The temporal relationship between anxiety disorders and (hypo)mania: a retrospective examination of 63 panic, social phobic and obsessive–compulsive patients with comorbid bipolar disorder

Abstract

Background: The relationship between anxiety and depressive disorders has been conventionally limited to unipolar depression. Recent studies from both clinical and epidemiologic samples have revealed intriguing associations between anxiety and bipolar (mainly bipolar II) disorders. The present report examines the temporal sequence of hypomania to panic (PD), obsessive–compulsive (OCD) and social phobic (SP) disorders. Methods: Specialty-trained clinicians retrospectively evaluated the foregoing relationships in 63 patients meeting the DSM-III-R diagnosis for PD, OCD and SP with lifetime comorbidity with bipolar disorders (87% bipolar II). Structured interviews were used. Results: In nearly all cases, SP chronologically preceded hypomanic episodes and disappeared when the latter episodes supervened. By contrast, PD and OCD symptomatology, even when preceding hypomanic episodes, often persisted during such episodes; more provocatively, nearly a third of all onsets of panic attacks were during hypomania. Limitations: Assessing temporal relationships between hypomania and specific anxiety disorders on a retrospective basis is, at best, of unknown reliability. The related difficulty of ascertaining the extent to which past antidepressant treatment of anxiety disorders could explain the anxiety-bipolar II comorbidity represents another major limitation. Conclusions: Different temporal relationships characterized the occurrence of hypomania in individual anxiety disorder subtypes. Some anxiety disorders (notably SP, and to some extent OCD) seem to lie on a broad affective continuum of inhibitory restraint vs. disinhibited hypomania. By contrast, and more tentatively, PD in the context of bipolar disorder, might be a reflection of a dysphoric manic or mixed hypomanic symptomatology. The foregoing suggestions do not even begin to exhaust the realm of possibilities. The pattern of complex relationships among these disorders would certainly require better designed prospective observations.

Introduction

Frequent comorbidity among panic agoraphobic (PD), social phobic (SP) and obsessive compulsive disorders (OCD) and depression has been widely reported in clinical (Marks and Lader, 1973, Boyd et al., 1984, Maser and Cloninger, 1990) and epidemiological studies (Robins and Regier, 1991, Lépine et al., 1993, Kessler et al., 1994). However, much of this research has been essentially limited to the co-occurrence of anxiety disorders and unipolar depression. This limited viewpoint is being increasingly challenged. We submit that the available evidence strongly suggests that comorbidity between anxiety and mood disorders — conventionally conceived as the relationship between anxiety and depression — extends into the domain of bipolar spectrum disorders (Perugi et al., 1999).

The occurrence of bipolar features in patients with anxiety disorders is counter-intuitive, but evidence for such a relationship has come from both epidemiological and clinical studies. In the National Comorbidity Survey (NCS), the reported risk of comorbid PD and SP is higher in bipolar disorder (Odds Ratios respectively of 11.0 vs. 4.6) compared to unipolar disorder (Odds Ratios respectively of 7.0 vs. 3.6). Chen and Dilsaver (1995), analyzing the ECA database, reported lifetime rates of OCD amongst probands with bipolar and unipolar disorder of 21.0 and 12.2%, respectively. More recently, in subjects meeting DSM-IV hypomania, Angst (1998) reported elevated rates of comorbidity with PD (12.2%), SP (5.1%) and OCD (9.8%) over population controls.

Concerning clinical studies, a history of manic or hypomanic episodes was observed in 19 (13.5%) of 140 patients with PD-Agoraphobia (Savino et al., 1993), and intra-episode PD was observed in six (23.1%) of 26 patients with pure mania and 12 (80%) of 15 with depressive mania (Shoaib and Dilsaver, 1995). In a clinical sample of 57 patients with SP, Van Ameringen et al. (1991) reported a comorbidity rate of 70.1% with major depression, and 3.5% with bipolar disorder. Major depression is also considered the most common comorbid condition in clinical samples of OCD patients, ranging from 13 to 75% (Rosemberg, 1968, Goodwin et al., 1968, Rasmussen and Tsuang, 1986). Less attention has been devoted to the comorbidity between OCD and bipolar disorder. The development of mania or hypomania in OCD patients in response to treatment with serotonin reuptake inhibitors has primarily been described in case reports (Van Scheyen and Van Kammen, 1979, White et al., 1986, Gordon and Rasmussen, 1986, Rhimer et al., 1996). A systematic evaluation was conducted by Kruger et al. (1995) in a sample of 149 in-patients with DSM-III major affective disorder; similar rates of OCD were observed in those with a unipolar (35.2%) or a bipolar (35.1%) course. In a clinical study conducted by our group on 345 outpatients with OCD (Perugi et al., 1997), lifetime comorbidity with bipolar disorder (primarily with bipolar II disorder) was 16%. More recently, we have investigated lifetime comorbidity between PD, SP and OCD on the one hand, and mood disorder on the other, utilizing the DSM-III-R criteria (Perugi et al., 1998). Major depression was the most common comorbid disorder in all groups; however, the rate of comorbid major depression was significantly higher in the SP (52.1%) and OCD (38%) compared with the PD (29.4%) group. Even bipolar II disorder was more frequently associated with SP (21.1%) and OCD (17.7%) than with PD (5.0%).

Concerning temporal relationship between anxiety and mood disorders, available data are currently limited only to major depression and there does not seem to be a consistent pattern. In some studies, depression was found to be more likely to precede anxiety disorders (Breier et al., 1985, De Ruiter et al., 1989, Sanderson et al., 1990); while other investigators reported the opposite tendency (Thompson et al., 1989). The latter is in accordance with the hypothesis that depression, at least in some cases, may be ‘secondary’ to anxiety disorders. However, the complexity of the comorbidity patterns — which seems to extend beyond unipolarity into the domain of bipolar spectrum disorders — does not permit a unified model of understanding anxiety-mood disorder comorbidity phenomena. In particular the putative link of anxiety disorder to bipolarity presents greater challenges to both theory and practice.

The aim of the present report is to explore the temporal relationship of (hypo)mania with anxiety states in patients with PD, SP, OCD and comorbid bipolar disorder. In particular, we focus on the temporal relationship of PD, SP and OCD relative to that of bipolar disorder. Finally, we examine whether the occurrence of (hypo)mania coincides with the symptomatic phases of PD, SP and OCD or follows their remission.

The sample comprised 63 outpatients enrolled consecutively from 1995 to 1997 at the Psychiatric Institute, University of Pisa. The study is part of the Pisa-San Diego collaborative project on affective and related disorders. The subjects came from a variety of sources, almost equally divided between self-referrals, those from general practitioners, various medical specialists and psychiatrists and, finally, those referred by former patients. All patients gave their informed consent for their participation in the present research, which had been approved by the ethical board of the University of Pisa.

The inclusion criterion for participation was fulfilment of the DSM-III-R criteria for PD (n=21), OCD (n=23) and SP (n=19), as a principal diagnosis and comorbid bipolar disorder type I (n=8, 12.7%) and (atypical) type II (n=55, 87.3%). In the present study the term ‘comorbidity’ is used to refer to the lifetime presence of two or more disorders in the same individual. A principal disorder is defined as the one that is judged to interfere the most with the patient’s overall functioning and/or to have prompted the patient to seek treatment. In order to assure group assignment on the basis of current diagnoses, patients showing current (past 6 months) comorbidity with other anxiety disorders were not included in the study, except for specific phobia and generalized anxiety disorder. Patients were also excluded if they had a lifetime history of schizophrenia or related psychotic disorders, organic mental syndrome, or uncontrolled or serious medical conditions.

A systematic face-to-face interview that consisted of structured and semi-structured components was used to collect data. Diagnostic evaluation and comorbidity with mood and other anxiety disorders were carried out by means of the SCID Up-R, the Structured Clinical Interview for DSM-III-R (Spitzer and Williams, 1988) developed to diagnose anxiety disorders (except for post-traumatic stress disorder), mood disorders, substance abuse and past or present psychotic disorders. Demographic and illness course characteristics were obtained by means of specific interviews for PD (Cassano et al., 1994, Toni et al., 1996, Perugi et al., 1998), SP (Perugi et al., 1995) and OCD (Lensi et al., 1996).

The criteria for major depressive disorder and (atypical) bipolar II are based on DSM-III-R; as in previous studies (Cassano et al., 1992), we used the Semistructured Interview for Depression (Cassano et al., 1989). This last instrument, which has shown its utility for clinical investigation in over 2000 patients at this writing, systematically collects anamnestic and clinical information, explores the presence of DSM-III-R criteria for major depressive episode and mania; history of previous hypomanic episodes; temperamental characteristics (hyperthymic or depressive) and first degree family history for mood and anxiety disorders, schizophrenia, as well as for drug and alcohol abuse.

The relationship between anxiety and mood disorders was explored by means of the National institute of Mental Health (NIMH) retrospective life-chart method (Leverich and Post, 1997), an innovative clinical tool for delineating the longitudinal course of the affective and related disorders.

‘Onset’ is defined as the moment when diagnostic criteria for affective episodes and each anxiety disorder were met for the first time. Retrospective evaluation of severity of mood dysregulation is based on the degree of functional impairment that arises from mild, moderate or severe depression and hypomania. This evaluation is by month and can be corroborated by medical records, physician notes, family members, friends, diaries, school records, work history and any other possible sources. With respect to the individual anxiety disorders under consideration, ‘remission’ was defined as no longer meeting their respective diagnostic criteria. As reliability and validity of this retrospective life-chart method are not fully defined, its use to assess the temporal relationships of (hypo)manic symptoms and PD, SP and OCD has to be considered exploratory. However, given the comparative nature of our study, possible bias in retrospective evaluations is likely to affect similarly all groups under examination; as a result, any such biases are unlikely to significantly impact our findings.

The clinical interviews and all ratings were completed by psychiatrists with at least 4 years of experience in anxiety and mood disorders. Each interviewer underwent a training program in the use of the interview instruments, which included direct observation of experienced interviewers, direct supervision of interviews and interrater reliability. High reliability and diagnostic concordance (Kappa from 0.82 to 0.97) have been documented in previous reports from our center (Benedetti et al., 1997, Perugi et al., 1999). Credible assessment of familial, developmental and psychopathological antecedents at the Pisa Center is considerably aided by the fact that access to past records is readily made available by patients and their physicians and, especially, by the willingness of many family members to corroborate personal history data obtained from patients. To address the limitation common to all retrospective studies, clinical data were reviewed by the interviewer team for the purpose of consensus. When questions arose, patients were recontacted for further clarification. Patients’ medical records were reviewed and information was obtained from family members and previous physicians.

Comparative analyses included ANOVA followed by the Scheffé F-test for continuous data and χ² for categorical data. Conservatively we used two-tail statistics. Because of the number of comparisons performed and the likelihood that some would be significant by chance alone, we chose a significance level of P<0.01.

The mean age did not differ significantly among PD, SP and OCD patients (respectively 35.1±10.2 vs. 29.5±7.9 vs. 28.7±12.7, F=2.32, P=ns). The gender distribution was similar in PD (female=14, 66.7%) and OCD (female=15, 65.2%), with lower percentage of female (n=7, 36.8) in SP group; however, this difference was not statistically significant (χ²=4.59, df=2, P=ns). The mean age at onset of the anxiety disorder was significantly different in the three groups, SP (mean±S.D.=12.2±6.0) had earlier onset compared with OCD (mean±S.D.=22.7±12.6) and PD (mean±S.D.=25.0±8.7) (F=8.93, P=0.0004, Scheffe F-test=SP<OCD, PD). Mean±S.D. age at onset of bipolar disorder was 23.9±9.1, similar in the three groups (SP=21.3±6.5, OCD=24.6±12.8, PD=23.8±14.1, F=1.91, P=ns). The same was true for mean±S.D. age at onset of (hypo)mania (mean=24.9±10.3) (SP=22.1±6.7, OCD=25.7±11.8, PD=25.9±14.1, F=1.31, P=ns).

Examining the temporal relationship among (hypo)mania and anxiety disorders (Fig. 1), significant differences were observed (χ²=21.18, df=4, P=0.0003). PD, SP and OCD preceded the first ascertainable (hypo)manic episode respectively in seven (33.3%), 18 (94.7%) and 12 (52.2%) cases. The onset of PD, SP and OCD followed the onset of hypomania respectively in eight (38.1%), one (5.3%) and 10 (43.5%) cases. Interestingly, a concurrent onset of the anxiety disorder during (hypo)manic episode was present in six (28.6%) cases of PD, but in only one of the OCD (4.3%) patients, and in none of the SP subjects.

As shown in Fig. 2, a complete remission of the anxiety disorder during (hypo)manic episode was recorded in 18 (94.7%) SP, in 9 (42.9%) PD and 11 (47.8) OCD patients (χ²=13.58, df=2, P=0.001).

Age, age at onset and sex distribution of PD, SP and OCD found in this study are in the range of those reported in other clinical studies (Aimes et al., 1983, Uhde and Nemiah, 1989, Lensi et al., 1996). SP showed an earlier age at onset and a different sex distribution and also tended to present a different pattern of association with (hypo)mania. Thus, SP preceded the onset of bipolar disorder and showed a complete remission during (hypo)mania in almost all cases. The only patient that reported the persistence of some performance anxiety suffered from a circumscribed form of SP, characterized by fear and avoidance of writing in public. Most of the other patients presented early onset disabling generalized subtypes of SP and avoidant personality disorder. These findings are consistent with the suggestion that different subtypes of social anxiety have a differential pattern of association with mood disorders.

Concerning PD and OCD, mean age at onset and pattern of remissions during (hypo)mania appeared similar. It is remarkable that PD and OCD symptomatology tend to persist in more than 40% of cases even during (hypo)manic periods. Prospective studies are necessary to confirm these findings, and further clarify which panic-agoraphobic and obsessive–compulsive subtypes are more likely to persist during (hypo)manic phases. Interestingly, PD turned out to be the only anxiety disorder that can begin during (hypo)mania in a significant proportion of cases (28.6%). This observation is consistent with previous reports of high frequency of intra-episode PD during mania and mixed mania (Shoaib and Dilsaver, 1995).

The foregoing findings and considerations fly against a common perception among both mood and anxiety disorder specialists that the relationship between these disorders is largely limited to unipolar depression and dysthymia (Dunner, 1980, Maser and Cloninger, 1990). It is our contention that the relative neglect in clinical and epidemiological research for the presence of bipolar spectrum disorders in patients with anxiety disorders is due to: (1) the relative underdiagnosis of bipolar II disorders, often misdiagnosed as unipolar or personality disorders (Akiskal, 1996); (2) the lack of utilization of structured interviews for the diagnosis of anxiety disorders in (hypo)manic patients (Angst, 1998), and vice versa (Himmelhoch, 1998). Expert interviewing has been recently validated as a means for improving, often doubling, the reported clinical prevalence of bipolar II disorder (Dunner and Kai Tay, 1993, Hantouche et al., 1998).

Even if comorbidity of bipolar and anxiety disorders reported in this paper was partly due to the fact that a university center specializing in these illnesses is referred the most complex patients in this realm, there remains the challange of how best to treat them in the long term. The lack of information about anxiety-bipolar connections may have a negative impact on treatment choice and management. Anxiety disorders are usually treated with antidepressants that might trigger (hypo)manic or mixed states in patients with bipolar diathesis (Van Scheyen and Van Kammen, 1979, White et al., 1986, Rhimer et al., 1996, Himmelhoch, 1998, Levy et al., 1998). On the other hand, some antimanic agents like classical neuroleptics might worsen anxious, phobic and obsessive–compulsive symptomatology and in some cases, contribute to interepisodic and residual chronicity (Van Putten and Marder, 1987, Argyle, 1990). Syndromic complexity may be made apparent by the contrasting effects of treatment, which may improve one disorder, while worsening another, making the adoption of pharmacological combination mandatory in practice (Savino et al., 1993, Perugi et al., 1997).

The main limitation of the present study — shared with other studies reporting on lifetime comorbidity — is its retrospective design. To elicit criteria for lifetime comorbid disorders and their temporal relathionships depended heavily on patient recall of past history. However, interviewers were trained clinicians who utilized structured diagnostic instruments, and who obtained collateral information from relatives in order to minimize the impact of such memory distortion. Retrospective methodology barred us from estimating the proportion of ‘comorbid’ anxiety disorder-hypomania that could be accounted by antidepressant induction of hypomania. This is certainly a possibility (Himmelhoch, 1998). We submit that such pharmacologic induction typically occurs on a prepared terrain, i.e. bipolar family history (Akiskal et al., 1983). The relationship between anxiety disorders and bipolar II has been reported in epidemologic samples (Angst, 1998) and a primary care setting (Manning et al., 1997), indicating that clinical referral bias is unlikely to be the main explanation for the comorbidity patterns reported herein. Obviously, prospective studies are needed to confirm and extend the temporal relationships of hypomania to the different anxiety disorders uncovered by retrospective examination in the present investigation. Although the present findings, heavily biased by our retrospective procedures, are far from being definitive, we submit on clinical grounds that the validity of the phenomenon of anxiety-bipolar II comorbidity should no longer be in doubt. In at least one prospective study of depressed patients (Akiskal et al., 1995), the presence of phobic anxiety, obsessive–compulsive symptoms and somatic preoccupations of subpanic proportions — co-existing with trait mood lability with cyclothymic features — were predictive of eventual clinically diagnosable hypomania over a 12-year period.

As elaborated elsewhere (Himmelhoch, 1998, Perugi et al., 1999), the relationship of SP (and to some extent that of OCD and PD) is in part explained in a model whereby SP and OCD represent extremes on a broad temperamental dimension of constraint vs. hypomanic disinhibition. By contrast, and more tentatively, many cases of PD in the context of bipolar disorder might reflect mixed (hypo)manic symptomatology (Akiskal, 1990). These are merely some hypothetical possibilities for a prevalent clinical phenomenon that is in search of better explanatory models of comorbidity.