Objective: The boundary between borderline personality disorder (BPD) and bipolar disorder (BD) is a controversial subject. Clinically, it can be difficult to diagnose patients who present with both affective instability and impulsivity. This paper reviews concepts and challenges related to the overlap of these disorders.

Methods: A Medline search was conducted, using the key words borderline personality disorder, bipolar disorder, affective disorder, and personality disorder. Reference lists from articles generated were also used. Publications from the last 20 years were considered.

Results: Studies demonstrate a greater cooccurrence between these 2 disorders than between BPD and other Axis I disorders or between BD and other Axis II disorders. Some authors suggest that many patients diagnosed with BPD are better described as having BD, that the bipolar classification is too narrow, or that BPD should be considered a variant of affective disorders. Others present evidence supporting BPD as a valid construct. Hypotheses about the relation between the 2 disorders and suggestions for clinical practice are offered.

Conclusions: There appears to be sufficient evidence to consider BPD to be a valid diagnosis. Both disorders apply to heterogeneous populations, and their characteristics require further clarification. In diagnostically challenging situations, careful consideration of a patient’s longitudinal history is essential. Future research will be important to ensure that our diagnostic classifications reflect clinically useful entities.

Clinical Implications

* Differentiating borderline personality disorder from bipolar disorder in a clinical situation can be challenging.

* The 2 disorders have been found to cooccur, and the boundary between them requires further clarification.

* To avoid misdiagnosing patients presenting with both affective instability and impulsivity, a detailed longitudinal history is essential.


* Although an attempt was made to conduct a thorough review, there may be relevant publications that have not been summarized in this paper.

The distinction between personality and mood disorders is a controversial issue in psychiatry (1). Defining the boundary between borderline personality disorder (BPD) and bipolar disorder (BD) has been particularly controversial, given the extent to which the symptoms of these 2 disorders overlap. I patients with a history of both affective instability and impulsivity, it is often difficult to determine whether these symptoms occur within the context of discrete episodes, whether they represent a long-standing pattern of functioning, or whether they are a combination of both. Patients with relationship instability and persistent maladaptive behaviours who also have intermittent affective symptoms can be especially difficult to classify (2,3). Making a diagnosis can be particularly problematic in cross-sectional assessments, because patients with either disorder can present with similar symptomatology at a given time.

Both BPD and BD are evolving concepts. For example, when the diagnosis of schizotypal personality disorder was introduced in the DSM-III (4), the definition of BPD was narrowed after the psychotic element was largely removed from the category. However, although the category has narrowed, this diagnosis continues to apply to a heterogeneous population (4). Conversely, in the case of BD, the trend has been toward broadening the diagnosis to include a bipolar spectrum that extends beyond BD I and BD II (5,6). As a result, this diagnosis also applies to a heterogeneous population.

Compared with other diagnostic dilemmas in psychiatry, the relation between BPD and BD is particularly interesting, because it affects our understanding of the boundary between Axis I and Axis II disorders (2). The DSM-III drew a boundary between mood disorders on Axis I and personality disorders on Axis II, ensuring that attention is given to patients’ long-standing personality traits (2,7). However, many DSM Axis II categories have changed with subsequent revisions (7). It is important to continue to question the accuracy of diagnostic categories and to refine these so that they are clinically useful.

This paper describes the prevalence and comorbidity of BPD and BD. Current perspectives on the bipolar spectrum are then presented, followed by evidence for BPD as a valid construct. Perspectives on the overlap between the 2 disorders and implications for clinical practice are discussed.

Prevalence and Comorbidity

According to the results of the Epidemiologic Catchment Area study, mania and hypomania occur in 1.2% of the general population over a lifetime (6). When the diagnostic criteria are expanded to include a broader bipolar spectrum, studies have found prevalence rates ranging from 3.0% to 8.8% in the general population (6,8). Owing to the broadened diagnosis, the condition is being identified with increasing frequency in clinical settings (9). BPD is thought to occur in 2% to 3% of the general population; it is the most common personality disorder encountered in clinical settings (7).

Before considering the difficulty of differentiating BPD from BD, it is important to consider the extent to which these disorders occur together. Several studies have investigated the occurrence of BD in patients with BPD. Deltito and others report that, in their study, 5/16 patients with BPD had a history of a spontaneous manic or hypomanic episode (10). Akiskal and others found that affective disorders were the most common comorbid Axis I disorders in 100 outpatients with BPD, with 7 patients having cyclothymia and 17 having BD II (11). During their study’s 6- to 36-month follow-up period, significantly more patients in the BPD group developed mania or hypomania, compared with those in the control group not suffering from an affective personality disorder. Zanarini and others found that significantly more patients with BPD had comorbid BD II, compared with patients diagnosed with other personality disorders (12). Similarly, Levitt and others found that cyclothymia occurred significantly more frequently in patients with BPD than in patients with other personality disorders (13).

Several studies have also found that BPD is one of the personality disorders occurring most frequently in patients with BD (14–18). Brieger and others report on the results of their comorbidity study pooled with 6 other similar studies: overall, 14.5% of patients with BD were also found to meet criteria for BPD, with a range across the 7 studies of 7% to 30% (15). Üçok and others found that significantly more BD I patients (10%) met criteria for BPD, compared with a control group of subjects having no history of psychiatric disorder (1.7%) (19). Benazzi found that, among patients presenting for treatment of a major depressive episode (MDE), a significantly higher prevalence of BPD existed in BD II patients, compared with BD I patients (20).

Although some studies have found that other personality disorders are more frequently diagnosed in patients with BD (21–23), most suggest that BPD and BD occur together more frequently than BPD with other Axis I disorders or BD with other Axis II disorders. However, this does not further our understanding of the underlying reasons: it is not clear whether we are seeing true comorbidity or, rather, variations of a single disorder.
BPD Within the Bipolar Spectrum

The concept of BD is currently evolving. There have been suggestions that current diagnostic criteria are too narrow, leading to misdiagnosis of certain patients. According to Akiskal, the bipolar spectrum includes not only mania and hypomania but also more subtle forms of hypomania that can occur as part of either a cyclothymic or hyperthymic temperament (5). Although temperaments are enduring patterns, Akiskal suggests that cyclothymic or hyperthymic temperaments fall within the larger scope of a “soft bipolar spectrum” (5). A person with a cyclothymic temperament could present with a history of biphasic affective disturbances starting in early teenage or adult years and characterized by abrupt shifts between phases, with each phase lasting only a few days and with infrequent euthymic periods (5). This description is also characteristic of the affective instability seen in people considered to have BPD. Akiskal argues that there is a tendency to see character flaws rather than affective instability in patients with these temperaments and that many patients diagnosed with personality disorder would be more appropriately considered to have a bipolar spectrum disorder (5). Similarly, Deltito and others propose that the difference between BPD and BD is a matter of degree, not quality, and that BPD is essentially an ultrarapid-cycling disorder (10).

Hirschfeld also describes a broad concept of the bipolar spectrum (8). He suggests that the spectrum includes mixed states, hyperthymic temperament, MDEs, and depressive mixed states. He describes depressive mixed states as periods of dysphoria, irritability, extreme fatigue, racing thoughts, anxiety, panic attacks, and suicidal obsessions and impulses. Again, a patient presenting with this combination of symptoms could likely be diagnosed with BPD if the symptoms were frequent and long-standing.

Ghaemi and Goodwin have proposed that an additional category—bipolar spectrum disorder—should be created to accommodate the broadening of the BD diagnosis (6). They suggest that this category should include all forms of BD other than BD I and BD II and that the diagnosis should require at least 1 MDE and no spontaneous manic or hypomanic episode. Further, one of several combinations of the following would be required: a family history of BD in a first-degree relative, antidepressant-induced mania or hypomania, hyperthymic personality, recurrent MDEs, brief MDEs, atypical depressive symptoms, psychotic MDEs, early onset of MDE, postpartum depressions, antidepressant “wear-off,” and a lack of response to at least 3 antidepressant treatment trials (6). These criteria have yet to be validated, but it seems that certain patients diagnosed with BPD could meet sufficient criteria to be considered as having bipolar spectrum disorder. To illustrate, Deltito and others found that 81% of their subjects showed evidence of bipolarity when they included bipolar temperaments, bipolar pharmacologic response, and bipolar family history as criteria for BD in their study of patients with BPD (10). Akiskal and others also demonstrated some of these features, finding a high rate of affective disorders, pharmacologically induced hypomania, and familial BD in a BPD group, compared with a control group with other personality disorders (11).

In addition to the overlap of symptoms already described, several other aspects of bipolar illness have been suggested to explain the possible misdiagnosis of BD patients as having BPD. First, prior to meeting criteria for BD, for months or years patients often experience chaotic fluctuations of mood and behaviour resembling those observed in patients with BPD (9). Also, the subsyndromal episodes that often occur between full manic or depressive episodes can be “fertile ground for interpersonal conflicts and pathological character developments” that result in personality disorder diagnoses (24). Similarly, the diagnosis of BD II can be complicated when patients are cyclothymic between distinct affective episodes, because they are more unstable than those who have more clear-cut episodes (20). In addition, it has been proposed that ultrarapid-cycling forms of BD exist, in which labile moods with irritable, mixed features characterize a person’s usual state, leading clinicians to interpret these symptoms as characterological rather than as symptoms of an affective disorder (5). Howland and Thase also suggest that problems such as marital discord, promiscuity, poor work performance, and substance abuse may often be attributed to a personality disorder but can also be understood as psychosocial complications of an underlying mood disorder (1).

Thus the relation between the 2 disorders remains controversial. Several authors argue that many patients diagnosed with BPD would be better diagnosed and more effectively treated if they were considered to have a BD. Some authors have suggested that the BPD category does not represent a unique disorder, that it does not appear to represent a definable personality type, and that patients with this diagnosis actually have disorders that lie along the affective spectrum (10,11,25,26).
BPD as a Valid Construct

Given the controversy surrounding the diagnosis of BPD, several investigators have conducted studies to look more closely at its validity. Sanislow and others (4) recently tested 2 models on the basis of DSM-IV BPD criteria. Their study population included patients with borderline, schizotypal, avoidant, and obsessive–compulsive personality disorders; comparison groups included patients who met criteria for major depressive disorder but not for personality disorder. They tested the category as a unique construct and then tested a model that included 3 distinct factors: 1) disturbed relatedness, including unstable relationships, identity disturbance, chronic feelings of emptiness, and stress-related paranoid ideation; 2) behavioural dysregulation, including impulsivity in 2 areas and suicidal or self-mutilative behaviour; and 3) affective dysregulation, including affective instability, inappropriate anger, and avoidance of abandonment. Using confirmatory factor analysis, the authors found a good statistical fit with the 1-factor model, supporting the diagnostic construct of BPD. The results also supported the 3-factor model, which showed a significantly better fit to the data.

Other studies have attempted to find features that distinguish BPD from other disorders. In the comorbidity study described earlier, Zanarini and others ran logistic regressions and found that a summary measure, “complex comorbidity,” better discriminated patients with BPD from control subjects than did each of the Axis I disorders studied (12). They also found that this measure had strong positive predictive power for the diagnosis. Their data demonstrated that meeting criteria for both an affective disorder (such as a mood disorder or an anxiety disorder) and an impulse-control disorder (such as a substance use disorder or an eating disorder) over a lifetime has strong sensitivity and specificity for the BPD diagnosis. These findings suggest that, rather than complicating the diagnosis, the multiple Axis I disorders often observed over time can provide support for the diagnosis of BPD.

In a relatively early study, Gaviria and others compared patient records of 13 patients diagnosed with BD plus BPD with records of 75 patients diagnosed with BD only (27). Patients with both disorders had a higher frequency of childhood and adolescent psychopathology, poorer school performance in childhood and adolescence, and more losses or separations from family members during childhood. They also reported their first affective episode at an earlier age and reported significantly more psychotic symptoms during affective episodes than did the BD-only group. Patients with both diagnoses often showed poor compliance and were frequently terminated from the follow-up phase by the clinic staff. In contrast, the patients with BD only had significantly better social functioning between episodes and were more likely to negotiate their own termination. There were no significant between-group differences in the number of previous affective episodes and hospitalizations. The authors suggest that patients with both BD and BPD can be distinguished from those without this additional diagnosis.

Atre-Vaida and Hussain evaluated 10 patients with BPD and 13 patients with BD to determine whether the 2 diagnoses are distinct or whether they lie on a continuum of the same disorder (25). None of the BD patients had a diagnosis of personality disorder, and all patients were asymptomatic at the time of the evaluation. The study used Cloninger and others’ model of personality (28). According to this model, character dimensions (such as self-directedness and cooperativeness) can discriminate between normal patients and those with a personality disorder, and temperament dimensions (such as novelty seeking, harm avoidance, reward dependence, and persistence) can discriminate among various Axis II categories. BD patients have previously been shown to demonstrate normal scores on all personality dimensions (25). The authors proposed that, if BPD is a variant of BD, patients diagnosed with BPD would have scores similar to patients with BD. The results showed that the temperament and character scores for BD patients did not differ from normal scores. However, the BPD group scored significantly higher on harm avoidance, impulsivity, and disorderliness and lower on self-directedness, cooperativeness, and persistence. The ability to distinguish BPD patients and BD patients according to deviations in temperament and character supports the notion that BPD is not simply a variant of BD (25).

These studies help to define BPD as a unique construct. In addition to using statistical methods to show that the disorder is distinct, several studies have demonstrated significant differences between the characteristics of patients with BPD and the characteristics of those without this disorder. The boundaries have to be clarified, but it appears that sufficient evidence supports keeping the diagnosis of BPD in our classification system.
The Relation Between the Disorders

Several hypotheses have been proposed to explain the relation between BPD and BD. For example, it has been suggested that the affective disorder could be the primary problem and that the personality disorder could result from developmental factors associated with the Axis I disorder (2,29). Alternatively, it could be that personality disorders contribute to the development of BD by unmasking bipolar symptoms (2,29). Third, affective disorders and BPD could be unrelated, with each having a relatively high incidence in the population (29). Finally, both BPD and affective disorders may be diagnosed with signs and symptoms arising from many sources, resulting in the cooccurence of affective and borderline symptoms (29).

Based on their finding of an earlier onset-age of affective illness in patients with both BD and BPD, Gaviria and others suggest that BPD may make patients with a genetic predisposition to affective disorders more vulnerable to earlier development of affective symptoms (27). Similarly, individuals may start with a biological vulnerability that increases their risk of being psychologically impaired in early development, and that this may create vulnerability to either or both disorders, with the actual presentation varying depending on later reactions to environment and temperament (29).

Gunderson and Phillips suggest that the evidence best supports the hypothesis that BPD and affective disorders coexist but are otherwise unrelated (30). They base their conclusions in part on the differences observed between depressive episodes in patients with BPD, which are generally experienced as more empty and lonely and appear to be more developmentally and interpersonally based, and depressive episodes in patients with affective disorders. They also point to family history data, which reveal patterns of familial transmission that are distinct for each disorder. In addition, they suggest that, since depression in patients with BPD is less responsive to antidepressants, there may be a different neurophysiologic substrate for these depressed states, compared with those of patients with affective disorders. Gunderson and Phillips also propose that, despite the independent occurrence of both disorders, a spectrum of affective disorders exists, ranging from the more episodic disorders on Axis I to the enduring character types identified on Axis II.

Based on their study of the 2 models of BPD, Sanislow and others propose that the 3-factor model comprising disturbed relatedness, behavioural dysregulation, and affective dysregulation may help to clarify the boundaries between BPD and other diagnoses (4). The affective dysregulation component, for example, may have elements in common with affective disorders, which may help to explain the overlap between BPD and affective disorders (4).
Clinical Considerations

Given that the boundary between the 2 disorders is not well defined, there will continue to be challenging clinical situations in which symptoms are difficult to categorize. Nevertheless, an accurate diagnosis is important if the patient is to receive appropriate treatment. Further, several studies have indicated that an additional diagnosis of a personality disorder in patients with BD leads to poorer outcomes, including poorer medication compliance, more days in hospital, lower rates of recovery, more severe mood symptoms, lower levels of functioning, and increased incidence of substance use disorders (21,31,32). These factors have an impact on the interventions chosen and on their eventual effectiveness.

To assess thoroughly the possibility of BPD in addition to or in the place of BD, it is necessary to review certain elements of a patient’s history in detail (33). This includes examining closely the nature of present and past mood episodes. When the mood changes are examined for evidence of mania, hypomania, or depression, it is important to carefully characterize the precipitating events, onset, duration, and quality of the mood changes (34). Patients with BPD often have very transient and rapid changes of affect that range from anger, euphoria, agitation, and irritability to sadness and loneliness; these often occur within hours, with each lasting only for a few hours and rarely for more than a few days (34). It is also important to determine whether mood shifts are precipitated by interpersonal events, such as perceived abandonment (33).

Tyrer and Brittlebank point to clues that could indicate a BD diagnosis in patients with histories of impulsive behaviour, poor interpersonal relationships, and frequent overdoses of drugs (3). They suggest paying close attention to the age of onset and type of depression that such patients experience—looking closely, for example, for evidence of adolescent-onset and recurrent depressive episodes characterized by marked psychomotor retardation. They also emphasize the importance of assessing for past hypomanic episodes that may not have received attention (3).

In addition to carefully examining mood episodes, it is essential to obtain a detailed family history, developmental course, background history, relationship history, and history of response to treatment (33). When obtaining the patient’s history, it is important to consider the context and timing of the evaluation. Time-limited, cross-sectional assessments can be problematic for these patients. The presence of an acute Axis I condition can lead to symptoms and behaviours that may suggest a personality disorder when none exists; alternatively, it can mask an underlying personality disorder, impeding such a diagnosis (35,12). Studies have found that euthymic patients with BD assessed by structured interviews show lower rates of comorbid personality disorders than those who are symptomatic (21,24,36). When possible, consideration should be given to delay the Axis II assessment until the Axis I episode has subsided.

The relation between BPD and BD has been controversial ever since the BPD category was included in the DSM-III. The overlap of symptoms between the 2 disorders has prompted discussion and investigation of the boundary between the 2 disorders and has led to controversy over the validity of the BPD diagnosis. There appears to be sufficient evidence to consider BPD a unique entity, distinct from BD. However, both categories include heterogeneous subgroups whose characteristics remain to be clarified (30).

Given the similarity between the diagnostic criteria for both disorders, there will continue to be diagnostically challenging clinical situations. To offer a patient the most accurate diagnosis, it is important to carefully assess several aspects of the patient’s illness history, as well as the patient’s personal and family history. These can provide clues about the nature and course of the patient’s illness and can uncover long-standing patterns. It is also important to consider, given the high degree of comorbidity found in several studies, that a clinical picture which seems difficult to classify may represent the co- occurence of both disorders.

The evolving classification of these disorders, especially the recent broadening of the concept of BD to the bipolar spectrum, creates further challenges for diagnosis and research. Different criteria may be used by different investigators or clinicians to determine a diagnosis, even though the official DSM classification has not changed. It remains to be seen whether these recent conceptualizations of bipolar spectrum disorders will be reflected in future editions of the DSM. It will be important to continue research in this area to ensure that the diagnostic classifications reflect clinically useful entities.