Secondary Logo

Journal Logo


Maintenance Treatment of Bipolar Disorder: Applying Research to Clinical Practice


Author Information
Journal of Psychiatric Practice®: September 2006 - Volume 12 - Issue 5 - p 283-299


Bipolar disorder as defined in the Diagnostic and Statistical Manual of Mental Disorders (DSM-IV-TR) is conceptualized as a disorder of episodes, specifically episodes of mania, hypomania, and depression.1 Most of the treatments recommended for bipolar disorder were initially studied in acute mania and some were subsequently also studied in maintenance. The medications initiated during the acute episode are often continued into maintenance. However, patients spend considerably more time in maintenance than in acute treatment,2 and the clinical issues that drive treatment decisions differ during acute and maintenance treatment. In acute treatment, the emphasis is on symptom reduction while the emphasis is on tolerability during maintenance treatment.

In recent years, renewed research interest in maintenance treatment has resulted in several large, industry-sponsored, double-blind, placebo-controlled maintenance studies in bipolar I disorder, after a period of over two decades during which no such studies were published. This sudden production of high quality information has important clinical implications.

In this review, we discuss the controlled maintenance trials that have been done, outline their strengths and limitations, and make recommendations for both clinical practice and future research. In the interest of focusing on maintenance treatment, no studies of treatment for acute mania, hypomania, or depression are included except as noted.


This review examines the literature on controlled studies of maintenance treatment for bipolar disorder. We included only randomized controlled studies that involved patients with bipolar I or II disorder who were entered into the study while not acutely symptomatic. To identify such studies, we searched the MEDLINE (1966-October 2005) and EMBASE (1980-2005) databases using the following terms related to diagnosis: bipolar disorder, bipolar disorder/prevention and control, recurrence/prevention and control, and double-blind method as well as combinations of these terms. We also searched terms related to drug therapy including bipolar disorder/drug therapy, anticonvulsants, atypical antipsychotics, lithium, carbamazepine, valproate, valproic acid, olanzapine, quetiapine, aripiprazole, clozapine, ziprasidone, and risperidone. Although the literature search examined citations through 2005, some citations were subsequently updated with more recent published reports.

Studies that enrolled subjects during an acute episode who were then continued into maintenance treatment were included only if a separate analysis for the maintenance phase was published. Both placebo-controlled and active-controlled studies were included. All studies described as blinded, including single-blind studies, were included regardless of quality of the blinding. Different blinding methods introduce varying degrees of bias: in single-blind studies, the rater is blinded to the treatment, but the trial subject is not. This allows the possibility that the patient's knowledge of the treatment may influence the outcome. In contrast, double-blind studies, in which both the subject and the rater are blinded to the treatment, are the least likely to introduce any bias that would influence the outcome. Studies that allowed subjects with schizoaffective disorder or unipolar depression to be enrolled were included only if separate analyses were reported for the subset of subjects with bipolar disorder, and only those separate analyses are included here. Studies that combined the study drug with another drug (i.e., combination studies) were included. Responder-enriched studies were included, but with a discussion of the potential effects of enrichment such as selection bias. Responder-enriched studies systematically exclude subjects who are already known either to be unable to tolerate the study drug or to be refractory to the study drug. These studies go beyond short-term studies of drug effects and look at the ability of the drug to sustain its effect over time. They are powerful because they most closely simulate clinical practice, where a drug's ability to maintain its effects over time has implications for relapse and recurrence of the patient's illness. When data sets were reported in multiple publications, only the primary publication was included. Pooled analyses of multiple studies were not included.

Both authors independently reviewed the studies generated by the search strategy to ensure that they met the inclusion criteria described above. Any disagreements were resolved by a consensus discussion. Information included in the tables was independently extracted by both authors.

All results for relapse to mania, mixed episode, or depression are reported. Studies that reported dropout rates but failed to separate relapse from other reasons for dropout were excluded. The primary analyses are emphasized for each study; important secondary analyses are mentioned parenthetically. The primary outcome measure is time to relapse (survival analysis), although some of the earlier studies did not use survival analysis methods and instead just reported number of relapses. Survival analysis is preferable to simple relapse rates because of its ability to account for how long patients remained well. Trials that found statistical significance rates of difference between treatments of > 0.05 are reported as negative studies.


Early Studies of Lithium

Table 1 lists five early placebo-controlled studies of lithium maintenance treatment that did not use survival analysis methods. Most of these trials did not utilize power calculations in advance and had a low sample size. We included these studies for historical completeness. While these five studies indicate lithium's efficacy as a maintenance treatment, at least three of these studies3,5,6 were enriched with lithium responders. The total number of subjects randomized to lithium was 91, with 97 randomized to placebo.

Table 1
Table 1:
Early placebo-controlled studies of lithium maintenance without survival analysis methods: Principal results

There was also one early lithium maintenance study that did use survival analysis techniques.8 This study had only 18 subjects, 9 randomized to lithium and 9 to placebo (Table 2). Interestingly, this was the only early lithium maintenance trial that included a statistical analysis, which did not demonstrate a statistically superior outcome for lithium.

Table 2
Table 2:
Early placebo-controlled lithium maintenance trial with survival analysis

These six studies produced early confidence in the efficacy of lithium as a maintenance treatment in subjects most of whom had previously responded to lithium for acute mania. Lithium soon became the gold standard for bipolar maintenance. Since these early studies, lithium has also become the standard active control for bipolar maintenance trials. While directional efficacy (mania prevention versus depression prevention) was not addressed in most of these studies, the report from Dunner et al.7 suggested efficacy in preventing hypomania even though the subjects, described as having bipolar II disorder, were selected for a stronger history of depression. The suggestion that lithium might have greater efficacy in preventing mania than in preventing depression was further developed in more modern studies. While most early studies did not separately report relapse into depression versus relapse into mania, the fact that most of the subjects in these trials had most recently had a manic episode would suggest a directional maintenance efficacy in prevention of mania.9,10

Two randomized, single-blind, placebo-controlled studies of lithium in recently manic and recently depressed bipolar subjects have also been done11-14(Table 3). In these two studies, lithium was superior to placebo in preventing mania in recently manic subjects as well as superior to placebo in preventing depression in recently depressed subjects. While these studies were single-blind, responder-enriched, and did not use survival analysis techniques, they did provide additional evidence of lithium's efficacy as a bidirectional maintenance agent-that is, in providing maintenance of remission from both mania and depression. These studies also suggest that the direction of the index episode preceding maintenance treatment predicts the most likely direction of subsequent relapse during the maintenance phase. This was also supported by all subsequent placebo-controlled bipolar maintenance trials we reviewed. Nearly all of these trials used patients in an index manic episode. While the bidirectional maintenance efficacy of lithium compared with placebo was demonstrated in the two studies listed in Table 3, these studies did not attempt to compare its relative efficacy in preventing mania versus preventing depression.

Table 3
Table 3:
Randomized, placebo-controlled, single-blind lithium bipolar maintenance trials: Principal results

Trials of Other Agents That Did Not Use Survival Analysis

Table 4 summarizes placebo-controlled bipolar maintenance trials involving other treatments, some of which were also compared to lithium. Okuma et al.15 were unable to show a statistically significant difference between carbamazepine and placebo in a small study. Kane et al.16 studied a small number of depressed subjects with bipolar II disorder and reconfirmed that lithium had efficacy superior to either imipramine or placebo in preventing relapse. Prien et al.17 demonstrated lithium's efficacy in preventing mania. This dataset was later reanalyzed by Shapiro et al.18 using survival analysis (see discussion later in this section). Naylor et al.19 suggested that adding methylene blue to lithium might improve maintenance efficacy. Giannini et al.,20 in a poorly designed study, could not demonstrate a difference between lithium and verapamil in maintenance efficacy. Hartong et al.21 compared lithium and carbamazepine separately in a responder-enriched sample and a non-responder-enriched sample. They suggested that lithium had superior efficacy for maintenance in subjects with bipolar II disorder and in subjects with mild bipolar I disorder who were not previously on maintenance medication. Pazzaglia et al.22,23 used a placebo crossover design in subjects with lithium-refractory, ultra rapid-cycling bipolar disorder and found that the calcium channel blocker nimodipine reduced cycle frequency more than placebo.

Table 4
Table 4:
Randomized, double-blind, controlled bipolar maintenance trials of other treatments without survival analysis methods: Principal results
Table 4
Table 4:

Randomized, Double-Blind Studies Using Survival Analysis

Table 5 summarizes randomized, controlled studies that did not use a placebo control but still provided clinically useful information. Gelenberg et al.24 randomly assigned known lithium responders to maintenance treatment with either standard (0.8-1.0 mEq/L) or low (0.4-0.6 mEq/L) lithium levels and showed that the standard level had superior efficacy for maintenance treatment although it produced more side effects. Unfortunately, this study did not examine maintenance lithium levels between 0.6-0.8 mEq/L, which are often used in clinical practice. Perlis et al.26 subsequently reanalyzed this dataset and suggested that relapse was most likely in subjects who experienced a reduction in their lithium levels, since those subjects who had been on low levels prior to the study and continued in the low dose group did not have higher rates of relapse.

Table 5
Table 5:
Randomized, double-blind bipolar maintenance trials using survival analysis methods (not placebo-controlled)

Denicoff et al.25 randomly assigned a nonenriched bipolar sample to lithium, carbamazepine, or to a combination of lithium and carbamazepine and found that the combination was superior to either drug alone in preventing mania. These results extended findings from an earlier study that had suggested a similar conclusion for acute mania.27 The subjects in this maintenance trial were possibly more severely ill than those in most studies as indicated by 61% having a history of rapid cycling and 52% a history of psychosis

Prien et al.17 and Shapiro et al.18 compared lithium, imipramine, and a combination of lithium and imipramine in bipolar maintenance treatment. The study allowed enrollment of both recently manic and recently depressed subjects as long as they had responded to a lithium-imipramine combination. They found imipramine alone to be less efficacious in preventing mania than either lithium alone or lithium plus imipramine. There was no indication that including imipramine in the regimen had any advantage over lithium alone. Interestingly, they also found that lithium combined with imipramine was superior to imipramine alone in preventing depression, again demonstrating that lithium has efficacy in preventing depression. However, imipramine alone was numerically superior to lithium alone in preventing depression in recently depressed subjects.

Table 6 summarizes eight randomized, placebo-controlled maintenance studies that used survival analysis. An early study by Quitkin et al.28 compared lithium plus imipramine to lithium alone and found that imipramine increased the risk of mania; however, this study was not designed to be a controlled study of lithium. The other seven studies, which were published between 2000 and 2006, represent a new era of research on the maintenance treatment of bipolar disorder.8,9,29-33 All of these studies were funded by the manufacturers of the drugs under study, and most were designed with the intent to obtain Food and Drug Administration (FDA) approval for maintenance treatment. Each of these studies was considerably larger than the earlier studies described above. In total across these 7 studies, over 500 subjects were randomized to placebo, over 370 to lithium, over 350 to lamotrigine, and over 250 to olanzapine, compared with a total from the literature prior to 1985 of about 400 subjects randomized to lithium monotherapy and about 220 randomized to placebo monotherapy.

Table 6
Table 6:
Randomized placebo-controlled double-blind bipolar maintenance trials using survival analysis methods
Table 6
Table 6:

In addition to size, these studies have other advantages that reflect improvements in research methodology including use of survival analysis, better description of subjects, use of improved rating instruments, and better conceptualized hypotheses, including hypothesized differences concerning the direction of efficacy (preventing mania versus preventing depression).



While lithium has not recently enjoyed the research attention that has been given to drugs the pharmaceutical industry has recently been promoting, based on the extensive evidence from earlier trials reviewed above, lithium continues to be considered the gold standard for maintenance treatment. Lithium is the drug that has been best studied in bipolar maintenance and has shown efficacy in preventing both mania and depression.7,17,18 Notably, in the two modern maintenance trials that focused on lamotrigine9,10 lithium was the active control. In these two studies, lithium was found to have greater maintenance efficacy than placebo, even though these studies were partially enriched with lamotrigine responders and were primarily designed to study lamotrigine. These two studies also found that lithium had greater maintenance efficacy in preventing mania than depression, supporting similar findings from earlier studies.7,11-14,17,18

It has also been suggested that lithium has efficacy in preventing suicide. Uncontrolled retrospective studies have found the use of lithium to be associated with significant reductions in suicidality.34-36 A retrospective study suggested that lithium was associated with less suicidality than divalproex.37 In addition, discontinuation of lithium has been associated with a rebound increase in suicidality above baseline levels.38 A recent meta-analysis of randomized trials of lithium versus placebo that examined suicidal behavior39 further confirmed that lithium reduces suicidality, although many of these studies involved unipolar depression. Lithium has also been found to have greater efficacy in preventing suicide than carbamazepine in a prospective maintenance study.40


Calabrese et al.30 reported an ambitious attempt to demonstrate the efficacy of lamotrigine in maintenance treatment of rapid-cycling patients. This study allowed enrollment of subjects with both bipolar I and II disorder. A flexible dose design was used and the mean lamotrigine dose was 288 mg/day. The endpoint used in this study was simply a need for additional clinical intervention rather than relapse as defined on a rating scale. While the primary analysis was negative, some secondary analyses, including time to study discontinuation, favored lamotrigine. Lamotrigine appeared to have more efficacy in rapid-cycling subjects with bipolar II than bipolar I disorder, suggesting efficacy in preventing depression.

Two additional maintenance studies led to the FDA approval of lamotrigine for maintenance treatment of bipolar I disorder.9,10 These two studies used parallel designs but differed in that one included only subjects whose most recent episode had been manic9 while the other included only subjects whose most recent episode had been depressed.10 While multiple fixed doses of lamotrigine were initially planned, only a single fixed dose of 200 mg/day was finally reported. A large number of subjects were lost during a lead-in period designed to select only patients who tolerated lamotrigine. Together, these two trials demonstrated that lamotrigine is more efficacious than placebo in preventing relapse in this partially responder-enriched sample (i.e., in patients known to tolerate lamotrigine). Lamotrigine was also more efficacious in preventing depression regardless of the direction of the index episode in these lamotrigine-responder-enriched samples. Notably, even though these studies enriched the samples with lamotrigine responders, lithium was found to be superior to placebo in preventing mania in both studies. These studies were the first bipolar maintenance studies to show differences in directional efficacy between treatments. Unfortunately, no inferences about dose-response are possible since only a single dose was reported. It should be noted that, prior to this lamotrigine study that enrolled recently depressed subjects,10 only a few attempts involving small numbers of subjects had been made to study maintenance treatment following an index depressed episode.17,21

Lamotrigine was extremely well tolerated in these studies with no side effects occurring more often than with placebo under randomized conditions.41 Lithium caused more diarrhea and tremor than lamotrigine. Notably, while the labeling for lamotrigine includes a black box warning about serious rash, this risk has been shown to be directly related to the starting dose and rate of dose escalation. In these bipolar trials, lamotrigine was titrated gradually and produced no increased risk of either serious or benign rash compared to placebo.


Although commonly used and recommended for maintenance treatment, divalproex has not been approved by the FDA for a maintenance treatment indication. The pivotal divalproex maintenance trial of recently manic subjects by Bowden et al.,29 which was notable for not having used a responder-enriched design, failed to show statistically significant differences in efficacy among divalproex, lithium, and placebo. Two problems with this study included a lower than expected rate of relapse on placebo as well as a higher than expected rate of relapse on lithium. In fact, in this study, lithium was actually numerically (although not statistically significantly) inferior to placebo. The fact that the survival analysis showed a trend towards superiority of divalproex over placebo (p = 0.06) has been used to support use of divalproex for maintenance. In addition, in a secondary analysis of subjects who were taking divalproex prior to randomization (probable divalproex responders), divalproex was found to have greater maintenance efficacy than placebo, and this was due to preventing mania.29

Gyulai et al.42 explored the possibility that divalproex might have bidirectional maintenance efficacy in a posthoc analysis of the data from the Bowden et al. study,29 which suggested that divalproex may also have efficacy in preventing depression, although this was not one of the primary hypotheses of the study which, by design, allowed inclusion only of recently manic subjects.

Divalproex (mean level 84 μg/mL) was better tolerated than lithium in the study by Bowden et al.,29 but this may have been due to relatively high lithium levels (mean 1.0 mEq/L) causing increased side effects. Divalproex side effects included weight gain, tremor, sedation, alopecia, infection, and tinnitus. While this study was not designed to detect polycystic ovarian syndrome, other non-randomized reports have suggested that this may be a concern for divalproex in young women.43


Among the atypical antipsychotics, olanzapine has enjoyed the most research attention in bipolar maintenance trials. Two large placebo-controlled trials of olanzapine in recently manic patients who responded to olanzapine have been published. One was a monotherapy trial,33 and the other was a study of olanzapine versus placebo in combination with either lithium or divalproex.32 In these studies, olanzapine was found to be more efficacious than placebo in maintenance treatment either alone or in combination with lithium or divalproex. The maintenance efficacy of olanzapine has been primarily demonstrated in prevention of mania; since the olanzapine maintenance trials entered only recently manic subjects, its efficacy in preventing depression has not been adequately studied. All of the olanzapine maintenance trials have used flexible dose designs with doses ranging from 5-20 mg/day and mean doses ranging from 8-12 mg/day. The olanzapine combination was not superior on the primary outcome measure, but only in a subset of patients who achieved both syndromal and symptomatic recovery at the time of randomization. Fewer than 30% of the enrolled patients met criteria for randomization, despite treatment with full-dose olanzapine combined with lithium or valproate in the open phase. The superiority of the combination in the maintenance phase in the subset of patients for whom data were secondarily analyzed was for delayed time to mania, but not for time to depression. Olanzapine had higher rates than placebo of some troublesome side effects including weight gain, increased cholesterol, prolactin elevation, prolonged QTc on electrocardiogram, and akathisia.32


A recent study of aripiprazole in maintenance treatment randomized recently manic subjects who were responsive to aripiprazole to maintenance treatment with either flexible dose aripiprazole or placebo.31 This study found that aripiprazole was superior to placebo in efficacy for maintenance treatment, and this was due primarily to its efficacy in preventing mania. The study included few subjects who relapsed into depression and hence was not a suitable test for inferring whether aripiprazole might prevent depressive episodes. Flexible dosing of aripiprazole was used in this study with a mean dose of 24 mg/day.


The most obvious conclusion from this literature review is that bipolar maintenance trials are extraordinarily complicated to conduct. Research designs must cope with many important questions including use of responders only, enrolling subjects with mild versus more severe illness, active control or placebo, direction of index episode, monotherapy or combination, duration of stability required, bipolar I or II, and many others. Each study makes compromises that limit the generalizability of its results.

The vast majority of maintenance trials have used responder-enriched samples. While this might be viewed as a scientific compromise that is used to increase the likelihood of finding a difference from placebo, it does simulate standard clinical care in which acute treatments that are efficacious are continued into the maintenance phase of treatment. It is also clear that the vast majority of maintenance trials enrolled only recently manic subjects and thus primarily studied prevention of mania.

Placebo-controlled, responder-enriched studies have shown that lithium, lamotrigine, olanzapine, and aripiprazole were superior to placebo in their primary endpoints. Lithium, olanzapine, and aripiprazole have shown efficacy in preventing mania, while lithium and lamotrigine have shown efficacy in preventing depression. In a non-responder-enriched, placebo-controlled design, divalproex did not show maintenance efficacy in the primary analysis, but did show efficacy in mania prevention in a secondary responder-enriched analysis.

Medication Dosing

Unfortunately, to our knowledge, only one dose-finding study in bipolar maintenance treatment has been done.24 This study showed that higher lithium levels are more efficacious than lower lithium levels. In addition, a reanalysis of the results of this study26 found that patients whose lithium dose was reduced were the most likely to relapse. Other studies of lithium and all studies of other drugs either used flexible dosing or a single fixed dose. Thus, decisions about dosing must be made clinically and cannot rely on research evidence. Whether maintenance dosing should be lower than dosing used in acute treatment has also not been studied. Most maintenance studies have selected doses that were either comparable to, or only slightly lower than, doses used in acute mania studies.

What Is a Mood Stabilizer?

No medication has been approved by the FDA as a "mood stabilizer." This commonly used yet inconsistently defined term was also omitted from the American Psychiatric Association Practice Guideline for the Treatment of Patients with Bipolar Disorder.44 Bauer and Mitchner45 used a strict definition of the term that included only treatments that have shown both acute and maintenance efficacy for both depression and mania in placebo-controlled studies without destabilizing the disorder. A list of drugs meeting this definition would essentially be limited to lithium. Others have proposed broader definitions that might include dozens of drugs as possible mood stabilizers.46,47 It appears unlikely that a clear consensus will develop on the meaning of this term.

Should Maintenance Treatment Simply Be a Continuation of the Acute Treatment?

Acute episodes of bipolar disorder can be severely disruptive or disabling. Thus, when clinicians have stabilized a patient following an acute episode, there is often a hesitation to change the medication that helped. An attitude of not wanting to change something that is working can easily lead to simply continuing the acute treatment. In fact, the use of responder-enriched designs in most of the controlled maintenance literature mimics this practice.

Furthermore, the majority of relapses in maintenance trials have occurred early in the study, usually in the first 3 months. While this may suggest that many of the studies that were termed "maintenance" were in fact studying continuation treatment, both common sense and the literature suggest that reducing or discontinuing efficacious acute treatments too soon is unwise.

However, the two recent partially responder-enriched lamotrigine maintenance trials,9,10 because of their innovative design, raise the possibility of considering an alternative approach. These trials were unique in that they studied stable subjects who had been recently manic and recently depressed separately. The next episode following randomization was most often in the same direction as the index episode, but a minority of subjects did relapse into the opposite direction. In both studies, lamotrigine prevented depression while lithium prevented mania. This would suggest that, in bipolar subjects who have a history of frequent episodes in one direction, the predominant historical direction of episodes as well as the index episode should be considered.

How Long Should Maintenance Treatment Continue?

The APA Practice Guideline44 recommends indefinite maintenance treatment for most patients with bipolar I disorder. Many patients, however, are unwilling to comply with the daunting task of a lifetime of maintenance medication. As noted above, our review of maintenance trials revealed that most of the relapses in maintenance trials occurred early in the trial, on both placebo as well as on active medication. Patients who have been stable on maintenance medication often wonder after what period of time they might be able to discontinue the medication without a significantly increased risk of relapse. Clearly as more time passes after an episode, the risk of relapse diminishes.

Unfortunately, several lines of research do lead to a conclusion that indefinite maintenance treatment is truly necessary for bipolar I disorder. Discontinuation of lithium in stable patients has been associated with relapse, subsequent lithium-refractoriness, and a significant increase in suicidality.38,48 Reduction of lithium levels in stable patients has also been associated with increased relapse.26 Furthermore, a secondary analysis of the lamotrigine maintenance studies,49 which excluded relapse within the first 6 months of the trial, found that lamotrigine continued to be superior to placebo in maintenance treatment even for the last 12 months of the 18 month study. This suggests that even patients who have been stable for 3 months or longer still have an increased risk of relapse if their medication is reduced or discontinued.

In addition, any factors that are associated with a poorer prognosis in bipolar disorder such as severe or frequent episodes; subsyndromal symptoms; poor interepisode functioning; or the presence of suicidality, rapid cycling, or psychosis should all be additional reasons to continue efficacious maintenance treatment indefinitely.

Drug Combinations versus Monotherapy

While many patients may receive multiple medications during maintenance treatment, research has not addressed the issue of drug combinations versus monotherapy under controlled conditions. Whether partially nonresponsive symptoms or partial relapse should be handled by an increase in the primary drug or addition of a second drug has not been studied. In general, the APA Practice Guideline recommends first optimizing the dose of the current drug and then adding another agent if the response is still inadequate. Clearly, many patients with bipolar disorder are taking multiple medications.50

Clinically, one problem that develops with the use of multiple medications is interpreting drug response and side effects. The only way to make the determination that the combination is more effective than either drug is for a patient to first have failed to respond to adequate trials of monotherapy with both medications and then to have responded to a trial of the combination treatment.51 This ideal is certainly difficult to accomplish in clinical practice. If a patient has a clinical response (or develops a side effect) after a medication is added, how does the clinician decide if the response (or side effect) is due to the second medication or is uniquely due to the combination? Simplification of complex medication regimens may make interpretation of drug response clearer.

Continuing Antidepressants in Maintenance

Although antidepressants are generally not recommended for bipolar maintenance because of concerns about inducing a switch to mania and the possibility of causing cycling,52 they are commonly used both as a second-line treatment for acute bipolar depression and maintenance treatment of depression in bipolar I and II disorder. The APA Practice Guideline recommends that antidepressants should be discontinued as soon as possible in patients with bipolar disorder due to concerns about inducing a switch to mania.

This concern about induction of switching is due principally to earlier studies of imipramine in acute bipolar depression and may not be as relevant with newer drugs. Peet et al.53 suggested that newer antidepressants have not caused more switching than placebo in short-term studies, raising the question of whether the newer antidepressants really are more likely to cause switching. A recent report from Leverich et al.54 provided controlled evidence that switching is more likely with venlafaxine than with bupropion or sertraline. Thus, tricyclic antidepressants, venlafaxine, and probably monoamine oxidase inhibitors appear to be associated with an increased risk of switching. Whether this also applies to the other newer antidepressants has not been determined.

Whether antidepressants in bipolar maintenance are associated with an increased risk of mania is a parallel concern to that of switching during acute depression. All guidelines caution against using antidepressants longer than necessary in bipolar maintenance. However, based on a retrospective study, Altshuler et al.55 suggested that rapid discontinuation of antidepressants leads to higher rates of relapse into depression. However, the study used an "enriched" sample of patients who had done well acutely on antidepressants (i.e., who had responded without switching). This may suggest a subgroup with a distinct pattern of response.

Thus, while prolonged antidepressant use in maintenance may be associated with an increased risk of mania, early discontinuation may also be associated with an increased risk of depression in patients known to be good antidepressant responders.

Comorbid Substance Abuse

As many as 60% of patients with bipolar disorder have at some point had a comorbid substance use disorder. While these rates are very high for lifetime diagnosis, cross-sectionally, the rates in study samples are, in general, substantially lower.56-58 Patients with bipolar disorder who have comorbid substance abuse have poorer outcomes including more frequent episodes and rapid cycling,59 more severe manic episodes,60 diminished quality of life,61 more nonadherence to medication,62 and a very high lifetime rate of suicide attempts.63

Etiologic theories of comorbidity between bipolar disorder and substance abuse range from genetic to psychobehavioral, with genetic vulnerability theories growing in importance. Increased rates of substance abuse have been found in family members of patients who have bipolar disorder and comorbid substance abuse. In addition, patients with bipolar disorder without substance abuse who have a family history of substance abuse are more likely to later develop substance abuse themselves as well as to have more severe and frequent bipolar episodes.60 A related point is that the short-term implications for course of illness differ for lifetime substance use disorder and substance use disorder that co-occurs with presenting bipolar episodes. These findings suggest that preventive strategies for bipolar disorder should focus on substance abuse as well as on mood episodes.

Risk factors for comorbid substance abuse include male gender, lower education level, and other Axis I disorders, especially anxiety disorders.64 A specific new diagnosis of "dual mania" has been proposed to describe a manic or mixed episode with current or past substance use disorder.65 Patients with such comorbidity have also been found to have a high level of disability and to display increased self-harming behavior.

Most reviews advocate integrated/simultaneous treatment of comorbid bipolar disorder and substance abuse,66 with a focus on remission and improving functional outcomes rather than solely on symptom reduction.67 A logical approach would be to treat the bipolar disorder and substance use disorder concurrently using both neuromodulating agents for the mood component and anti-appetitive agents, including medications such as topiramate, naltrexone, acamprosate, rimonabant, and buprenorphine, for the addiction component. Unfortunately, such strategies have not been systematically studied.

Controlled maintenance trials have categorically excluded subjects with bipolar disorder who have comorbid substance abuse or dependence, so that there is limited research on maintenance treatment of patients with comorbid substance use disorders. Kosten and Kosten68 reviewed the characteristics of an ideal agent for this population and suggested that GABA agonists should be promising. One study of acute treatment of patients with bipolar I disorder (including depressed subjects) who had comorbid alcohol dependence69 found that divalproex added to lithium (and other treatments) reduced alcohol use more than placebo. McIntyre et al.66 also recommended topiramate and lithium and mentioned other potential treatments including gabapentin, carbamazepine, and atypical antipsychotics. Brown70 recently reviewed the literature in this area.

Several clinical conclusions are clear:

Bipolar patients with suspected substance abuse should receive specific treatment for the substance abuse disorder.

The diagnostic criteria for substance abuse and dependence may be too stringent for patients with bipolar disorder who are so vulnerable to many of the consequences of substance abuse.

Patients with bipolar disorder who use substances at levels that do not meet full DSM-IV criteria for substance abuse1 should be counseled regarding the special risks of worsening the course of bipolar disorder associated with substance abuse.

While treatment guidelines caution prescribers about the risks of abusing substances while taking medications for bipolar disorder, the thorny issue of whether to consider medication holidays during periods of substance abuse has not been systematically studied. In particular, if substance use increases the risk of relapse, then discontinuation of the medication is likely to further increase the risk of relapse. Clearly, continuing some medications is more of a concern than others. In particular, lithium has a low therapeutic index and may be subject to toxicity in patients who become dehydrated from using alcohol. Sedating medications may potentiate the effects of sedating drugs of abuse such as opioids or benzodiazepines. Alcohol or sedative/hypnotic withdrawal as well as intoxication with stimulants can lead to seizures which might be exacerbated by antipsychotics or antidepressants.

In contrast, antipsychotics and anticonvulsants are generally thought to be fairly safe in medically ill patients and are often used in intensive care settings. It seems probable that staying on these medications, or even on lithium, might be clinically helpful even in the face of substance abuse if the risk of acute medical problems is tolerable.

New Directions for Future Research

Available research on bipolar maintenance treatment leaves many clinical questions unaddressed. Key areas for future research include more studies of recently depressed and rapid cycling patients, and studies of patients with psychotic bipolar and schizoaffective disorder. In addition, innovative designs that can help differentiate the numerous treatments that are used in bipolar disorder would be welcome. In particular, the potential role of atypical antipsychotic medications as first-line options for bipolar maintenance treatment, either as monotherapy or in combination with lithium or anticonvulsants, should be better studied. The recent focus on metabolic complications of antipsychotics in schizophrenia should be expanded to patients with bipolar disorder who are exposed not only to antipsychotics but also to other medications, such as lithium and divalproex, with metabolic side effects. Although clonazepam and lorazepam have been studied in acute mania, no maintenance studies of benzodiazepines in bipolar disorder have been done, even though this class of drugs is often used in bipolar maintenance treatment. Research on maintenance treatment using newer antidepressants is sorely needed, since these drugs are commonly used. The development of genetic or other biological markers that are correlated with specific drug response may also be possible in the foreseeable future.


1. American Psychiatric Association. Diagnostic and statistical manual of mental disorders, 4th edition, text revision. Washington, DC: American Psychiatric Association; 2000.
2. Judd LL, Akiskal HS, Schettler PJ, et al. The long-term natural history of the weekly symptomatic status of bipolar I disorder. Arch Gen Psychiatry 2002;59:530-7.
3. Baastrup PC, Poulsen JC, Schou M, et al. Prophylactic lithium: Double blind discontinuation in manic-depressive and recurrent depressive disorders. Lancet 1970;2:326-30.
4. Coppen A, Nogubra R, Bailey J, et al. Prophylactic lithium in affective disorders. Lancet 1971;2:275-9.
5. Cundall RL, Brooks PW, Murray LG. A controlled evaluation of lithium prophylaxis in affective disorders. Psychol Med 1972;2:308-11.
6. Hullin RP, McDonald R, Allsopp MNE. Prophylactic lithium in recurrent affective disorders. Lancet 1972;1:1044-6.
7. Dunner DL, Stallone F, Fieve RR. Lithium carbonate and affective disorders, V: A double-blind study of prophylaxis of depression in bipolar illness. Arch Gen Psychiatry 1976;33:117-20.
8. Melia PI. Prophylactic lithium: A double-blind trial in recurrent affective disorders. Br J Psychiatry 1970;116:621-4.
9. Bowden CL, Calabrese JR, Sachs G, et al. 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;60:392-400.
10. Calabrese JR, Bowden CL, Sachs G, et al. A placebo-controlled 18-month trial of lamotrigine and lithium maintenance treatment in recently depressed patients with bipolar I disorder. J Clin Psychiatry 2003;64:1013-24.
11. Prien RF, Caffey EM, Klett CJ. Prophylactic efficacy of lithium carbonate in manic-depressive illness: Report of the Veterans Administration and National Institute of Mental Health Collaborative Study Group. Arch Gen Psychiatry 1973;28:337-41.
12. Prien RF, Klett CJ, Caffey EM. Lithium prophylaxis in recurrent affective illness. Am J Psychiatry 1974;131:198-203.
13. Prien RF, Caffey EM, Klett CJ. Factors associated with treatment success in lithium carbonate prophylaxis. Arch Gen Psychiatry 1974;31:189-92.
14. Prien RF, Klett CJ, Caffey EM. Lithium carbonate and imipramine in prevention of affective episodes: A comparison in recurrent affective illness. Arch Gen Psychiatry 1973;29:420-5.
15. Okuma T, Inanaga K, Otsuki S, et al. A preliminary double-blind study on the efficacy of carbamazepine in prophylaxis of manic-depressive illness. Psychopharmacology 1981;73:95-6.
16. Kane JM, Quitkin RF, Rifkin A, et al. Lithium carbonate and imipramine in the prophylaxis of unipolar and bipolar II illness: A prospective, placebo-controlled comparison. Arch Gen Psychiatry 1982;39:1065-9.
17. Prien Rf, Kupfer DJ, Mansky PA, et al. Drug therapy in the prevention of recurrences in unipolar and bipolar affective disorders. Arch Gen Psychiatry 1984;41:1096-104.
18. Shapiro Dr, Quitkin FM, Fleiss JL. Response to maintenance therapy in bipolar illness: Effect of index episode. Arch Gen Psychiatry 1989;46:401-5.
19. Naylor GJ, Martin B, Hopwood E, et al. A two-year double-blind crossover trial of the prophylactic effect of methylene blue in manic-depressive psychosis. Society of Biological Psychiatry 1986;21:915-20.
20. Giannini AJ, Taraszewski R, Loiselle RH. Verapamil and lithium in maintenance therapy of manic patients. J Clin Pharmacol 1987;27:980-2.
21. Hartong EG, Moleman P, Hoogduin CAL, et al. Prophylactic efficacy of lithium versus carbamazepine in treatment-naive bipolar patients. J Clin Psychiatry 2003;64:144-51.
22. Pazzaglia PJ, Post RM, Ketter TA, et al. Preliminary controlled trial of nimodipine in ultra-rapid cycling affective dysregulation. Psychiatry Res 1993;49:257-72.
23. Pazzaglia PJ, Post RM, Ketter TA, et al. Nimodipine monotherapy and carbamazepine augmentation in patients with refractory recurrent affective illness. J Clin Psychopharmacol 1998;18:404-13.
24. Gelenberg AJ, Kane JM, Keller MB, et al. Comparison of standard and low serum levels of lithium for maintenance treatment of bipolar depression. N Engl J Med 1989;321:1489-93.
25. Denicoff KD, Smith-Jackson EE, Disney ER, et al. Comparative prophylactic efficacy of lithium, carbamazepine, and the combination in bipolar disorder. J Clin Psychiatry 1997;58:470-8.
26. Perlis RH, Sachs GS, Lafer B, et al. Effect of abrupt change from standard to low serum levels of lithium: A reanalysis of double-blind lithium maintenance data. Am J Psychiatry 2002;159:1155-9.
27. Kramlinger KG, Post RM. Adding lithium carbonate to carbamazepine: Antimanic efficacy in treatment-resistant mania. Acta Psychiatr Scand 1989;79:378-85.
28. Quitkin FM, Kane J, Rifkin A, et al. Prophylactic lithium carbonate with and without imipramine for bipolar patients: A double-blind study. Arch Gen Psychiatry 1981;38:902-7.
29. Bowden CL, Calabrese JR, McElroy SL, et al. A randomized, placebo-controlled 12-month trial of divalproex and lithium in treatment of outpatients with bipolar I disorder. Divalproex Maintenance Study Group. Arch Gen Psychiatry 2000;57:481-9.
30. Calabrese JR, Suppes T, Bowden CL, et al. A double-blind, placebo-controlled, prophylaxis study of lamotrigine in rapidcycling bipolar disorder. J Clin Psychiatry 2000;61:841-50.
31. Keck PE Jr, Calabrese JR, McQuade RD, et al. A randomized, double-blind, placebo-controlled 26-week trial of aripiprazole in recently manic patients with bipolar I disorder. J Clin Psychiatry 2006;67:626-37.
32. Tohen M, Chengappa NR, Suples 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;184:337-45.
33. Tohen M, Calabrese JR, Sachs GS, et al. Olanzapine versus placebo in the prevention of relapse in bipolar disorder. Am J Psychiatry 2006;163:247-56.
34. Tondo L, Baldessarini RJ, Hennen J. Lithium and suicide risk in bipolar disorder. Primary Psychiatry 1999;6:51-6.
35. Baldessarini RJ, Tondo L, Viguera AC. Discontinuing lithium maintenance treatment in bipolar disorders: Risks and implications. Bipolar Disorders 1999;1:17-24.
36. Baldessarini RJ, Tondo L, Hennen J. Lithium treatment and suicide risk in major affective disorders: Update and new findings. J Clin Psychiatry 2003; 64(suppl 5):44-52.
37. Goodwin FK, Fireman B, Simon GE, et al. Suicide risk in bipolar disorder during treatment with lithium and divalproex. JAMA 2003;290:1467-73.
38. Baldessarini RJ, Tondo L, Hennen J. Effects of lithium treatment and its discontinuation on suicidal behavior in bipolar manic-depressive disorders. J Clin Psychiatry 1999;60(Suppl 2):77-84.
39. Cipriani A, Pretty H, Hawton K, et al. Lithium in the prevention of suicidal behavior and all-cause mortality in patients with mood disorders: A systematic review of randomized trials. Am J Psychiatry 2005;162:1805-19.
40. Thies-Flechtner K, Muller-Oerlinghausen B, Seibert W, et al. Effect of prophylactic treatment on suicide risk in patients with major affective disorders: Data from a randomized prospective trial. Pharmacopsychiatry 1996;29:103-7.
41. Goodwin GM, Bowden CL, Calabrese JR, et al. A pooled analysis of 2 placebo-controlled 18-month trials of lamotrigine and lithium maintenance in bipolar I disorder. J Clin Psychiatry 2004;65:432-41.
42. Gyulai L, Bowden CL, McElroy SL, et al. Maintenance efficacy of divalproex in the prevention of bipolar depression. Neuropsychopharmacology 2003;28:1374-82.
43. Joffe H, Cohen LS, Suppes T, et al. Valproate is associated with new-onset oligomenorrhea with hyperandrogenism in women with bipolar disorder. Biol Psychiatry 2006;59:1078-86.
44. American Psychiatric Association. Practice guideline for the treatment of patients with bipolar disorder. Am J Psychiatry 2002;159(suppl):1-50.
45. Bauer MS, Mitchner L. What is a "mood stabilizer"? An evidence-based response. Am J Psychiatry 2004;161:3-18.
46. Bowden CL. New concepts in mood stabilization: Evidence for the effectiveness of valproate and lamotrigine. Neuropsychopharmacol 1998;19:194-9.
47. Sachs GS. Bipolar mood disorder: Practical strategies for acute and maintenance phase treatment. J Clin Psychopharmacol 1996;16(2 Suppl 1):32S-47S.
48. Suppes T, Baldessarini RJ, Faedda GL, et al. Risk of recurrence following discontinuation of lithium treatment in bipolar disorder. Arch Gen Psychiatry 1991;48:1082-8.
49. Calabrese JR, Goldberg JF, Ketter TA, et al. Recurrence in bipolar I disorder: A post hoc analysis excluding relapses in two double-blind maintenance studies. Biol Psychiatry 2006;59:1061-4.
50. Blanco C, Laje G, Olfson M, et al. Trends in the treatment of bipolar disorder by outpatient psychiatrists. Am J Psychiatry 2002;159:1005-10.
51. Preskorn SH, Lacey RL. Polypharmacy: When is it rational? J Pract Psychiatry Behav Health 1995;1:92-8.
52. Wehr TA, Sack DA, Rosenthal NE, et al. Rapid cycling affective disorder: Contributing factors and treatment responses in 51 patients. Am J Psychiatry 1988;145:179-84.
53. Peet M. Induction of mania with selective serotonin re-uptake inhibitors and tricyclic antidepressants. Br J Psychiatry 1994;151:549-50.
54. Leverich GS, Altshuler LL, Frye MA, et al. Risk of switch in mood polarity to hypomania or mania in patients with bipolar depression during acute and continuation trials of venlafaxine, sertraline, and bupropion as adjuncts to mood stabilizers. Am J Psychiatry 2006;163:232-9.
55. Altshuler L, Suppes T, Black D, et al. Impact of antidepressant discontinuation after acute bipolar depression remission on rates of depressive relapse at 1-year follow-up. Am J Psychiatry 2003;160:1252-62.
56. Regier DA, Farmer ME, Rae DS, et al. Comorbidity of mental disorders with alcohol and other drug abuse. Results from the Epidemiologic Catchment Area (ECA) Study. JAMA 1990;264:2511-8.
57. Grant BF, Stinson FS, Dawson DA, et al. Co-occurrence of 12-month alcohol and drug use disorders and personality disorders in the United States: Results From the National Epidemiologic Survey on Alcohol and Related Conditions. Arch Gen Psychiatry 2004;61:361-8.
58. Kessler RC, Berglund P, Demler O, et al. Lifetime prevalence and age-of-onset distributions of DSM-IV disorders in the National Comorbidity Survey Replication. Arch Gen Psychiatry 2005;62:593-602.
59. Feinman JA, Dunner DK. The effect of alcohol and substance abuse on the course of bipolar affective disorder. J Affect Disord 1996;37:43-9.
60. Nolen WA, Luckenbaugh DA, Altshuler LL, et al. Correlates of 1-year prospective outcome in bipolar disorder: Results from the Stanley Foundation Bipolar Network. Am J Psychiatry 2004;161:1447-54.
61. Weiss RD, Ostacher MJ, Otto MW, et al. Does recovery from substance use disorder matter in patients with bipolar disorder? J Clin Psychiatry 2005;66:730-5.
62. Sherwood Brown E, Suppes T, Adinoff B, et al. Drug abuse and bipolar disorder: Comorbidity or misdiagnosis? J Affect Disord 2001;65:105-15.
63. Dalton EJ, Cate-Carter TD, Mundo E, et al. Suicide risk in bipolar patients: The role of co-morbid substance use disorders. Bipolar Disord. 2003;5:58-61.
64. Sonne SC, Brady KT, Morton WA, et al. Substance abuse and bipolar affective disorder. J Nerv Ment Dis 1994;182:349-52.
65. Haro JM, van Os J, Vieta E, et al. Evidence for three distinct classes of typical, psychotic and dual mania: Results from the EMBLEM study. Acta Psychiatr Scand 2006;113:112-20.
66. McIntyre RS, Konarski JZ, Yatham LN. Comorbidity in bipolar disorder: A framework for rational treatment selection. Hum Psychopharmacol 2004;19:369-86.
67. O'Brien CP, Charney DS, Lewis L, et al. Priority action to improve the care of persons with co-occurring substance abuse and other mental disorders: A call to action. Biol Psychiatry 2004;56:703-13.
68. Kosten TR, Kosten TA. New medication strategies for comorbid substance abuse and bipolar affective disorders. Biol Psychiatry 2004;56:771-7.
69. Salloum IM, Cornelius JR, Daley DC, et al. Efficacy of valproate maintenance in patients with bipolar disorder and alcoholism: A double-blind placebo-controlled study. Arch Gen Psychiatry 2005;62:37-45.
70. Brown ES. Bipolar disorder and substance abuse. Psychiatr Clin North Am 2005;28:415-25.

bipolar disorder; maintenance treatment; lithium, divalproex; carbamazepine; lamotrigine; olanzapine; aripiprazole; anticonvulsant; antipsychotic

Copyright © 2006 Wolters Kluwer Health, Inc. All rights reserved.