Abstract
There is a well-known association between migraine and affective disorders, but the information is sparse concerning the prevalence of migraine in subgroups of the affective disorders. The present study was undertaken to investigate the prevalence of migraine in unipolar depressive, bipolar I and bipolar II disorders. Patients with major affective disorders (n = 62), consecutively admitted to an open psychiatric ward, were examined with a semi-structured interview based on DSM-IV diagnostic criteria, combined with separate criteria for affective temperaments. Diagnosis of unipolar and bipolar I disorders followed the DSM-IV criteria, while bipolar II disorder encompassed patients with either discrete hypomanic episodes or a cyclothymic temperament. Migraine was diagnosed according to IHS-criteria. Symptoms of migraine were found to be common in these patients, both in those with unipolar depression (46% prevalence of migraine) and in those with bipolar disorders (44% prevalence). Among the bipolar patients there was, however, a striking difference between the two diagnostic subgroups, with a prevalence of 77% in the bipolar II group compared with 14% in the bipolar I group (P = 0.001). These results support the contention that bipolar I and II are biologically separate disorders and point to the possibility of using the association of bipolar II disorder with migraine to study both the pathophysiology and the genetics of this affective disorder.
Introduction
Several epidemiological studies have shown a clear connection between migraine headaches and depression (1–3). In a study from Detroit (1) the lifetime prevalence of major depression was found to be three times higher among patients with migraine compared with patients without migraine (27% vs. 9%). Bipolar I and II disorders also occurred more frequently among the patients with migraine, of whom 5% had a bipolar I disorder (vs. 1% in persons without migraine) and 4% had a bipolar II disorder (vs. 1% in those without migraine). However, there have been few studies examining the prevalence of migraine among patients seeking treatment primarily for psychiatric disorders. Endicott (4) found in a study from private specialist practice in New York that migraine was common among patients with major affective disorders, and occurred with highest frequency in a group with characteristics similar to DSM-IV bipolar II patients. Mahmood et al. (5) found in a questionnaire study of bipolar patients that 26% fulfilled the criteria for a diagnosis of migraine.
Genetic factors are important both in migraine (6) and in the affective disorders (7). If there should be a specific association between migraine and one subtype of affective disorders this clearly would be important, pointing to the possibility of using more homogeneous groups for further genetic studies of the affective disorders. In a similar vein the association of bipolar disorders with panic disorder has been used to define a subgroup with a possible common genetic background (8). More homogenous groups would clearly also be useful in studies of biochemical mechanisms associated with affective dysregulation, for instance alterations in serotonergic systems, which are thought to be involved in the pathogenesis of both migraine (9) and the affective disorders (10).
Drug treatment of migraine and the affective disorders is to a large extent different, but there are drugs having effect on both disorders, the most interesting of which may be valproate (11, 12). A closer study of the association between these disorders could hopefully contribute towards better treatment.
The purpose of the present study has been, on this background, to examine the prevalence of migraine in patients with bipolar and unipolar affective disorders.
Methods
Subjects were consecutively admitted patients to a 12-bed psychiatric ward, which is open 5 days a week. The ward is part of a university hospital in Bergen. Most of the patients have initially been admitted to one of the closed wards of the hospital, and have subsequently been transferred to this open ward, when they are in the process of recovering. Criteria for admission to the ward are the need for further inpatient treatment, combined with the ability to function at home with their families during weekends. Patients were not selected on the basis of psychiatric diagnosis or the presence of headaches. The length of the inpatient treatment is usually 1–3 months. Most of the patients admitted to this ward have affective disorders. Patients were included if they at intake presented with a major affective syndrome (major depression or mania) that was not clearly secondary to an organic or substance abuse disorder, were between 18 and 65 years old, and gave informed consent to participate. Patients were excluded if they did not speak Norwegian with sufficient fluency to be interviewed without an interpreter. The study protocol was approved by the local ethics committee.
Information was collected by direct interview with patients, supplanted with data from the hospital records where appropriate. A semi-structured interview based on DSM-IV criteria (13) for affective disorders (major depressive episode, mania, hypomania) and anxiety disorders was used. Suicide attempt during the present or previous illness episodes and number of depressive episodes were recorded. Cyclothymic affective temperament was diagnosed according to the criteria of Akiskal and Akiskal (14) and hyperthymic temperament according to Akiskal and Mallya (15). Criteria for the cyclothymic temperament requires at least three of five attributes of each of the following two sets, with an indeterminate early onset (< 21 years): first group, (1) hypersomnia vs. decreased need for sleep, (2) introverted self-absorption vs. uninhibited people-seeking, (3) taciturn vs. talkative, (4) unexplained tearfulness vs. buoyant jocularity, and (5) psychomotor inertia vs. restless pursuit of activities; second group, (1) lethargy and somatic discomfort vs. eutonia, (2) dulling of senses vs. keen perceptions, (3) slow-witted vs. sharpened thinking, (4) shaky self-esteem alternating between low self-confidence and overconfidence, and (5) pessimistic brooding vs. optimism and carefree attitudes. The hyperthymic temperament requires at least five of the following characteristics, with an indeterminate early onset (< 21 years): (1) irritable, cheerful, over-optimistic or exuberant, (2) naive, over-confident, self-assured, boastful, bombastic or grandiose, (3) vigorous, full of plans, improvident and rushing off with restless impulse, (4) over-talkative, (5) warm, people-seeking, or extroverted, (6) over-involved and meddlesome, and (7) uninhibited, stimulus-seeking or promiscuous.
Unipolar depressive and bipolar I disorders were diagnosed according to DSM-IV criteria. Bipolar II disorder was in this study defined to include patients fulfilling either DSM-IV criteria for the disorder or the criteria for a cyclothymic or hyperthymic temperament as described above, in addition to one or more major depressive episodes.
The criteria of the International Headache Society (16) were used to establish the diagnosis of migraine. In addition to migraine with and without aura the occurrence of migraine aura without headache was specifically asked for and recorded. The frequency of attacks during the last year were also recorded (at least once a week, at least once a month, less than once a month or no attacks last year).
χ2 test, t-test (two-tailed) or
Results
A total of 62 patients were included in the study. The mean age was 37 ± 8.9 years (range 20–57 years). Forty-four (71%) of the patients were women. The index episode was depression in 55 of the patients (89%) and mania in 7 (11%). Bipolar disorders were found to be almost as common as unipolar depressive disorder (n = 27, 44% vs. n = 35, 56%). Fifty per cent (22/44) of the women and 28% (5/18) of the men had a bipolar disorder. There were approximately an equal number of bipolar I (n = 14, 23%) and bipolar II (n = 13, 21%) patients. None of the patients fulfilled the criteria for a hyperthymic temperament, so the bipolar II group consisted of patients with hypomanic episodes (n = 9) or a cyclothymic temperament (n = 4). In addition two of the patients with discrete hypomanic episodes also had a cyclothymic temperament, making a total of six patients with cyclothymic temperament in the bipolar II group.
In Table 1 some of the clinical characteristics, apart from migraine, of the patients with bipolar I, bipolar II and unipolar depressive disorders are presented. Significant differences between the groups are found in the age of onset of the first affective episode, the number of depressive episodes, and the proportion of patients who were holding a job or studying, had alcohol or substance abuse, and any type of anxiety disorder. It is noteworthy that all of the bipolar II patients had a comorbid anxiety disorder, compared with 50% of the bipolar I patients.
Characteristics of the patients with bipolar I (BP I), bipolar II (BP II) and unipolar depressive (UP) disorders
∗
†χ2 test.
Twenty-eight (45%) of the patients had migraine, 48% of the women and 39% of the men. Ten (16%) had migraine without aura, 12 (19%) had migraine with aura and 6 (10%) had aura symptoms without headache. Age of onset of migraine (mean ± SD) was 23 years (± 9.9). The frequency of migraine attacks was significantly higher in patients with bipolar II disorder than in patients with unipolar depressive disorder (P = 0.04, chi-square test): in the bipolar II group 7/10 patients had attacks at least once a week, two had attacks at least once a month and one did not have any migraine attacks during the last year, while in the unipolar depressive group only 3/16 patients had attacks at least once a week, four had attacks at least once a month, six had attacks less than once a month and three did not have any migraine attacks during the last year.
There was a somewhat higher prevalence of anxiety disorders in the patients with migraine (Table 2), especially evident for panic disorder (50% vs. 26% prevalence, P = 0.056). The age of onset of anxiety disorders or major affective episodes among the migraine sufferers was not significantly different from that of the group without migraine (Table 2). Other characteristics of the patients, grouped according to the presence or not of migraine, are presented in Table 2.
Characteristics of the patients with and without migraine
∗ t-test, two-tailed.
†χ2 test. NS = not significant.
Looking at the whole group of bipolar patients there was no difference in the prevalence of migraine compared with the unipolar group (44% vs. 46%). However, when comparing the patients with bipolar I and bipolar II disorders there was a substantial difference: 14% (2/14) of the bipolar I patients and 77% (10/13) of the bipolar II patients had a diagnosis of migraine (P = 0.001, comparing the two bipolar groups, Fig. 1).
Percentage of patients with (▪) and without () migraine in unipolar depressive (UP, n = 35), bipolar II (BP II, n = 13) and bipolar I (BP I, n = 14) disorders (P = 0.001 for the comparison between the two bipolar groups, χ2 test).
In the bipolar II group five of the patients had migraine with aura, one had aura symptoms without headache and four had migraine without aura. All of the six bipolar II patients with a cyclothymic temperament had migraine. In the bipolar I group five patients had a cyclothymic temperament and one of these had migraine.
Discussion
This study has shown that migraine is a common comorbid condition among inpatients with major affective disorders, with a prevalence (45%) clearly in excess of what one would expect from epidemiological data (17, 18). The prevalence of migraine was highest among patients with bipolar II disorder (77%), and this was significantly different from the prevalence among bipolar I patients (14%). The unipolar group had a prevalence in between (46%) the two bipolar groups. This is in broad agreement with the study by Endicott (4), who found a 51% prevalence of migraine in patients with characteristics similar to bipolar II patients as defined in the present study, compared with a 22% prevalence of migraine in bipolar I patients and 27% in strictly unipolar depressive patients. Other studies have examined only patients with bipolar I disorder or have not distinguished between bipolar I and bipolar II disorders. In a large group (n = 327) of patients with bipolar I disorder 27% of the women and 14% of the men had a history of migraine headaches (19). Mahmood et al. (5) found a migraine prevalence of 26% among bipolar patients, but did not report separate figures for bipolar I and bipolar II disorders. Support for the association of migraine and bipolar disorders has also been obtained from a study of migraine sufferers attending a headache clinic, where the prevalence of bipolar spectrum disorders was found to be 8.6% (20).
Bipolar II disorder in this study is defined more broadly than according to DSM-IV criteria. There is substantial evidence supporting the separation of patients with cyclothymic temperament from the unipolar depressive group (21, 22). It might be argued that there are several subgroups in this realm of ‘soft bipolar’ conditions (21), but until more definite evidence is presented it seems reasonable to let the bipolar II group encompass patients with either discrete hypomanic episodes or a cyclothymic temperament. Concerning patients with hyperthymic temperament, it may be more problematic to include these in the bipolar II group without reservation (21, 22), but as stated there were no patients with hyperthymic temperament in the present sample.
Bipolar II disorder often is associated with different comorbid conditions, especially substance abuse and character pathology. It has been argued that when patients are selected that are free from these conditions the characteristics of bipolar II patients resemble those with bipolar I disorder (23). In the present study the bipolar II patients do have an excess of alcohol and substance abuse compared with bipolar I patients, and they also more frequently have anxiety disorders, but gender, age, age of onset of first major affective episode, marital and job status, presence of psychotic symptoms during depressive episodes, current or previous suicide attempt and number of depressive episodes, are not significantly different. The strong connection between bipolar II disorder and migraine supports the contention that bipolar I and II disorders represent two different nosological conditions (24) and indicates that a tendency to migraine headaches may be an integral part of the biological underpinning of bipolar II disorder. However, the study does not give any obvious clue as to what feature of the bipolar II syndrome cluster might explain the association with migraine. One might conjecture that excessive mood lability, as seen in the patients with a cyclothymic temperament, represents the link to migraine. This is supported by the fact that all of the bipolar II patients with this temperament had migraine, but on the other hand only one of five patients with bipolar I disorder and cyclothymic temperament had migraine, making it difficult to uphold this hypothesis.
The present findings underscore the importance of including questions concerning migraine when taking the history of patients with affective disorders. Decisions concerning pharmacotherapy for a patient with a major affective disorder should take into consideration such a common comorbid condition as migraine, which is often not reported spontaneously by the patient because he or she does not think this is relevant during a psychiatric consultation.
The high frequency of migraine among the men in this study (39%, compared with 48% among the women) is different from what is usually found in epidemiological studies. Rasmussen and Olesen (17) found in an investigation in Denmark a male : female ratio of 1 : 2 for migraine with aura and 1 : 7 for migraine without aura. However, the present findings are similar to those of Mahmood et al. (5), who found an equal proportion of male (25%) and female (27%) migraine sufferers in their sample of bipolar patients, in contrast to the findings of Blehar et al. (19) described above. The present study thus indicates that major affective disorders in males and females may be associated with an approximately equal frequency of migraine.
In the present study there is furthermore a comparatively greater number of patients having migraine with aura compared with migraine without aura than usually found in epidemiological studies (12 vs. 10), and this is even more pronounced if the patients having aura symptoms without headache are included in the former group (18 vs. 10). Silberstein and Lipton (18) reported that in a large sample of 1400 migraineurs 27% of the men and 28% of the females had migraine with aura. In contrast to the present findings, Endicott, in his study from 1989 (4), found a higher prevalence of migraine without aura (26%) than migraine with aura (17%) in patients with affective disorders. However, a high frequency of migraine with aura among patients with a combination of migraine and affective disorders is to be expected from findings in epidemiological studies, where the association of migraine and major depressive disorder is especially strong for migraine with aura (1).
Mahmood et al. (5) found that bipolar patients with migraine had an earlier age of onset of their psychiatric disorder than those without migraine and suggested that the presence of migraine might define a subgroup with a more serious affective disorder. In this study the onset of an anxiety disorder did occur earlier in the migraine group than in the non-migraine group, although the difference is not statistically significant. The lack of statistical significance may be the result of the small sample sizes. The mean age of onset of the first anxiety disorder (14 years) for patients with migraine was earlier than the onset of migraine (23 years), which again occurred earlier than the first major affective episode (26 years). This is in agreement with studies using samples from the general population (11).
The limitations of the present study are a small sample size and the difficulty of knowing to what extent the patients are similar to a community sample of patients with major affective disorders. It would also have been desirable to make a comparable diagnostic assessment with patients presenting migraine as their primary complaint. The diagnostic assessment in this study is based on direct interview with patients, combined with information contained in the hospital journals. Relatives have not been systematically interviewed. More detailed information from relatives could perhaps have given a better separation of the bipolar II group from the unipolar group. Patients were interviewed in a recovery phase and were at that time without psychotic or serious cognitive symptoms. However, it is possible that interview in a more clearly euthymic phase might have improved the quality of the information obtained, especially concerning the affective temperaments.
The close relation that apparently exists between migraine and bipolar II disorder may contribute to the delineation of the pathophysiology of both migraine and the affective disorders. Serotonergic systems are strongly implicated in the pathophysiology of migraine (9) and there is evidence that there may be a ‘migraine generator’ in the brainstem, perhaps associated with the monoaminergic nuclei located there (25). This is clearly relevant to the bipolar disorders, where serotonergic systems probably are of importance both in the pathophysiology and in the treatment of the disorders (10). Combining findings from biological studies of migraine and bipolar disorders may shed better light on both types of disorders. In the study of migraine pathophysiology a small subgroup (familial hemiplegic migraine with a dominant inheritance) has pointed to mechanisms (defect in calcium channels) that may be relevant to migraine in general (6). Similar mechanisms ought to be considered also in relation to the pathophysiology of bipolar II disorder.
It might be conjectured that the presence or absence of migraine defines more homogenous subgroups of patients with major depression (from a biochemical and genetic point of view) than the presence or absence of hypomania or cyclothymic temperament. Especially in relation to genetic studies of bipolar disorders, where the difficulty of defining ‘caseness’ for unipolar relatives is considerable (26), including questions concerning migraine in study protocols might prove useful.
Footnotes
Acknowledgements
I want to thank Per Bergsholm MD for helpful discussions concerning the diagnosis of affective temperaments, and Professor Kjell Hole for comments during the preparation of the manuscript. I am also grateful for financial support from the legacy of Gerda Meyer Nyquist Gulbrandson & Gerdt Meyer Nyquist.