately and sequentially is quite limited. Whether automatically prescribing combination treatments makes sense for completely unresponsive patients is questionable, as most treatments cause some adverse events.

The recommendation to combine either lithium or divalproex with an antipsychotic for severe mania also does not consider the option of antipsychotic monotherapy. This is one of several examples in the practice guideline that favors combination therapy or polypharmacy over monotherapy with a different medication. The clinical trials of antipsychotics in mania have included at least moderately ill subjects, which suggests that antipsychotic monotherapy may be efficacious for these patients. Furthermore, the addition of lithium has never been demonstrated to produce additional efficacy for acute manic patients who receive an efficacious dose of an antipsychotic.34,35,37

The possibility that using lithium or divalproex together with an antipsychotic might reduce antipsychotic dosing has been studied at least twice under controlled conditions. One haloperidol study34 found that combining lithium with a low dose of haloperidol for acute mania produced efficacy comparable to a higher dose of haloperidol monotherapy. Muller-Oerlinghausen and colleagues38 also found in a multicenter placebo-controlled acute mania study that when adding valproate versus placebo to typical antipsychotics dosed openly, the antipsychotic-valproate group was treated with lower doses of antipsychotic and had a greater clinical response than the antipsychotic-placebo group. Thus, considering a reduced-dose antipsychotic as the primary treatment and lithium or divalproex as the added treatment may be an alternative approach to combination therapy. For manic patients not fully responsive to atypical antipsychotic monotherapy, it might be reasonable to increase the dose of the antipsychotic first, and then consider adding lithium or divalproex (similar to what is often done with antidepressants and lithium in unipolar depression).

The practice guideline has also avoided considering the possibility of adding carbamazepine to either lithium or divalproex instead of an antipsychotic, in the case of severe or nonresponding patients. This combination is not considered despite the fact that double-blind data, albeit from a small number of subjects, suggests that it has efficacy in refractory patients.30

The practice guideline for mania recommends considering oxcarbazepine as an alternative to carbamazepine, under the assumption that oxcarbazepine has comparable efficacy. Oxcarbazepine has never been tested in a placebo-controlled clinical trial for acute mania. The available double-blind literature for oxcarbazepine and mania is limited to two small studies from the same principal investigator who found oxcarbazepine comparable in efficacy to either haloperidol or lithium.39 These two studies have limited utility in determining the antimanic efficacy of oxcarbazepine, due to the absence of a placebo control. Why the practice guideline has elevated oxcarbazepine to a prominence which has been reserved only for other treatments that have been tested in placebo-controlled trials is unclear (Table 6). The assumption that oxcarbazepine is as efficacious as carbamazepine for treatment of mania should be considered unproven at best.

The role of benzodiazepines has also been largely ignored. Despite the evidence that clonazepam and lorazepam are efficacious in mania, albeit at high doses (clonazepam 16 mg/day and lorazepam 30 mg/day), their use has been largely discouraged due to concerns about dependence. There are many cases, however, in which benzodiazepines have been used for long-term treatment with apparent prolonged efficacy. The true incidence rate and risk of dependence with chronic use when prescribed under the supervision of a psychiatrist is unknown and may have been overestimated in the interest of being cautious.

Treatment of Refractory Mania

The same concerns described above for mania or mixed episode also apply to the refractory mania recommendations. The guideline recommends lithium, divalproex, or an antipsychotic for refractory patients, followed by carbamazepine and oxcarbazepine, as well as clozapine or ECT (Table 7).1 Again, the question of the definition of an adequate trial of monotherapy for mania is avoided. It is notable that neither lithium nor divalproex was shown to have efficacy in treatment of a manic episode in <7 days,2 suggesting that >1 week may be necessary to define nonresponse. The guideline for refractory mania does not clearly define the condition itself. Should clinicians wait for 7–10 days before considering a patient refractory? If this decision is made sooner, has lithium or divalproex been given a reasonable opportunity to work? What blood level of lithium or valproate should be achieved before a patient is considered refractory? Should the same duration criteria be used for atypical antipsychotics that appear to have more rapid onset of efficacy?25

The recommendation to add an antipsychotic to lithium or divalproex for refractory mania suffers from the same problems as discussed for manic episode. Specifically, why should an atypical antipsychotic be added to lithium or divalproex instead of substituted? If lithium or divalproex alone is not working, why would they be continued if an antipsychotic is going to be used anyway? Haloperidol and lithium data suggest that if an adequate dose of antipsychotic is used, adding lithium produces no additional benefit.34

Treatment of Bipolar Depressive Episode

The early research basis of treatments for bipolar depression has been thoroughly reviewed by Zornberg and Pope.40 Many problems have limited the interpretability of these earlier studies, though, including the use of both unipolar and bipolar depressed subjects in early lithium trials; slow antidepressant response to either antidepressants or lithium in bipolar depression; and low overall rates of response. The guideline makes several assumptions about bipolar depression treatment (Table 8). The practice guideline recommends lithium or lamotrigine as first-line treatments, although it places a higher level of confidence in lithium, followed by a combination of an antidepressant with lithium or electroconvulsive therapy. In addition, interpersonal and cognitive-behavioral psychotherapies are suggested, although the limited evidence of their efficacy is acknowledged (Table 9).1

Since the guideline was published, a new pivotal study (Figure 7)20 has led to the first FDA approval for a treatment for acute bipolar depression, a combination of olanzapine and fluoxetine. This study, with over 700 randomized subjects, showed that both olanzapine and an olanzapine/fluoxetine combination have acute antidepressant efficacy in bipolar depression.20 Olanzapine alone was statistically superior to placebo in acute efficacy but had a relatively small clinical effect. More interestingly, a combination of olanzapine and fluoxetine was superior to both placebo and olanzapine monotherapy. The average dose of olanzapine was 9.7 mg/day. The average doses of olanzapine and fluoxetine in combination were 7.4 mg/day and 39.3 mg/day, respectively. Neither active treatment caused more switching into mania than placebo.20

This is the first study to show acute antidepressant efficacy for an antipsychotic in bipolar depression. It seems likely that studies of other atypical antipsychotics for acute bipolar depression will be conducted rapidly, hopefully creating more treatment options. Prospective studies in this area should consider whether to use atypical antipsychotic monotherapy; a combination of an atypical antipsychotic with an antidepressant; or an atypical antipsychotic combined with lithium, divalproex, or lamotrigine.

The clear and robust efficacy of the olanzapine-fluoxetine combination for acute bipolar depression may suggest a wider role for antidepressants in bipolar depression. However, as the first study showing acute efficacy of either an antipsychotic or an antipsychotic-antidepressant combination, the study raises many new questions. Unfortunately, this study did not have a cell on fluoxetine alone. Is antidepressany monotherapy a reasonable option? If not, and if antidepressants should always be combined with an antimanic drug, is lithium the logical choice, or are other mania treatments, such as olanzapine, worth considering? Furthermore, how long should the antidepressant be continued following remission? Perhaps the biggest question raised by this study is that of whether the terminologies of antipsychotic and antidepressant are really relevant to bipolar disorder.

Other research in this area includes an acute bipolar I depression trial of lamotrigine (Figure 8), which demonstrated clinically significant antidepressant efficacy and an overall mania switch rate equal to placebo.3 This study led to the inclusion of lamotrigine as a first-line treatment in the guideline. The onset of efficacy was seen by 3 weeks, although at that point the lamotrigine dose was only 50 mg/day; response increased further with subsequently increased doses up to 200 mg/day. This trial provides evidence of antidepressant efficacy in bipolar depression and suggests that lamotrigine could be considered as a first-line treatment. Furthermore, although lamotrigine must be titrated slowly to reduce the risk of serious rash, the statistically significant separation from placebo by 3 weeks of treatment, even at a dose of only 50 mg/day, indicates that the antidepressant response to lamotrigine may be no slower than that of antidepressants. Unfortunately, there is no evidence of efficacy in acute mania.

On the other hand, there are several arguments supporting the preference of lithium over lamotrigine for acute bipolar depression. These include the demonstrated efficacy of lithium in both acute mania and acute bipolar depression, efficacy in prevention of mania in both responder-enriched and nonresponder-enriched samples, strong antisuicide efficacy based on retrospective comparisons to suicide attempts prior to taking lithium, and extensive clinical experience. There is no available trial directly comparing lithium and lamotrigine in acute bipolar depression.

The recommendation against using antidepressant monotherapy for treatment of bipolar depressive episodes is so widespread and well-accepted that it warrants inclusion in the practice guideline. Even the Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition,41 allows the savvy clinician to easily diagnose substance-induced mood disorder (mania) secondary to an antidepressant. The question of whether the manic episode was caused by the antidepressant is left to the clinician’s judgement—specifically, no minimum dose of antidepressant or duration of exposure is required. Without a controlled trial, the individual clinician faced with an antidepressant-treated patient who becomes manic assumes that the manic episode is antidepressant induced because there is no way to determine if the patient would have become manic even without an antidepressant. The rate of switching of bipolar depressed patients receiving acute treatment with placebo has ranged from 2% to 5%.3,4,6

Research evidence supporting this recommendation is not as convincing as might be assumed. While TCAs clearly have a high rate of switching into mania, the data for newer antidepressants is far from convincing. Recommendations for using bupropion and paroxetine arise from one small non-placebo–controlled study of bupropion suggesting a lower switch rate than desipramine,5 and one larger study of paroxetine.4 Notably, in the paroxetine trial, there was no switching at all reported in paroxetine-exposed depressed patients, although it is possible that some subjects had other adverse events that were consistent with mania but were not reported as such (Figure 9).4 Even if these are considered switches, the rate of switching on paroxetine is still low. It seems likely that different antidepressants have different liabilities for causing switching into mania and that newer antidepressants are safer. One report of a large series of 242 bipolar patients treated with SSRIs6 suggests a switch rate of only 3.7%, which was comparable to placebo.

The recommendation to use antidepressants with lithium (if antidepressants are used at all) to “cover” and reduce the possibility of switching into mania is more widely accepted. Unfortunately, the evidence basis for this recommendation in controlled trials is, at best, limited. In fact, in a paroxetine versus imipramine versus placebo in acute bipolar depression study,4 all patients were given combination treatment with lithium in addition to the study drug. In this trial, imipramine-treated patients had a switch rate of 8% despite being on lithium, while placebo-treated patients had a switch rate of only 2%. Imipramine-treated patients did have a lower switch rate when lithium was at a serum level of >0.8 mEq/L, but it was still higher than placebo. The fact that the switch rates reported in this trial represent a total of only four switches out of 117 subjects in the entire study should be considered.

A new study with >700 randomized subjects showed that both olanzapine and a combination of olanzapine-fluoxetine have antidepressant efficacy in bipolar depression.20 In this study, olanzapine alone was statistically superior to placebo but had a relatively small clinical effect. More interestingly, a combination of olanzapine and fluoxetine was superior to both placebo and to olanzapine monotherapy. This study suggests that both olanzapine monotherapy and an olanzapine-fluoxetine combination should be recommended as first-line acute bipolar depression treatments also.

The recommendation for psychosocial treatments in bipolar depression, including psychotherapy, has been essentially universally accepted by the academic community. Unfortunately, there is almost no controlled research on psychotherapy in bipolar disorder. The guideline recommendations on psychosocial or psychotherapeutic treatments are largely derived from the unipolar depression literature, where cognitive-behavioral therapy (in addition to medication) has been shown to be helpful. Psychodynamic psychotherapy, while widely used, also remains untested for bipolar depression.

Treatment of Rapid Cycling

The practice guideline recommendations for treatment of rapid cycling (Table 10)1 are based on a very small and incomplete research database (Table 11). The majority of the literature on treatments for rapid cycling is either in the form of case series or retrospective analyses and only clearly shows that rapid cycling is associated with poor response to lithium and also possibly carbamazepine. Hence, the assumptions that are made elsewhere in the practice guideline reappear in the rapid-cycling section (Table 12). For example, while efficacious mania treatments including lithium and valproate are recommended, atypical antipsychotics and benzodiazepines are omitted despite their documented efficacy in mania. The widely accepted notion that antidepressant use is associated with rapid cycling has been consistently documented in retrospective analyses of current rapid-cycling patients42,43 but has never been tested in a prospective trial (Table 13).42,44-56 The concept of adding rather than substituting, which was discussed earlier, is repeated in this section of the guideline.

The guideline for rapid cycling also discusses whether bipolar I and bipolar II rapid cycling are different. A recently published and well-designed study of rapid-cycling suggests that, in fact, they are quite distinct.57 In this trial, rapid-cycling subjects who were able to tolerate withdrawing from most of their medications took at least 100 mg/day of lamotrigine, and were randomly assigned to either continue on flexible-dose lamotrigine or placebo (withdrawal of lamotrigine) for 6 months. Lamotrigine was efficacious in reducing rapid cycling, and was found in a post hoc analysis be effective in the bipolar II subset but not the bipolar I subset (Figure 10). Due to the use of outcome measures which were not primarily related to relapse into either mania or depression, the potential efficacy of lamotrigine to selectively prevent depression but not mania could not be assessed from the study. However, the results suggest a differential drug response between bipolar I and II rapid-cycling patients.

Maintenance Treatment of Bipolar Disorder

The research literature about maintenance treatment is certainly more modest than that available for manic episode. The most important new development since the publishing of the practice guideline is the FDA approval of lamotrigine for maintenance treatment of bipolar I disorder. Since lithium and lamotrigine are now the only two FDA-approved medications for bipolar maintenance, clearly they should both be considered first-line treatments.

One important and often underappreciated feature of the literature for maintenance treatment is that nearly all of the controlled trials showing that any treatment has maintenance efficacy superior to placebo have done so based on responder-enriched samples. In other words, maintenance studies have tested medications primarily in subjects who were either known acute responders to that medication or had at least been shown to tolerate the study drug before being randomized. For example, the early studies documenting lithium efficacy in maintenance used subjects who had been acutely stabilized on lithium. Recent olanzapine and lamotrigine bipolar maintenance trials have selected only subjects who were either known drug responders or at least able to tolerate the drug being studied.19,57-59

In fact, the one major study that did not select a responder-enriched sample was the divalproex maintenance trial.60 Using an unselected study sample, divalproex could not be shown to have convincing and statistically significant maintenance efficacy when compared to lithium or placebo. However, when a post hoc subanalysis was performed on subjects treated with divalproex in their acute episode prior to the maintenance trial (and hence enriched with divalproex responders) divalproex was found to be efficacious.

Two lamotrigine maintenance trials58,59 selected lamotrigine-tolerating subjects for randomization and what could be considered lamotrigine-responder–enriched samples. The two studies differed in terms of the patients entered: one trial enrolled recently manic subjects, while the other enrolled recently depressed subjects. Both trials found lithium and lamotrigine superior to placebo in maintenance of bipolar disorder when relapse to either mania or depression was analyzed. More interestingly, when the direction of relapse was analyzed separately, lithium was superior to placebo in preventing mania (while lamotrigine was not), and conversely, lamotrigine was superior to placebo in preventing depression (while lithium was not). These studies also provide convincing confirmation of lithium’s efficacy in prophylaxis of mania since the study samples were definitely not enriched with lithium responders.

The results of these well-designed maintenance trials raise two important clinical points. The first is that results were similar regardless of whether the most recent episode was mania or depression. This could have important clinical implications. Common sense might suggest that the most important type of episode to prevent in each patient would be the same as the most recent episode (in other words, recently manic patients should have maintenance treatment focused primarily on prevention of mania while recently depressed patients should have maintenance treatment focused primarily on prevention of depression). These trial results suggest that this distinction may not be necessary and that long-term tendencies to relapse in either direction may be more important.

Secondly, the two active treatments, lithium and lamotrigine, have differences in the direction of their prophylactic efficacy. This may indicate that patients who historically had primarily manic or hypomanic episodes require one type of treatment (either lithium or valproate), while those with a history of depressive episodes require a different type of treatment (either lamotrigine or lithium). Thus, longitudinal course may be more important than recent history in terms of choosing a maintenance treatment. Furthermore, those having both types of episodes may require two treatments simultaneously.

There is nothing inherently unacceptable about using responder-enriched samples in a maintenance trial; in fact, this reflects the almost universal clinical practice of continuing efficacious acute treatments during maintenance. The principal conclusion from all of the bipolar maintenance trials confirms this practice by demonstrating that the first-choice maintenance treatment should be whatever worked for the acute episode. The exception to this observation is the finding that lithium was shown to be an efficacious maintenance treatment for prophylaxis of mania in the lamotrigine trials,59,60 which did not use a lithium-responder–enriched design. This demonstrates that lithium has mania prophylaxis efficacy even in nonresponder-enriched samples.

The inclusion of oxcarbazepine in this section of the guideline is highly questionable. There is no maintenance study of this medication at all in bipolar disorder, much less a placebo-controlled trial. Again, apparently the assumption is that oxcarbazepine works as well as carbamazepine for maintenance. This assumption has never been tested in bipolar disorder.

The issue of antipsychotics in maintenance treatment has largely been avoided by the practice guideline, which recommends antipsychotics only for psychosis or relapse prevention. These drugs, although commonly used for bipolar disorder, lack a clear role in maintenance treatment because there is little clinical research on their effectiveness. Recently, two large trials documented olanzapine’s maintenance efficacy in responder-enriched samples.19,61 The first trial randomly assigned olanzapine mania responders to continue on olanzapine or to withdraw from olanzapine, and to be treated with placebo. Olanzapine was superior to placebo in preventing both mania and depression.19 In the second trial,61 olanzapine combined with either lithium or valproate was compared to placebo combined with either lithium or valproate in maintenance treatment of recently manic responders. The olanzapine plus lithium or valproate was superior to the lithium or valproate alone (olanzapine withdrawal) in preventing relapse to mania.

These studies suggest that olanzapine is a legitimate first-line candidate for both acute and maintenance treatment and that olanzapine should clearly be considered a mood stabilizer. A major concern, though, particularly for olanzapine, is long-term tolerability with regard to metabolic side effects. Olanzapine has been associated with significant weight gain, diabetes, and hyperlipidemia in long-term studies of schizophrenia. Moreover, the magnitude of weight gain is greater than that estimated for lithium or divalproex. Other atypical antipsychotics with fewer long-term metabolic concerns may be important options for maintenance treatment, but there are no data yet available to demonstrate efficacy.

Lithium, carbamazepine, divalproex, lamotrigine, and olanzapine all have some placebo-controlled evidence of maintenance efficacy (Table 14). However, there is limited data available allowing direct comparisons of the efficacy of these medications in bipolar maintenance (Tables 15 and 16). Therefore, treatment choices are typically made based on tolerability and continuing the medication that worked in the acute episode.

Who Should Receive Maintenance Treatment?

The practice guideline recommends maintenance treatment for all bipolar patients. While this common sense recommendation is clearly popular, at least among mental health providers who prescribe medication, it should be considered in the context of the supporting research data. Obviously, all clinical maintenance trials have been conducted only with clearly identified bipolar patients and have included subjects with moderate and severe histories. Clinical trials nearly always intentionally attempt to exclude mildly ill patients because of the likelihood that such subjects may have a high placebo response rate, potentially undermining the ability of the study to show drug-placebo differences.

There are several rationales that suggest that maintenance medication may not be helpful for all patients. First, subjects in clinical trials may not actually represent the patients seen in standard clinical care. It is certainly likely that milder patients may benefit less from maintenance treatment. Second, while placebo has often performed worse than active medication in bipolar maintenance trials, this has not always been the case. For example, a lower-than-expected placebo relapse rate in the divalproex maintenance trial was one of several factors leading to the overall negative results of the study. Lithium was also unable to produce efficacy greater than placebo.61 Finally, the recommendation considers only efficacy in relapse prevention and ignores side effects. It is left to the individual clinician to decide whether the tolerability issues of any specific medication outweigh the potential benefits.


Bipolar disorder is a complex disorder often requiring a complex treatment plan. The APA practice guideline has summarized the research literature and expanded its conclusions on the basis of standard clinical practice. However, the guideline makes assumptions that have limited supporting evidence, including: atypical antipsychotics are more poorly tolerated than lithium or divalproex; combining a new treatment with one that is not working is preferable to switching to a new treatment; oxcarbazepine works as well as carbamazepine; all antidepressants cause mania; and adding lithium to an antidepressant will reduce chances of switching.

The practice guideline does represent a major synthesis of research and clinical opinion distilled into easily understood recommendations. However, most patients do not fit perfectly into any of its recommendations. In order to apply the practice guideline, it is important to understand the underlying assumptions, as well as the limitations of, the evidence on which the guideline is based. Only then can the clinician make a more educated decision about when to go beyond the recommendations in the guideline. PP


1. American Psychiatric Association. Practice guideline for the treatment of patients with bipolar disorder. Am J Psychiatry. 2002;159(suppl 4):1-50.

2. Bowden CL, Brugger AM, Swann AC, et al, for the Depakote Mania Study Group. Efficacy of divalproex vs lithium and placebo in the treatment of mania. JAMA. 1994;271(12):918-924. Erratum in: JAMA. 1994;271(23):1830.

3. Calabrese JR, Bowden CL, Sachs GS, Ascher JA, Monaghan E, Rudd GD. A double-blind placebo-controlled study of lamotrigine monotherapy in outpatients with bipolar I depression. J Clin Psychiatry. 1999;60(2):79-88.

4. Nemeroff CB, Evans DL, Gyulai L, et al. Double-blind, placebo-controlled comparison of imipramine and paroxetine in the treatment of bipolar depression. Am J Psychiatry. 2001;158(6):906-912.

5. Sachs GS, Lafer B, Stoll AL, et al. A double-blind trial of bupropion versus desipramine for bipolar depression. J Clin Psychiatry. 1994;55(9):391-393.

6. Peet M. Induction of mania with selective serotonin reuptake inhibitors and tricyclic antidepressants. Br J Psychiatry. 1994;164(4):549-550.

7. Chou JC. Recent advances in treatment of acute mania. J Clin Psychopharmacol. 1991;11(1):3-21.

8. Ahlfors UG, Baastrup PC, Dencker SJ, et al. Flupenthixol decanoate in recurrent manic depressive illness. A comparison with lithium. Acta Psychiatr Scand. 1981;64(3):226-237.

9. Esparon J, Kolloori J, Naylor GJ, McHarg AM, Smith AH, Hopwood SE. Comparison of the prophylactic action of flupenthixol with placebo in lithium treated manic-depressive patients. Br J Psychiatry. 1986;148:723-725.

10. Chouinard G, Young SN, Annable L. Antimanic effects of clonazepam. Biol Psychiatry. 1983;18(4):451-466.

11. Chouinard G. The use of benzodiazepines in the treatment of manic-depressive illness. J Clin Psychiatry. 1988;49(suppl):15-20.

12. Chouinard G. Clonazepam in acute and maintenance treatment of bipolar affective disorder. J Clin Psychiatry. 1987;48(suppl):29-37.

13. Bradwejn J, Shriqui C, Koszycki D, Meterissian G. Double-blind comparison of the effects of clonazepam and lorazepam in acute mania. J Clin Psychopharmacol. 1990;10(6):403-408.

14. Modell JG, Lenox RH, Weiner S. Inpatient clinical trial of lorazepam for the management of manic agitation. J Clin Psychopharmacol. 1985;5(2):109-113.

15. Lenox RH, Modell JG, Weiner S. Acute treatment of manic agitation with lorazepam. Psychosomatics. 1986;27(suppl 1):28-32.

16. Kane JM, Woerner M, Weinhold P. Incidence and severity of tardive dyskinesia in affective illness. In: Gardos G, Casey DE, eds. Tardive Dyskinesia and Affective Disorders. Washington, DC: American Psychiatric Press; 1984:21-28.

17. Tohen M, Sanger TM, McElroy SL, et al, for the Olanzapine HGEH Study Group. Olanzapine versus placebo in the treatment of acute mania. Am J Psychiatry. 1999;156(5):702-709.

18. Tohen M, Jacobs TG, Grundy SL, et al, for the Olanzapine HGGW Study Group. Efficacy of olanzapine in acute bipolar mania: a double-blind, placebo-controlled study. Arch Gen Psychiatry. 2000;57(9):841-849.

19. Tohen M, Bowden C, Calabrese J, et al. Olanzapine’s efficacy for relapse prevention in bipolar disorder: a randomized double-blind placebo-controlled 12-month clinical trial. Bipolar Disord. 2004;6(4):26-27.

20. Tohen M, Vieta E, Calabrese J, et al. Efficacy of olanzapine and olanzapine-fluoxetine combination in the treatment of bipolar I depression. Arch Gen Psychiatry. 2003;60(11):1079-1088. Erratum in: Arch Gen Psychiatry. 2004;61(2):176.

21. Sachs GS, Grossman F, Ghaemi SN, Okamoto A, Bowden CL. Combination of a mood stabilizer with risperidone or haloperidol for treatment of acute mania: a double-blind, placebo-controlled comparison of efficacy and safety. Am J Psychiatry. 2002;159(7):1146-1154.

22. Jody D, Marcus R, Keck P, et al. Aripiprazole vs. placebo in acute mania. Int J Neuropsychopharmacol. 2002;5(suppl 1):S57.

23. Mullen J, Devine N, Sweitzer D. Quetiapine adjunctive therapy for acute mania associated with bipolar disorder (SIAM). Poster presented at: American Psychiatric Association Annual Meeting; May 17–22, 2003; San Francisco, CA.

24. Bourin M, Auby P, Marcus RN, et al. Aripiprazole versus haloperidol for maintained treatment effect in acute mania. Poster presented at: Annual Meeting of the American Psychiatric Association; May 17–22, 2003; San Francisco, CA.

25. Keck PE Jr, Versiani M, Potkin S, West SA, Giller E, Ice K, for the Ziprasidone in Mania Study Group. Ziprasidone in the treatment of acute bipolar mania: a three-week, placebo-controlled, double-blind, randomized trial. Am J Psychiatry. 2003;160(4):741-748.

26. Sachs GS, Grossman F, Ghaemi SN, Okamoto A, Bowden CL. Combination of a mood stabilizer with risperidone or haloperidol for treatment of acute mania: a double-blind, placebo-controlled comparison of efficacy and safety. Am J Psychiatry. 2002;159(7):1146-1154.

27. Inoue K, Arima S, Tanaka K, Fukui Y, Kato N. A lithium and carbamazepine combination in the treatment of bipolar disorder—a preliminary report. Folia Psychiatr Neurol Jpn. 1981;35(4):465-475.

28. Nolen WA. Carbamazepine, a possible adjunct or alternative to lithium in bipolar disorder. Acta Psychiatr Scand. 1983;67(4):218-225.

29. Lipinski JF, Pope HG Jr. Possible synergistic action between carbamazepine and lithium carbonate in the treatment of three acutely manic patients. Am J Psychiatry. 1982;139(7):948-949.

30. Kramlinger KG, Post RM. Adding lithium carbonate to carbamazepine: antimanic efficacy in treatment-resistant mania. Acta Psychiatr Scand. 1989;79(4):378-385.

31. Desai NG, Gangadhar BN, Channabasavanna SM, Shetty KT. Carbamazepine hastens therapeutic action of lithium in mania. Paper presented at: International Conference on New Directions in Affective Disorders; April 5–9, 1987; Jerusalem, Israel.

32. Keck PE Jr, McElroy SL, Vuckovic A, Friedman LM. Combined valproate and carbamazepine treatment of bipolar disorder. J Neuropsychiatry Clin Neurosci. 1992;3:319-322.

33. Solomon DA, Ryan CE, Keitner GI, et al. A pilot study of lithium carbonate plus divalproex sodium for the continuation and maintenance treatment of patients with bipolar I disorder. J Clin Psychiatry. 1997;58(3):95-99.

34. Chou JC, Czobor P, Charles O, et al. Acute mania: haloperidol dose and augmentation with lithium or lorazepam. J Clin Psychopharmacol. 1999;19(6):500-505.

35. Biederman J, Lerner Y, Belmaker RH. Combination of lithium carbonate and haloperidol in schizo-affective disorder. Arch Gen Psychiatry. 1979;36(3):327-333.

36. Prien RF, Kupfer DJ, Mansky PA, et al. Drug therapy in the prevention of recurrences in unipolar and bipolar affective disorders. Report of the NIMH Collaborative Study Group comparing lithium carbonate, imipramine, and a lithium carbonate-imipramine combination. Arch Gen Psychiatry. 1984;41(11):1096-1104.

37. Garfinkel PE, Stancer HC, Persad E. A comparison of haloperidol, lithium carbonate and their combination in the treatment of mania. J Affect Disord. 1980;2(4):279-288.

38. Muller-Oerlinghausen B, Retzow A, Henn FA, Giedke H, Walden J, for the European Valproate Mania Study Group. Valproate as an adjunct to neuroleptic medication for the treatment of acute episodes of mania: a prospective, randomized, double-blind, placebo-controlled, multicenter study. J Clin Psychopharmacol. 2000;20(2):195-203.

39. Emrich HM. Experience with oxcarbazepine in acute mania. In: Modigh K, Robak OH, Vestergaard P, eds. Anticonvulsants in Psychiatry. Petersfield, UK: Wrightson Biomedical Publishing; 1994:29-36.

40. Zornberg GL, Pope HG Jr. Treatment of depression in bipolar disorder: new directions for research. J Clin Psychopharmacol. 1993;13(6):397-408.

41. Diagnostic and Statistical Manual of Mental Disorders. 4th ed. Washington, DC: American Psychiatric Association; 1994.

42. Altshuler LL, Post RM, Leverich GS, Mikalauskas K, Rosoff A, Ackerman L. Antidepressant-induced mania and cycle acceleration: a controversy revisited. Am J Psychiatry. 1995;152(8):1130-1138.

43. Wehr TA, Sack DA, Rosenthal NE, Cowdry RW. Rapid cycling affective disorder: contributing factors and treatment responses in 51 patients. Am J Psychiatry. 1988;145(2):179-184.

44. Crane GE. The psychiatric side-effects of iproniazid. Am J Psychiatry. 1956;112(7):494-501.

45. Arnold OH, Kryspin-Exner K. The problem of control of manic-depressive processes by antidepressants [in German]. Wien Med Wochenschr. 1965;115(45):929-934.

46. Till E, Vukovic AJ. Int Pharmacopsychiatry. 1970;5:13-17.

47. Coppen A, Brooksbank BW, Peet M. Tryptophan concentration in the cerebrospinal fluid of depressive patients. Lancet. 1972;1(7765):1393.

48. van Scheyen JD, van Kammen DP. Clomipramine-induced mania in unipolar depression. Arch Gen Psychiatry. 1979;36(5):560-565.

49. Siris SG, Chertoff HR, Perel JM. Rapid-cycling affective disorders during imipramine treatment: a case report. Am J Psychiatry. 1979;136(3):341-342.

50. Kukopulos A, Reginaldi D, Laddomada P, Floris G, Serra G, Tondo L. Course of the manic-depressive cycle and changes caused by treatment. Pharmakopsychiatr Neuropsychopharmakol. 1980;13(4):156-167.

51. Lerer B, Birmacher B, Ebstein RP, Belmaker RH. 48-hour depressive cycling induced by antidepressant. Br J Psychiatry. 1980;137:183-185.

52. Ko GN, Leckman JF, Heninger GR. Induction of rapid mood cycling during L-dopa treatment in a bipolar patient. Am J Psychiatry. 198;138(12):1624-1625.

53. Mattsson A, Seltzer RL. MAOI-induced rapid cycling bipolar affective disorder in an adolescent. Am J Psychiatry. 1981;138(5):677-679.

54. Extein I, Pottash AL, Gold MS. Does subclinical hypothyroidism predispose to tricyclic-induced rapid mood cycles? J Clin Psychiatry. 1982;43(7):290-291.

55. Oppenheim G. Drug-induced rapid cycling: possible outcomes and management. Am J Psychiatry. 1982;139(7):939-941.

56. Ghaemi SN, Boiman EE, Goodwin FK. Diagnosing bipolar disorder and the effect of antidepressants: a naturalistic study. J Clin Psychiatry. 2000;61(10):804-808.

57. Calabrese JR, Suppes T, Bowden CL, et al, for the Lamictal 614 Study Group. A double-blind, placebo-controlled, prophylaxis study of lamotrigine in rapid-cycling bipolar disorder. J Clin Psychiatry. 2000;61(11):841-850.

58. Bowden CL, Calabrese JR, Sachs G, et al, for the Lamictal 606 Study Group. A placebo-controlled 18-month trial of lamotrigine and lithium maintenance treatment in recently manic or hypomanic patients with bipolar I disorder. Arch Gen Psychiatry. 2003;60(4):392-400.

59. Goodwin FK, Bowden CL, Calabrese JR, et al. Maintenance treatments for bipolar I depression (lithium, lamotrigine, and placebo). Poster presented at: Annual Meeting of the American Psychiatric Association; May 17–22, 2003; San Francisco, CA.

60. Bowden CL, Calabrese JR, McElroy SL, et al, for the Divalproex Maintenance Study Group. A randomized, placebo-controlled 12-month trial of divalproex and lithium in treatment of outpatients with bipolar I disorder. Arch Gen Psychiatry. 2000;57(5):481-489.

61. Tohen M, Chengappa KN, Suppes T, et al. Relapse prevention in bipolar I disorder: 18-month comparison of olanzapine plus mood stabiliser v. mood stabiliser alone. Br J Psychiatry. 2004;184:337-345.

Dr. Chou is research associate professor at New York University School of Medicine in New York City.

Disclosure: Dr. Chou is a consultant for Abbott, Bristol-Myers Squibb, GlaxoSmithKline, and Pfizer; is on the speaker’s bureaus of Abbott, AstraZeneca, Bristol-Myers Squibb, GlaxoSmithKline, Hoechst, Janssen, Merck, Novartis, Ortho-McNeil, Otsuka, and Pfizer; receives grant and/or research support from Abbott, AstraZeneca, Bristol-Myers Squibb, GlaxoSmithKline, Hoechst, Janssen, Merck, Novartis, Ortho-McNeil, Otsuka, and Pfizer; and is a major stock holder in Bristol-Myers Squibb and Pfizer.

Please direct all correspondence to: James C-Y. Chou, MD, New York University School of Medicine, Department of Psychiatry, NBV 20 20W 13A, 462 1st Ave, New York, NY 10016; Tel: 212-263-6202; Fax: 914-359-7029; E-mail: