Secondary Logo

Journal Logo


Predictors of Relapse in Bipolar Disorder: A Review


Author Information
Journal of Psychiatric Practice®: September 2006 - Volume 12 - Issue 5 - p 269-282


This review examines factors associated with relapse, recurrence, and outcome in bipolar disorder. Bipolar disorder is the sixth leading cause of disability in the United States, affecting a total of 2.3 million U.S. adults.1 Two subtypes of bipolar disorder have been described.1 Bipolar I disorder is characterized by the presence or history of one or more manic or mixed episodes with or without a history of one or more major depressive episodes, while bipolar II disorder requires the presence of one or more major depressive episodes and at least one hypomanic episode.1 Based on National Comorbidity Survey Replication data published in 2005 (N = 9,282), the lifetime prevalence of bipolar I and II disorders is 3.9%.2

In 1991, it was estimated that bipolar disorder placed a $45 billion annual burden on the U.S. economy, including $7 billion in direct treatment costs.3 Over 50% of patients are not correctly diagnosed with bipolar disorder until more than 5 years after their first episode, and this high rate of misdiagnosis adds further to the burden this disorder places on both patients and society.4 Even when correctly diagnosed with type I or II bipolar disorder, fewer than 50% of patients are treated successfully,5 and 10%-15% may take their own lives. Therefore, approximately 250,000 patients in the U.S. who are currently diagnosed with bipolar disorder may eventually die as a result of suicide. According to more recent data from a longitudinal study with over 15,000 bipolar patients, the suicide rate for bipolar patients remains 22 times that of the general population for women and 15 times that of the general population for men.6

Treatment for bipolar disorder remains less than ideal. While the pharmacological guidelines for treatment are well established,7-12 most individuals still have breakthrough episodes or significant residual symptoms while on medication.5 In addition, functional deficits often remain even when patients are in remission.5 Unfortunately, little is known about the factors that may precipitate recurrent bipolar episodes. Because many patients with bipolar disorder remain symptomatic, even while fully adherent to their medication regimens, the need for greater understanding of the reasons behind relapse and recurrence is all the more urgent.5

Problems with adherence further undermine response to treatment and can lead to earlier relapse. Non-adherence also exacerbates social and occupational problems associated with episodes of bipolar disorder. Measured rates of nonadherence to long-term prophylactic pharmacotherapy in bipolar disorder range from 20% to 60%.13

Further complicating the treatment of bipolar disorder is the high comorbidity of substance abuse. More than 60% of patients with bipolar I disorder and 48% of those with bipolar II disorder also meet criteria for a lifetime diagnosis of substance abuse or dependence.14 The relationship between substance use and bipolar disorders has been extensively researched.15,16 Substance use can precipitate an affective episode, prolong its duration, shorten remission, influence severity, and complicate treatment.

This review examines predictors of relapse in bipolar disorders exclusive of substance use. Although problems associated with comorbid bipolar and substance use disorders clearly merit further research, they are beyond the scope of this review. To our knowledge, no reviews of research on predictors of relapse in bipolar disorder as outlined here are currently available. Improved recognition of these predictors may help prevent affective episodes and suggest better strategies for treatment.

Definitions of outcome have been clearly established in the literature on depression.17 However, the terms "relapse," "recurrence," and "episode" are often used interchangeably in the literature on bipolar disorders. Therefore, for the purpose of this review, "episode" refers to the first or index episode as well as any subsequent episode, relapse, or recurrence. "Survival time" refers to the time between episodes (i.e., duration of remission). "Outcome" refers to an evaluation of a number of factors, including improvement or deterioration in number of episodes, need for medication, level of social functioning, and survival time. The term "relapse" refers to the recurrence of an acute mood episode shortly after recovery (e.g., within 8 weeks), while the term "recurrence" generally refers to later episodes. However, since the majority of articles discussed in this review did not distinguish between these two terms, we do not make this distinction here either.


We conducted a literature search on Medline for original research articles published between 1996 and March 2006, using the keywords "relapse," "recurrence," "compliance," "adherence," "life events," "kindling," and "predictors" in combination with the term "bipolar disorder(s)." We supplemented the literature search with a review of the references in the articles that were identified during the initial search and found over 1,000 articles. We then narrowed the search by excluding articles that met the following criteria:

Review articles and meta-analyses

Articles focused on the following sub-populations of bipolar patients: pediatric, elderly, postpartum, Medicaid recipients, HIV positive patients, and patients with comorbid substance use disorders (in order to ensure that our review applied broadly to the population of adult patients with bipolar disorder)

Articles that evaluated the efficacy of bipolar medications but did not predict the effects of medication on future outcomes

Articles that evaluated treatment adherence but did not evaluate the effect of these factors on relapse

Articles evaluating suicide risk but not relapse

Genetic or neurobiological studies that did not predict future outcomes

Studies evaluating specific aspects of bipolar disorder symptoms and diagnosis that did not address future outcomes.

Thirty-eight original research articles met our search criteria. Although predictors of bipolar disorder have historically been understudied, the recent data from the first 2,000 participants in the Systematic Treatment Enhancement Program for Bipolar Disorder (STEP-BD) study addressed many gaps in the current knowledge base concerning these important issues. This longitudinal study, funded by the National Institute of Mental Health (NIMH), allowed evaluation of a very large cohort of patients with bipolar disorder and has high generalizability to the clinical setting.18,19

Based on the content of our final set of articles, we grouped the factors that predicted episodes and outcome of bipolar disorder into six categories:

1. Number of past episodes and interval between episodes

2. Stressful life events

3. Medication treatment

4. Psychodemographic/psychosocial factors

5. Psychotherapy

6. Clinical factors.

Categories 1 and 2 were guided by Post's "kindling" paradigm,20,21 which proposes that repeated subthreshold stimulation (e.g., repeated stress) can induce an episode and that an increased sensitivity and vulnerability to stress occurs with the progression of the illness (i.e., minimal stress can cause a subsequent episode). While this theory remains rather speculative, it has inspired several studies on the correlation between stressful life events and the number of previous episodes as predictors of subsequent episodes.


Number of Past Episodes and Interval Between Episodes

Overall, the number of previous episodes was found to be highly predictive of future relapse (Table 1). In a review of hospital admission records of nearly 3,000 patients with bipolar disorder, Kessing et al.,24,26,27 found that the propensity for relapse increased as the number of past episodes increased, suggesting a deteriorating pattern for bipolar disorder (at least in a subgroup of patients). In theory, these effects could be false artifacts of the data analysis, since those patients with more episodes and more rapid recurrences will comprise a disproportionate amount of patients sampled from a clinical setting. Moreover, within a given observation period, those patients with a greater number of episodes will necessarily have shorter inter-episode intervals. The proposed solution to this sampling bias, which is referred to as "Slater's Fallacy" after Slater's 1938 paper,29 involves within-subject analyses, so that patients are studied individually.30 Nonetheless, in the study by Kessing et al.,27 the individual rate of recurrence still increased with number of episodes, even after controlling for the above effects with statistical regression models. Similarly, in outpatients, Gitlin et al.23 found that patients with a greater number of previous episodes relapsed significantly earlier than those with fewer episodes, suggesting that the time interval between episodes decreases as patients experience more episodes.

Table 1
Table 1:
Studies evaluating the number of past episodes and intervals between episodes as a predictor of relapse in bipolar disorder

In a related (but not entirely corroborative) finding, Goldberg et al.22 observed that of subjects who had been hospitalized with acute mania, those with an interval of less than 1 year between past episodes were more likely to relapse than those with more than a year between past episodes or bipolar controls who had no history of hospitalization. There were no significant differences in the number of previous episodes between the two study groups with histories of hospitalization, so the effect of the number of previous episodes could not be observed. It must be noted, however, that the sample size of this study (N = 39) was significantly lower than that of other studies.24,26,27

In contrast, two studies did not support the previously observed association between number of previous episodes and propensity to relapse. Hlastala et al.25 hypothesized that increased number of previous episodes would predict increased vulnerability to stress, which would then prompt episode onset. However, they found that, regardless of the external stressors experienced by the patients, number of episodes was not a significant predictor of vulnerability to future episodes. Furthermore, Bromet et al. found no predictive effect of number of episodes on propensity for relapse.28 They speculated that their inability to replicate this generally robust finding might be due to the fact that their sample consisted of first-admission patients rather than prevalence samples of consecutively admitted patients as is more usually the case. Moreover, both studies had relatively small sample sizes (N = 64 and N = 123, respectively25,28).

Overall, the majority of research supports the conclusion that an increased number of previous episodes and decreased intervals between these episodes are reliable predictors of relapse in bipolar disorder. Unfortunately, this robust finding remains slightly hindered by the potential confounds originally described by Slater in 1938. As discussed above, while several studies suggested that previous episodes and decreased periods of wellness predict poor outcomes, we would expect a disproportionate number of patients with frequent episodes to be found in most clinical samples. This can lead to higher cycle counts in pooled samples of subjects, which adds a degree of uncertainty to the predictive power of the factors discussed above.30

Stressful Life Events

Seven studies (Table 2) indicated a strong relationship between stressful live events and increased likelihood of episode recurrence18,31,32,34-36 or longer time to recovery.33,35 However, only one of these studies,18 a publication from the STEP-BD study, had a sample size of larger than 100 patients. Those who experienced a bipolar episode during the 2-year follow-up period in one study were significantly more likely to have experienced a severe stressor during the previous 6 months than those who did not.35 Also, patients who experienced a severe stressor after the onset of an episode had a significantly longer recovery time than those who did not.35

Table 2
Table 2:
Studies evaluating stressful life events as a predictor of relapse in bipolar disorder

Although all of the studies we reviewed substantiated these findings, the impact of stressful life events was modified by additional factors. For example, it was found that the effects of stressful life events on relapse were moderated by personality traits, since patients who had high ratings on introversion and obsessionality were more likely to relapse.32

Malkoff-Shwartz et al. hypothesized that the effect of stress was mediated by the degree to which a given stressful event affects the sleep-wake cycle.34 Ratings of social rhythm disruption (the degree to which an event affects the sleep-wake cycle) were observed to have a unique association with the onset of a bipolar episode. This association was significantly stronger for patients with bipolar mania than for those with unipolar or bipolar depression or for those with rapid cycling. This suggests that sleep reduction in response to stress may be a significant predictor of relapse, especially in manic patients; it also suggests that predictors of depressive and manic episodes may not be identical.

While modifying factors do affect individual outcomes, the majority of research in the field, in both small and large studies, supports the notion that stressful life events have an overall negative impact on outcomes in bipolar disorder, both in terms of contributing to relapse and in lengthening time to recovery. Nonetheless, there are limitations on our ability to evaluate the effect of stress on relapse and outcome. Specifically, the definition of a stressful event may vary across studies and individuals, and the reliance on retrospective reporting can introduce a significant amount of recall bias.37

Medication Treatment

Although it is generally presumed that pharmacotherapy reduces the symptoms of bipolar disorder, several studies, the majority of which had sample sizes larger than 100 patients,18,26,28,38,39 revealed that this effect is not as pronounced as might be expected (Table 3). These results, however, must be viewed in light of the indication for which medications are prescribed, a significant potential confound which is discussed in greater detail below. Nonetheless, Post et al. found that, despite appropriate medication treatment and adherence during a 1-year observation period, one-quarter of study participants remained extremely ill, and two-thirds had persistent symptoms.39 In a 5-year study, Keller et al. found that over 50% of patients who relapsed were being treated prophylactically with the highest recommended maintenance doses of medication during the month preceding relapse.38 There is in fact some evidence to suggest that patients who receive more aggressive medication treatment have shorter survival times than those who are treated less aggressively.23 This appears to be the case regardless of the type of medication used. For example, it has been observed that both discontinuing antipsychotic, antimanic, and antidepressant medication and taking antipsychotics and antidepressant medications were associated with longer time to remission, in contrast to never having used them.28 Moreover, in Kessing's analysis of hospital admission records discussed above, it was found that rate of relapse was not reduced despite the introduction and national adoption of newer-generation antidepressant medications.26 In data from the first year of the STEP-BD study, Martinez et al.18 found that a higher number of prior psychiatric medications was predictive of a serious adverse event (e.g. suicidal ideation/ attempt, hospitalization, or relapse). There is also limited evidence to suggest that the propensity for rapid cycling may be lessened if anti-depressants are not introduced early in the course of illness for patients with initial manic or mixed episodes.36

Table 3
Table 3:
Studies evaluating medication treatment or adherence as predictors of relapse in bipolar disorder

Further research is needed, however, with larger and/or more heterogeneous samples to determine whether or not the observed persistence or apparent worsening of symptoms described above was due to confounding factors. It is certainly possible that those with the most severe illness are those who are more aggressively treated and thus causality needs to be more thoroughly addressed. For example, in two of the studies23,39 patients were recruited from university hospitals, which may have meant that they were more likely to be severely ill. Also, in one of these studies,39 the patients with the poorest outcome were more likely to have a positive family history of substance abuse, 10 or more prior depressive episodes, or limited occupational functioning prior to entering the study. In another study, the patients who relapsed despite medication therapy reflected only a subset of the study sample.38 This suggests that the medications were effective in the majority of patients, but that high doses of maintenance medication do not adequately prevent relapse in a subset of severely ill patients. Thus, with regard to all of the studies, it is difficult to discern without further investigation whether the decreased survival time is related to the aggressiveness of the medication therapy or characteristics inherent in the patient's illness. Nonetheless, the possibility that, in some highly refractory bipolar patients, aggressive medication treatment may be contra-indicated does merit investigation.

Psychodemographic/Psychosocial Factors

A number of studies have investigated the relationship between psychodemographic and psychosocial factors and course and outcome in bipolar disorder (Table 4). Among the factors considered were gender, age at onset, self-esteem, social adjustment, job functioning, house-hold income, and attribution style. Factors related to the patients' family and other caregivers, including family psychiatric history, expressed emotion, and family burden, were also analyzed.

Table 4
Table 4:
Studies evaluating psychodemographic/psychosocial factors as predictors of relapse in bipolar disorder

Of five studies that investigated the effects of gender22-24,40,46 only one24 found that female gender puts patients at increased risk for having a future episode. In that case, the authors speculated that the large sample size (N = 2,903) allowed them to find significance for the moderate effects of gender in bipolar disorder. In contrast, three of the five studies had sample sizes of fewer than 100,22,23,40 while the remaining study examined a sample of fewer than 300.46

Earlier age at onset was associated with poor outcome in several cases, being linked with increased risk of recurrence,24,28 and with developing comorbid substance abuse and dependence,36 while it was negatively correlated with episode frequency.36,46 However, age at onset was not found to be a significant variable in several of the studies.18,23,39,42 Moreover, the predictive nature of age at onset in bipolar disorder remains a complex issue. Difficulties include retrospective determination of early onset illness, lack of clarity about cut-points for early, middle, and late age of onset, and lack of prospective studies. This issue is presented in more detail in a review by Leboyer et. al.48

Patients' self-evaluation was not found to be related to patient outcome. Two small studies (N = 27 in both)41,42 investigated the effect of self-esteem on risk of recurrence and found no effect. In one of those studies, however, it was found that poor social adjustment was predictive of an episode.42 Adaptive coping styles may also play a role in preventing recurrence; in a qualitative self-report study of 100 patients with bipolar disorder who had avoided relapse symptoms for over 2 years, it was found that participants' ability to remain attentive to their symptoms contributed to their ability to create and implement intervention strategies.47

Social factors also appear to play a role in predicting relapse. Social support was associated with longer survival time while lack of support predicted relapse,41 and depressive recurrence in particular.35 This relationship was not, however, found in all studies.42 Job functioning was also found to be an important factor. One study observed a strong association between poor occupational functioning and outcomes,40 and three found poor job functioning to be predictive of relapse.22,23,39 However, it should be noted that at least one other small study found no significant association between work functioning and survival time.41 It is notably difficult to establish causality in this situation, since severely ill patients who experience more episodes often function poorly in the workplace. With regard to household income, lower household income may be associated with serious psychiatric adverse events as was found in 1,000 patients in the STEP-BD study who had been monitored for at least 1 year.18 Again, the direction of causality in this case is not clear. Social status, on the other hand, was not found to be significantly associated with survival time in a small study of 27 patients.41

A number of studies investigated the effects of family characteristics and the course of bipolar disorder. Fisfalen et al.46 found a significant association between episode frequency of probands and their affected relatives, and Post et al.39 concluded that a positive family history of substance abuse predicted poorer outcomes. A small study by Stefos et al.41 reported a trend toward an association between survival time and positive psychiatric family history, suggesting that a family history of psychiatric illness might be found to be associated with relapse in patients with bipolar disorder in a larger sample. However, a study by Staner et al. with only 27 patients42 found no significant association between family history and recurrence of episodes.

The expressed emotion of family members has also been investigated. High expressed emotion (in the form of critical comments) has been found to be a predictor of symptom severity, regardless of the type of psychotherapeutic treatment the patient is receiving.43 This significant effect was observed for depressive symptoms only; for manic symptoms, it only approached significance. These authors reported no relation between emotional overinvolvement and manic or depressive symptoms.

Corroborating these findings, Yan et al.45 found a significant effect of expressed emotion on the recurrence of depressive episodes only, but not on manic or any other type of episode. Using a path analysis, Perlick et al.44 observed that higher levels of emotional overinvolvement by family members was related to lower levels of medication adherence at the 7-month follow-up. While it is not discernible from the study data whether family members' expressed emotion was a reaction to patient non-adherence or a cause of it, it is clear that low levels of medication adherence were predictive of a major affective episode at the final follow-up. Interestingly, these findings were unrelated to the perceived burden of the caregivers; perceived burden at baseline did not have an effect on the patients' 7-month medication adherence nor on 15-month outcome.

In summary, while psychosocial (expressed emotion, job functioning, and social support) and psychodemographic (age at onset) factors were found to be associated with patient outcome, not all studies provided conclusive support for these findings and further investigation into the directionality of these effects, especially those supported only by studies with small sample sizes, is warranted.


Seven studies, the majority of which analyzed 100 or more patients,50-52,54,55 addressed the effects of psychotherapy in reducing episode recurrence and improving outcome in bipolar disorder in concert with medication (Table 5). Different psychotherapeutic approaches were employed, including those aimed at helping patients to regulate their interpersonal and social rhythms, as well as cognitive, psychoeducational, and family-focused approaches.

Table 5
Table 5:
Studies evaluating the effects of psychotherapy on relapse in bipolar disorder

Working on the hypothesis that patients with better interpersonal and social rhythms would have improved outcomes, Frank et al.49 developed interpersonal and social rhythm therapy (IPSRT). IPSRT is a manual-based method aimed at enabling patients to reduce medication noncompliance, attain regularity in their daily routines, and cope with disruptions to those routines caused by life events. They found that, over a 2-year period, IPSRT was more effective in enabling patients with bipolar disorder to gain control over their social rhythms when compared to treatment by conventional review of clinical status and symptoms. However, no differences between the two study groups were observed when clinicians assessed changes in manic and depressive symptoms.

In order to assess whether the phase of the patient's illness moderated the effectiveness of IPSRT, Frank et al.54 compared it with an intensive clinical management (ICM) approach in both acute and maintenance phases of bipolar disorder. It was found that IPSRT during the acute phase led to a longer survival time regardless of maintenance treatment assignment. Moreover, those in the IPSRT group had more regular social rhythms by the termination of acute treatment, which was associated with a decreased propensity for recurrence during the maintenance phase. These findings, however, are limited by the fact that possible confounding variables, such as marital status and medical burden, were not equally distributed across the treatment groups.

Cognitive therapy, when designed specifically for relapse prevention in bipolar disorder, led to fewer and shorter episodes, reduced depression and mania, improved time to relapse, improved medication adherence, and higher social functioning than conventional treatment alone.50,55

Psychoeducational programs (defined as 20 individual sessions with the goal of improving patients' awareness of illness, treatment adherence, early detection of prodromal symptoms and relapse, and lifestyle regularity) were also found to be effective in a small group of patients with bipolar disorder and comorbid personality disorders.53 Those receiving psychoeducation had longer survival times, a significantly lower number of total, manic, and depressive relapses, and a reduced likelihood of meeting criteria for recurrence.

Family-focused psychoeducational programs have also proven to be effective, particularly in conjunction with other methods such as IPSRT.51 Patients assigned to individual and integrated family therapy, a psychosocial therapy that combines IPSRT with family-focused treatment, showed greater reductions in depressive symptoms over time than patients assigned to standard community treatment (which consisted of family educational sessions, medication treatment, and crisis management). Family-focused therapy also proved more effective than crisis management (both were conducted in conjunction with pharmacotherapy) in prolonging survival time and decreasing the number of relapses for patients with bipolar disorder. However, the results of this study should be interpreted with caution, because the family-focused treatment group received substantially more time and attention (21 sessions involving all available family members as opposed to two 1-hour sessions for the crisis-management group).52

Overall, these initial data regarding the effectiveness of psychotherapies aimed specifically at patients with bipolar disorder appear quite promising. Because of methodological difficulties such as small sample size (see Table 5 for a comparison of the sizes of samples used), lack of adequately matched control groups,51,52 and demographic differences across comparison groups,54 additional research is needed to confirm this conclusion and to further refine specific therapeutic approaches that target patients with bipolar disorder at different stages of their illness and with different levels of social support.

Clinical Factors

Various clinical characteristics were examined, including polarity sequence, psychiatric history, comorbid disorders, and personality traits (Table 6).

Table 6
Table 6:
Studies evaluating clinical factors as predictors of relapse in bipolar disorder

Whether patients experience predominantly manic or predominantly depressive episodes has been shown to have unique effects on relapse rates. In a 15-year study of 165 patients, Turvey et al.58 found that 33% of patients with a depressive episode at first evaluation experienced a chronic episode lasting through years 14 and 15 of follow-up, compared with 7% of those with a manic episode at first evaluation. Although the difference in these results was only marginally statistically significant, it does suggest that patients with bipolar disorder who have a tendency to have more depressive episodes may have slower recoveries and poorer long-term outcomes than patients with a propensity toward manic episodes. In another long-term study, Coryell et al.57 found that low level functioning in the 5 years before baseline assessment, and the persistence of depressive symptoms during the first 2 years of follow-up, predicted poorer outcome 15 years after initial assessment. These results suggest that both low-level functioning and persistent depressive symptoms may be reliable predictors of poor long-term prognosis in bipolar disorder. More recently, using data from subjects with bipolar I disorder entering the STEP-BD, Perlis et al.60 observed that depressive-onset bipolar disorder was associated with more lifetime depressive episodes and longer periods of depression and anxiety in the year prior to study entry.

Patients' psychiatric history was found to play a role in predicting relapse. In a study that followed 82 patients with bipolar disorder to evaluate outcome in the context of maintenance pharmacotherapy, a history of psychotic symptoms was associated with earlier relapse.23 This finding was replicated in an examination of the first 1,000 patients in the STEP-BD study, which found that a history of psychosis and syndromic baseline mood state were correlated with the occurrence of serious psychiatric adverse events.18

Current comorbidity also played a role in patient outcome. Psychotic features, comorbid alcoholism, and suicidal behavior have all been found to be more prevalent in those with a higher frequency of episodes.46 In an analysis of data from the STEP-BD participants, anxiety disorders have also been found to be correlated with younger age at onset, lower likelihood of recovery, inferior role functioning and life quality, shorter euthymic periods, and greater tendency to attempt suicide, suggesting that patients with comorbid anxiety disorders should be carefully monitored and/or should be subject to special interventions.59 Another analysis of STEP-BD data showed that residual symptoms at recovery, whether depressive or manic, and proportion of days with depressed, anxious, or elevated mood in the preceding year have also been associated with shorter time to recurrence, once again suggesting that close attention should be paid to patient symptomatology even at recovery.19 Finally, Bromet et al. found that clinically elevated baseline levels of anxiety and depression predict lengthened time to remission.28

In addition, the effect of certain personality features has been investigated with mixed results. Patients who had introverted and/or obsessive personalities were more likely to relapse even under conditions of relatively low stress (i.e., minor life events). Conversely, extroversion and low levels of obsessionality appeared to provide protection from relapse, even under highly stressful conditions such as the death of a loved one or divorce.31 Finally, Hammen et al. found that an individual's level of sociotropy (deriving a sense of self-worth from contact with others) or autonomy (deriving self-worth from autonomous achievements) may not be predictive of increased vulnerability to either negative interpersonal or autonomous achievement events, as previously hypothesized.56


Although the available literature on predictors of bipolar episodes has historically been rather sparse, it has been growing recently, most notably with the publication of the first data from the NIMH-funded STEP-BD study.59 At present, a greater number of previous episodes, shorter intervals between episodes, history of psychosis, history of anxiety, persistence of affective symptoms and episodes (particularly a history of depressive episodes and symptoms), and stressful life events are the strongest predictors of future episodes and poor outcome in patients with bipolar disorder. These factors can and should be assessed in the course of any psychiatric evaluation because they are likely to inform mental health professionals' future treatment choices in many significant ways.

There is some evidence to support the predictive value of poor job functioning, lack of social support, increased expressed emotion (i.e., criticism and overinvolvement) by family members, and the personality traits of introversion and obsessionality with regard to poor outcome. These factors, although less reliably linked to outcome, could also contribute to prognostic considerations at intake and throughout ongoing treatment and therefore should be monitored by clinicians on a regular basis. Finally, the data implicating demographic and psychosocial factors such as gender, age at onset, level of self-esteem, and poor social adjustment in predicting outcomes and episodes in patients with bipolar disorder are much less conclusive. At present these data are of weaker predictive value with regard to relapse and should be used for prognostic consideration only with caution.

Counterintuitively, some data indicate that pharmacological treatment alone may not necessarily improve outcomes in the treatment of severe bipolar illness. However this controversial finding could be confounded by sampling bias towards sicker patients receiving more complex and aggressive pharmacotherapy. The literature also points to several protective factors: adherence to treatment and good social support have been shown to moderately improve survival time between episodes. Moreover, medication management combined with some form of psychotherapy provides significant benefits and may in fact constitute the optimal treatment for this illness.

Finally, although patients with comorbid substance abuse were excluded from this review, given the high incidence of substance use disorders in this population, the prognostic impact of substance abuse coexisting with either bipolar I or II disorder merits careful investigation in future studies.

Future Directions

The available research on the predictors of relapse in bipolar disorder presented in this review suggests a number of areas for further investigation. Although both shortened intervals between episodes and increased number of previous episodes have been shown to be predictors of poor outcome in bipolar disorder, further research is warranted to determine whether these seemingly interrelated factors are independent predictors or are intercorrelated. The effect of the type of episode in predicting outcome also needs further study.

While there is clearly evidence that stressful life events can lead to bipolar episodes and lengthen recovery periods from these episodes, the role of possible mediating factors, such as personality traits, social support, and sleep, is also of interest. Larger studies that can explore the interactions between stressful life events and numerous possible mediating variables are therefore warranted. In addition, the role of expressed emotion within the families of patients with bipolar disorder and the efficacy of family interventions suggest the need for more research on the reciprocal effects of bipolar illness on families and of families' emotional responses on illness outcome. Specific investigation of such questions in patients with bipolar disorder who have young children would also be useful.

Several studies in this review suggest that a subset of patients with bipolar disorder do not respond to pharmacotherapy. As it is often the patients with the most severe bipolar illness who are nonresponsive to aggressive medication therapy, such treatment resistance could be due to illness progression or to desensitization to pharmaceutical agents over time. Medication nonadherence was also not evaluated as a possible confounding factor. Thus lack of response to medication may not be the sole explanation for these poor outcomes, warranting further investigation in this area. Future studies investigating the responsiveness of patients with bipolar disorder to traditional versus aggressive medication treatment, while controlling for severity of illness and medication adherence, are strongly warranted. In addition, more research is needed to clarify the impact of early use of antidepressants, varied dosing regimens, and individual classes of medications on relapse.

Findings from current research investigating the effect of demographic and psychosocial factors on the course of bipolar illness are inconclusive. One exception is fairly strong evidence supporting poor job functioning as a predictor of bipolar episodes. While this may seem to suggest that poor job function leads directly to increased episodes and poorer outcomes in bipolar disorder, it is somewhat of a "chicken and egg" phenomenon. Does low level of functioning in the workplace predict future episodes in bipolar disorder or do more severely ill patients who are prone to more episodes function poorly in the workplace? Is job-related stress a factor in relapse? Further research is needed to address these questions and to guide clinicians in their recommendations concerning job-related activities for patients and their caregivers.

Finally, the impact of caregivers' expressed emotion on the course of bipolar illness is very intriguing, with several recent studies suggesting that it can play a significant role in relapse. The lack of research on caregiver burden in bipolar illness compared to that in other chronic illnesses, such as cancer, Alzheimer's disease, and even schizophrenia, is striking. Additional studies are needed to tease out the effects of caregiver variables such as stress level and coping strategies on expressed emotion. Research such as this could potentially open the door to treatment strategies that would provide both patients and their families with effective tools to avert relapse.

Similarly, more research on the role of personality pathology in relapse is needed to help clinicians design specific psychotherapeutic strategies. All the studies in this review that evaluated psychotherapy combined with medication management in bipolar disorder, both with and without comorbid personality pathology, supported the benefits of such combination treatment over psychopharmacological management alone. However, more research is needed to establish which psychotherapeutic modalities may be most beneficial for patients with bipolar disorder in general as well as for those with various types of personality pathology.

Clinical Implications

Our review supports several basic guidelines for the evaluation and treatment of patients with bipolar illness:

The number of episodes and duration and sequence of intervals between episodes should be carefully monitored since this information is crucial in defining the progression of the illness and treatment response.

The treatment focus should be on aggressive episode prevention and on equally aggressive extension of the time when the patient is asymptomatic.

Patients with a history of psychosis or anxiety merit increased care and vigilance.

Patients with more frequent depressive episodes may have worse outcomes and longer recoveries than those with more frequent manic episodes and should therefore be monitored especially closely.

In the event of stressful life events, closer monitoring of the patient's condition is warranted with concomitant education on stress management and adaptive coping skills.

Medication treatment combined with psychotherapy may be superior to medication treatment alone. In addition, some modes of family psychotherapy and psychoeducation appear to improve outcomes.

Although research in this area is limited, trends suggest that clinicians should strive to reduce the impact of expressed emotion during patients' depressive and mixed episodes, and to provide patients with tools to more effectively communicate with their families and cope with family conflict.


In conclusion, bipolar disorder is a severe and crippling disease and one of the leading causes of disability in the United States. Despite remarkable advances in pharmacotherapy over the past several decades, treatment outcome is often poor and relapse remains a continuing problem. Identification of predictors of relapse and poor outcome is critical in maximizing prevention and optimizing treatment strategies. This review surveyed available literature on the topic and identified several promising avenues for future research. While the current findings are intriguing, there is still considerable need for additional investigation.


1. American Psychiatric Association. Diagnostic and statistical manual of mental disorders, 4th edition, text revision. Washington, DC: American Psychiatric Association; 2000.
2. 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.
3. Kleinman L, Lowin A, Flood E, et al. Costs of bipolar disorder. Pharmacoeconomics 2003;21:601-22.
4. Lewis L. Mood disorders: Diagnosis, treatment, and support from a patient perspective. Psychopharmacol Bull 2001;35:186-96.
5. National Institute of Mental Health (NIMH). Breaking ground, breaking through: The strategic plan for mood disorders research. Washington, DC: NIMH; July 2002 (available at
6. Osby U, Brandt L, Correia N, et al. Excess mortality in bipolar and unipolar disorder in Sweden. Arch Gen Psychiatry 2001;57:844-50.
7. Chou JC-Y. Review and update of the American Psychiatric Association practice guideline for bipolar disorder. Primary Psychiatry 2004;11:73-84.
8. American Psychiatric Association. Practice guideline for the treatment of patients with bipolar disorder (revision). Am J Psychiatry 2002;159(4 Suppl):1-50.
9. Keck PE, Perlis RH, Otto MW, et al. The expert consensus guideline series: Treatment of bipolar disorder 2004. Postgrad Med Spec Rep 2004(December):1-116.
10. Suppes T, Dennehy EB, Hirschfeld RM, et al. and the Texas Consensus Conference Panel on Medication Treatment of Bipolar Disorder. The Texas implementation of medication algorithms: Update to the algorithms for treatment of bipolar I disorder. J Clin Psychiatry 2005;66:870-86.
11. Yatham LN, Kennedy SH, O'Donovan C, et al. and the Canadian Network for Mood and Anxiety Treatments. Canadian Network for Mood and Anxiety Treatments (CAN-MAT) guidelines for the management of patients with bipolar disorder: Consensus and controversies. Bipolar Disord 2005;7(Suppl 3):5-69.
12. Kowatch RA, Fristad M, Birmaher B, et al. and the Child Psychiatric Workgroup on Bipolar Disorder. Treatment guide-lines for children and adolescents with bipolar disorder. J Am Acad Child Adolesc Psychiatry 2005;44:213-35.
13. Lingam R, Scott J. Treatment non-adherence in affective disorders. Acta Psychiatr Scand 2002;105:164-72.
14. Grant BF, Stinsin FS, Dawson DA, et al. Prevalence and cooccurrence of substance use disorders and independent mood and anxiety disorders. Arch Gen Psychiatry 2004;61:807-16.
15. Goldberg JF. Bipolar disorder with comorbid substance abuse: Diagnosis, prognosis, and treatment. J Psychiatr Pract 2001;7:109-202.
16. Tohen M, Greenfield SF, Weiss RD, et al. The effect of comorbid substance use disorders on the course of bipolar disorder: A review. Harv Rev Psychiatry 1998;6:133-41.
17. Frank E, Prien RF, Jarrett RB, et al. Conceptualization and rationale for consensus definitions of terms in major depressive disorder: Remission, recovery, relapse, and recurrence. Arch Gen Psychiatry 1991;48:851-5.
18. Martinez JM, Marangell LB, Simon NM, et al. Baseline predictors of serious adverse events at one year among patients with bipolar disorder in STEP-BD. Psychiatr Serv 2005;56:1541-8.
19. Perlis RH, Ostacher MJ, Patel JK, et al. Predictors of recurrence in bipolar disorder: Primary outcomes from the Systematic Treatment Enhancement Program for Bipolar Disorder (STEP-BD). Am J Psychiatry 2006;163:217-24.
20. Post RM, Rubinow DR, Ballenger JC. Conditioning and sensitization in the longitudinal course of affective illness. Br J Psychiatry 1986;149:191-201.
21. Post RM. Transduction of psychosocial stress into neurobiology of recurrent affective disorder. Am J Psychiatry 1992;149:999-1010.
22. Goldberg JF, Harrow M. Kindling in bipolar disorders: A longitudinal follow-up study. Biol Psychiatry 1994;35:70-2.
23. Gitlin MJ, Swendsen J, Heller TL, et al. Relapse and impairment in bipolar disorder. Am J Psychiatry 1995;152:1635-40.
24. Kessing LV, Andersen PK, Mortensen PB, et al. Recurrence in affective disorder: I. Case register study. Br J Psychiatry 1998;172:23-8.
25. Hlastala SA, Frank E, Kowalski J, et al. Stressful life events, bipolar disorder, and the "kindling model." J Abnorm Psychology 2000;109:777-86.
26. Kessing LV, Hansen MG, Andersen PK. Course of illness in depressive and bipolar disorders: Naturalistic study, 1994-1999. Br J Psychiatry 2004;185:372-7.
27. Kessing LV, Hansen MG, Andersen PK, et al. The predictive effect of episodes on the risk of recurrence in depressive and bipolar disorders: A life-long perspective. Acta Psychiatr Scand 2004;109:339-44.
28. Bromet EJ, Finch SJ, Carlson GA, et al. Time to remission and relapse after the first hospital admission in severe bipolar disorder. Soc Psychiatry Psychiatr Epidemiol 2005;40:106-13.
29. Slater E. Zur periodic des manisch-depressiven. Irreseins Z Gesamte Neurol Psychiatrie 1938;162:794-801.
30. Oepen G, Baldessarini RJ, Salvatore P. On the periodicity of manic-depressive insanity, by Eliot Slater (1938): translated excerpts and commentary. J Affect Disord 2004;78:1-9.
31. Swendsen J, Constance H, Heller T, et al. Correlates of stress reactivity in patients with bipolar disorder. Am J Psychiatry 1995;152:795-7.
32. Hammen C, Gitlin M. Stress reactivity in bipolar patients and its relation to prior history of disorder. Am J Psychiatry 1997;154:856-7.
33. Johnson SL, Miller I. Negative life events and time to recovery from episodes of bipolar disorder. J Abnorm Psychol 1997;106:449-57.
34. Malkoff-Schwartz S, Frank E, Anderson BP, et al. Social rhythm disruption and stressful life events in onset of bipolar and unipolar episodes. Psychol Med 2000;30:1005-16.
35. Cohen AN, Hammen C, Henry RM, et al. Effects of stress and social support on recurrence in bipolar disorder. J Affect Disord 2004;82:143-7.
36. Ernst CL, Goldberg JF. Clinical features related to age at onset in bipolar disorder. J Affect Disord 2004;82:21-7.
37. Dohrewend BP. Inventorying stressful life events as risk factors for psychopathology: Toward resolution of the problem of intracategory variability. Psychol Bull 2006;132: 477-95.
38. Keller MB, Lavori PW, Coryell W, et al. Bipolar I: A five-year prospective follow-up. J Nerv Ment Dis 1993;181:238-45.
39. Post RM, Denicoff KD, Leverich GS, et al. Morbidity in 258 bipolar outpatients followed for 1 year with daily prospective ratings on NIMH life chart method. J Clin Psychiatry 2003;64:680-90.
40. Goldberg JF, Harrow M, Grossman LS. Recurrent affective syndromes in bipolar and unipolar mood disorders at follow-up. Br J Psychiatry 1995;166:382-5.
41. Stefos G, Bauwens F, Staner L, et al. Psychosocial predictors of major affective recurrences in bipolar disorder: 4-year longitudinal study of patients on prophylactic treatment. Acta Psychiatr Scand 1996;93:420-6.
42. Staner L, Tracy A, Dramaix M, et al. Clinical and psychosocial predictors of recurrence in recovered bipolar and unipolar depressives: A one-year controlled prospective study. Psychiatry Res 1997;69:39-51.
43. Kim EY, Miklowitz DJ. Expressed emotion as a predictor of outcome among bipolar patients undergoing family therapy. J Affect Disord 2004;82:343-52.
44. Perlick DA, Rosenheck RA, Clarkin JF, et al. Impact of family burden and affective response on clinical outcome among patients with bipolar disorder. Psychiatric Serv 2004;55:1029-35.
45. Yan LJ, Hammen C, Cohen AN, et al. Expressed emotion versus relationship quality variables in the prediction of recurrence in bipolar patients. J Affect Disord 2004;83:199-206.
46. Fisfalen ME, Schulze TG, DePaulo JR, Jr, et al. Familial variation in episode frequency in bipolar affective disorder. Am J Psychiatry 2005;162:1266-72.
47. Russell SJ, Browne JL. Staying well with bipolar disorder. Aust NZ J Psychiatry 2005;39:187-93.
48 Leboyer HCM, Paillere-Martinot ML, Bellivier F. Age at onset in bipolar affective disorders: A review. Bipolar Disord 2005;7:111-8.
49. Frank E, Hlastala S, Ritenour A, et al. Inducing lifestyle regularity in recovering bipolar disorder patients: Results from the maintenance therapies in bipolar disorder protocol. Biol Psychiatry 1997;41:1165-73.
50. Lam DH, Watkins ER, Hayward P, et al. A randomized controlled study of cognitive therapy for relapse prevention for bipolar affective disorder. Arch Gen Psychiatry 2003;60:145-52.
51. Miklowitz DJ, Richards JA, George EL, et al. Integrated family and individual therapy for bipolar disorder: Results of a treatment development study. J Clin Psychiatry 2003;64:182-91.
52. Miklowitz DJ, George EL, Richards JA, et al. A randomized study of family-focused psychoeducation and pharmacotherapy in the outpatient management of bipolar disorder. Arch Gen Psychiatry 2003;60:904-12.
53. Colom F, Vieta E, Sanchez-Moreno J, et al. Psychoeducation in bipolar patients with comorbid personality disorders. Bipolar Disorders 2004;6:294-8.
54. Frank E, Kupfer DJ, Thase ME, et al. Two-year outcomes for interpersonal and social rhythm therapy in individuals with bipolar I disorder. Arch Gen Psychiatry 2005;62:996-1004.
55. Lam DH, Hayward P, Watkins ER, et al. Relapse prevention in patients with bipolar disorder: Cognitive therapy outcome after 2 years. Am J Psychiatry 2005;162:324-9.
56. Hammen C, Ellicott A, Gitlin M, et al. Sociotropy/autonomy and vulnerability to specific life events in patients with unipolar depression and bipolar disorders. J Abnorm Psychol 1989;2:154-60.
57. Coryell W, Turvey C, Endicott J, et al. Bipolar I affective disorder: Predictors of outcome after 15 years. J Affect Disord 1998;50:109-16.
58. Turvey CL, Coryell WH, Arndt S, et al. Polarity sequence, depression, and chronicity in bipolar I disorder. J Nerv Ment Dis 1999;187:181-7.
59. Simon NM, Otto MW, Wisniewski SR, et al. Anxiety disorder comorbidity in bipolar disorder patients: Data from the first 500 participants in the Systematic Treatment Enhancement Program for Bipolar Disorder (STEP-BD). Am J Psychiatry 2004;161:2222-9.
60. Perlis RH, Delbello MP, Miyahara S, et al. Revisiting depressive-prone bipolar disorder: Polarity of initial mood episode and disease course among bipolar I Systematic Treatment Enhancement Program for Bipolar Disorder participants. Biol Psychiatry 2005;58:549-53.

bipolar disorder; predictors of outcome; predictors of episodes; relapse; recurrence; mania; depression

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