Journal of Affective Disorders 59 (2000) S5–S30www.elsevier.com/ locate / jad

Review article

Re-evaluating the prevalence of and diagnostic composition withinthe broad clinical spectrum of bipolar disorders

a , b c d*Hagop S. Akiskal , Marc L. Bourgeois , Jules Angst , Robert Post ,e f¨ ¨Hans-Jurgen Moller , Robert Hirschfeld

aInternational Mood Center, University of California at San Diego, La Jolla, CA, USAbUniversity of Bordeaux, France

cZurich University Psychiatric Hospital, SwitzerlanddBiological Psychiatry Branch, National Institute of Mental Health, Bethesda, USA

ePsychiatric Hospital of the University of Munich, GermanyfDepartment of Psychiatry and Behavioral Sciences, University of Texas Medical Branch, Galveston, USA

Received 19 November 1999; accepted 15 February 2000

Abstract

Until recently it was believed that no more than 1% of the general population has bipolar disorder. Emerging transatlanticdata are beginning to provide converging evidence for a higher prevalence of up to at least 5%. Manic states, even those withmood-incongruent features, as well as mixed (dysphoric) mania, are now formally included in both ICD-10 and DSM-IV.Mixed states occur in an average of 40% of bipolar patients over a lifetime; current evidence supports a broader definition ofmixed states consisting of full-blown mania with two or more concomitant depressive symptoms. The largest increase inprevalence rates, however, is accounted for by ‘softer’ clinical expressions of bipolarity situated between the extremes offull-blown bipolar disorder where the person has at least one manic episode (bipolar I) and strictly defined unipolar majordepressive disorder without personal or family history for excited periods. Bipolar II is the prototype for these intermediaryconditions with major depressions and history of spontaneous hypomanic episodes; current evidence indicates that mosthypomanias pursue a recurrent course and that their usual duration is 1–3 days, falling below the arbitrary 4-day cutoffrequired in DSM-IV. Depressions with antidepressant-associated hypomania (sometimes referred to as bipolar III) alsoappear, on the basis of extensive international research neglected by both ICD-10 and DSM-IV, to belong to the clinicalspectrum of bipolar disorders. Broadly defined, the bipolar spectrum in studies conducted during the last decade accounts for30–55% of all major depressions. Rapid-cycling, defined as alternation of depressive and excited (at least four per year),more often arise from a bipolar II than a bipolar I baseline; such cycling does not in the main appear to be a distinct clinicalsubtype — but rather a transient complication in 20% in the long-term course of bipolar disorder. Major depressionssuperimposed on cyclothymic oscillations represent a more severe variant of bipolar II, often mistaken for borderline or otherpersonality disorders in the dramatic cluster. Moreover, atypical depressive features with reversed vegetative signs, anxietystates, as well as alcohol and substance abuse comorbidity, is common in these and other bipolar patients. The properrecognition of the entire clinical spectrum of bipolarity behind such ‘masks’ has important implications for psychiatric

*Corresponding author.VA Psychiatry Service (116A), 3550 La Jolla Village Drive, San Diego, CA 92161, USA. Tel.: 1 1-619-552-8585ext. 2226; fax: 1 1-619-534-8598.

E-mail address: hakiskal@ucsd.edu (H.S. Akiskal).

0165-0327/00/$ – see front matter 2000 Elsevier Science B.V. All rights reserved.PI I : S0165-0327( 00 )00203-2

S6 H.S. Akiskal et al. / Journal of Affective Disorders 59 (2000) S5 –S30

research and practice. Conditions which require further investigation include: (1) major depressive episodes wherehyperthymic traits — lifelong hypomanic features without discrete hypomanic episodes — dominate the intermorbid orpremorbid phases; and (2) depressive mixed states consisting of few hypomanic symptoms (i.e., racing thoughts, sexualarousal) during full-blown major depressive episodes — included in Kraepelin’s schema of mixed states, but excluded byDSM-IV. These do not exhaust all potential diagnostic entities for possible inclusion in the clinical spectrum of bipolardisorders: the present review did not consider cyclic, seasonal, irritable-dysphoric or otherwise impulse-ridden, intermittentlyexplosive or agitated psychiatric conditions for which the bipolar connection is less established. The concept of bipolarspectrum as used herein denotes overlapping clinical expressions, without necessarily implying underlying genetichomogeneity. In the course of the illness of the same patient, one often observes the varied manifestations described above— whether they be formal diagnostic categories or those which have remained outside the official nosology. Some form oflife charting of illness with colored graphic representation of episodes, stressors, and treatments received can be used todocument the uniquely varied course characteristic of each patient, thereby greatly enhancing clinical evaluation. 2000Elsevier Science B.V. All rights reserved.

Keywords: Bipolar spectrum; Epidemiology; Bipolar I; Bipolar II; Bipolar III; Antidepressant-induced hypomania; Rapid-cyclings; Mixedstate; Life charting

1. Introduction (Akiskal, 1983, 1996, 1999), and which embraces aspectrum involving schizomanic and manic (Gershon

What today is officially termed ‘bipolar disorder’ et al., 1982; Marneros, 1999), as well as predomi-has been recognized throughout much of this century nantly depressive expressions with hypomania and/as ‘manic-depressive psychosis.’ Nonetheless, or cyclic recurrent course, including rapid-cyclingKraepelin (1921) had included in the latter rubric (Dunner et al., 1976, 1977; Angst, 1998). We willmany recurrent depressive conditions that he consid- document that patients with major depressions andered to belong to manic-depression either on the spontaneous hypomania — as well as those withbasis of lesser degrees of excitement (hypomania), hypomanic or manic episodes occurring during anti-temperamental dispositions of a cyclothymic, irrit- depressant treatment — belong to the bipolar spec-able or manic types, or family history for manic- trum. An important thrust of recent research withindepressive illness. In current classificatory schemas the bipolar spectrum is the greater recognition offormally adopted by the American Psychiatric As- manic and depressive admixtures variously termedsociation (DSM-IV, 1994) and the World Health ‘mixed mania’, ‘depressive mania’, or ‘dysphoricOrganization (WHO, 1992), Kraepelin’s position has mania’. Again we review the need and evidence forbeen compromised in favor of unipolar and/or major less restrictive definitions for these bipolar mixeddepressive disorders. What is retained in the bipolar states (McElroy et al., 1992; Perugi et al., 1997).category conforms to a narrower definition of the Finally, we consider the usefulness of charting theillness in which manic excitement, often of psychotic course of affective illness (Leverich and Post, 1998)proportion, alternates with depression; while hypo- as a way of better documenting — and following themania is recognized, its diagnostic threshold is set progress of — the entire spectrum of affectivetoo high with many exclusionary clauses. As a result, manifestations for a given bipolar patient.publications deriving from instruments or research The recognition of the entire clinical spectrum ofbased on such conservative criteria, have estimated bipolar disorders is of major public health concernbipolarity to account for about 1% of the population, because, despite the increasing availability of newand only for 10–15% of all mood disorders (Regier treatments, under-diagnosis or long delay in diag-et al., 1988; Weissman et al., 1996). nosis, and gross under-treatment continue to plague

This paper challenges these rates for bipolar our field (Lish et al., 1994; Hirschfeld et al., 1997).disorder in light of a broader concept of bipolarity ‘‘Lesser’’ manifestations — often of a subthreshold,which has evolved during the past two decades depressive or labile nature and intermixed with

H.S. Akiskal et al. / Journal of Affective Disorders 59 (2000) S5 –S30 S7

‘‘conduct’’ disturbances — precede the overt mani- which incorporates cyclic depressions, has beenfestations of the illness in the offspring and bio- embraced by the definitive contemporary text onlogical kin of adult bipolars (Akiskal et al., 1985). manic-depressive illness (Goodwin and Jamison,Clinicians must be prepared to embrace a broader 1990).view of bipolarity if such antecedents are to be Originally, Dunner et al. (1976) identified theseappreciated as possible indicators of the disease. less-than-manic patients as bipolar II on the basis ofAlthough a plea, on methodological grounds, has hospitalization for depression, and excited periodsbeen recently made to maintain the ‘‘integrity of the that did not require hospitalization. Fieve and Dun-bipolar disorder concept’’ and to prevent its ‘‘prema- ner (1975) had reserved bipolar I for those bipolarture widening and dilution’’ (Baldessarini, 2000), a patients whose excitements were of such severitygreat deal of sound clinical research reviewed in this that hospitalization had been required. Although onepaper justifies considerable widening beyond the can argue that hospitalization is an artificial criterionconservative positions of DSM-IV and ICD-10. for defining the diagnostic threshold for mania, theWhere the evidence is inconclusive or tenuous, we work of these authors nonetheless represented anhave refrained from endorsement of bipolar status for important advance — for both research methodology‘‘border’’ conditions, which would require more and clinical work — in paving the way for thesystematic data. recognition of the large universe of bipolar patients

whose excited periods remained ambulatory.The ‘soft bipolar spectrum’ (Akiskal and Mallya,

2. Concepts and terminology 1987) is a more inclusive term for bipolar conditionsbeyond classic mania, and which modifies the

Although the connection of melancholia to mania foregoing definitions of bipolar II by incorporatinghad been observed by Greek physicians about 2000 depressions with hypomanic episodes, cyclothymicyears ago, it wasn’t fully documented in clinical and hyperthymic traits, as well as those with familialpractice until the 19th century when French alienists bipolarity; the spectrum also includes hypomanic(Baillarger, 1854; Falret, 1854) and the German periods which occur during pharmacotherapy orpsychiatrist Kraepelin (1921) devoted detailed de- other somatic treatments. Alternative terms used inscriptions to alternating forms of depression and referring to these less-than-manic bipolar conditionsexcitement in their psychotic, as well as milder, with depressive presentation include ‘‘Dm’’ (Angstambulatory forms. The unipolar–bipolar distinction et al., 1980), ‘‘Unipolar-L’’ (Kupfer et al., 1975),(as proposed, among others, by Angst, 1966/1973 and ‘‘pseudo-unipolar’’ depression (Mendels, 1976).and Winokur et al., 1969), gained momentum during The last two designations do highlight the provoca-the last third of the 20th century, with the net result tive possibility that many apparently unipolar pa-of uncoupling depressive disorders from more strict- tients could be related to bipolar disorder on thely defined bipolar disorders. This distinction, which basis of pharmacological response to lithium carbon-has proven to be of great heuristic value for clinical ate (Bowden, 1978).research, left undefined many affective conditions Taylor and Abrams (1980), Akiskal et al. (1983),lying in the interface of unipolar and bipolar dis- and Egeland (1983) were among the first to urge fororders (Winokur, 1980). We learn about these pa- the necessity to return to a broader concept of bipolartients when we examine the pedigrees of bipolar disorder. Klerman (1981) spoke of a spectrum ofprobands and discover many affected individuals manic conditions that extended from classic psychot-with predominantly depressive manifestations (Ger- ic mania through various degrees of hypomania,shon et al., 1982; Tsuang et al., 1985; Akiskal et al., subclassified into six types. A related proposal made1985b). Based on such data and clinical observa- by Endicott (1989) extended the spectrum to includetions, Akiskal and Mallya (1987) estimated that cyclic depressions — without clear-cut hypomania4–5% of the general population belongs to a broad — but abrupt onset and offset. In a series of morebipolar spectrum with predominantly depressive formal bipolar spectrum proposals (Akiskal, 1983,phenomenology coupled with less-than-manic excite- 1996; Akiskal and Akiskal, 1988), bipolarity isments. This enlarged concept of bipolar disorder, categorized into type I (mania with or without

S8 H.S. Akiskal et al. / Journal of Affective Disorders 59 (2000) S5 –S30

depression), type II (depression with hypomania and/ the spectrum and mood-incongruent psychosesor cyclothymia), and type III (hypomania associated beyond the boundaries of the classic affective psy-with antidepressants, as well as depressions with choses at the severe end of the spectrum.hyperthymic temperament and/or bipolar family Two recent review papers represent contrastinghistory). Just a few years after the publication of views on the boundaries of bipolar disorder. CassanoDSM-III — and in response to a request by the et al. (1999) declare the bipolar spectrum to be aAmerican Psychiatric Association as to whether the ‘‘clinical reality’’ in search of diagnostic and assess-new diagnostic manual provided adequate coverage ment methodology. The other paper by Baldessarinifor all affective diagnoses — in a consecutive case (2000) argues for a more cautious approach whichseries of personally examined affective states (Akisk- would restrict any broadening of the presently ac-al and Mallya, 1987), the foregoing bipolar spectrum cepted official boundaries; it is curious, however,conditions were shown to be at least as prevalent as that Baldessarini’s tabulation of the historical evolu-their unipolar counterparts (Table 1). tion of the bipolar concept includes largely authors

The spectrum concept of bipolarity has been or investigators whose work supports a broad con-greatly enriched by the epidemiological studies of cept of bipolar disorder. This is because the thrust ofAngst (1998), who demonstrated the high prevalence historical development in bipolar disorder has beenof brief hypomanic episodes below the threshold of 4 for a broad spectrum. Indeed, a great deal of workdays required in such formal classifications as the has been done in validating such a concept, includingDSM-IV. This work, conducted by one of the clinically usable criteria. Where the Baldessarini andoriginal group of researchers who was highly in- Cassano approaches come together is their require-fluential in promoting the unipolar–bipolar dich- ment of methodological purity which may be neces-otomy (Angst, 1966/1973), persuasively argues for sary for certain research operations versus practicethe need to enlarge bipolarity at the severe (psychot- considerations which, we would contend, do notic manic) and the subthreshold (brief hypomania) require pristine methodological designs, nor un-ends of the spectrum. What is remarkable here is that wieldy assessment instruments which are unrealisticthe so-called ‘subthreshold’ manifestations from a in a clinical setting.symptomatological point of view have proven — in To recapitulate, candidates proposed for inclusionassociation with depression — to have significant in a broadly conceived bipolar spectrum areadverse psychosocial consequences. schizobipolar disorder, mania, mixed states, depres-

The important research of Bertelsen et al. (1977) sions with hypomania (irrespective of duration) oris perhaps the most convincing evidence for the pharmacologically mobilized hypomania, as well asbroad concept of bipolarity: Monozygotic twins those in association with cyclothymic and hyper-discordant for strictly defined mood disorders, were thymic temperaments, and finally recurrent (pseudo-broadly concordant for such conditions as mood- unipolar) depressions with bipolar family history orlabile temperaments at the milder (untreated) end of cyclic depressions responsive to lithium (and by

extension to other mood stabilizers). Because there isrelatively little controversy about manic states, much

Table 1 of this paper enlarges upon the lesser-known entitiesaPrimary affective diagnoses in 102 patients in a community within the bipolar spectrum.bmental health centre

Diagnosis %

Bipolar I 18 3. Epidemiological studiesBipolar II 18Bipolar III 9 Factors which influence rates of psychiatric dis-Cyclothymia 5

orders and — particularly those for bipolar spectrumUnipolar 44disorders — are listed in Table 2. Historically,Dysthymia 6

a epidemiological research has neglected the less-than-This sample is based on taking every 50th patient over amanic forms of bipolar disorder. An exception is a10-year period and eliminating those with non-affective diagnoses.

b Based on Akiskal and Mallya (1987). study by Weissman and Myers (1978) which, using

H.S. Akiskal et al. / Journal of Affective Disorders 59 (2000) S5 –S30 S9

Table 2 when conforming to narrow definitions of bipolarMethodologic factors influencing rates of bipolarity disorder as reflected by instruments in which bipolar? Breadth of criteria II is inadequately defined.? Instrument used Relatively new epidemiological data (also listed in? Lay versus clinical interviewers Table 3) have challenged the foregoing figures.? Population studied (e.g., students, community subjects)

These data expand the concept of bipolar disorder to? Sample sizeinclude subthreshold expressions of hypomania, in-? Single versus repeated observations

? Interview of patient versus relatives cluding those with less than 4 days. Thus, the new? Timing of interview rates include mania, hypomania, brief hypomania,

and cyclothymia. The most extensive of these studiesThe Schedule for Affective Disorders and Schizo- were conducted by Angst (1998) in the canton ofphrenia (SADS), reported on bipolar I, II, and Zurich, and yielding rates for bipolar disorder up tocyclothymic personality in New Haven. However, the age of 35 of 5.5% for DSM-IV mania andthe difficulties of case ascertainment have generally hypomania, and a further 2.8% for brief hypomania.led to the subsequent exclusion of the symp- The modal duration for these hypomanias, whichtomatologically milder expressions of bipolarity could be brief recurrent or sporadic, range from one(Weissman et al., 1996). This is exactly the opposite to three days. Lewinsohn et al. (1995), also usingof what has happened in the epidemiology of depres- broader definitions that went beyond the convention-sion, which has focused on the entire severity of al duration thresholds for mania and hypomania,depressive disorders (Angst and Merikangas, 1997; found that 5.7% of adolescents in a community studyJudd et al., 1997). in Oregon met criteria for bipolar disorder. The

It is therefore not surprising that the rates for foregoing studies justify the inclusion of thesebipolar disorder, based primarily on ascertaining subthreshold conditions within the bipolar spectrumhistory of mania, have been under 1%. The same is on the basis of family history of mood disorders, aeven more true for bipolar II disorder. Table 3 history of suicide attempts, treatment seeking forprovides a breakdown of different studies conducted depression, or social impairments; comorbidity within many countries since the availability of structured substance abuse and anxiety disorders was also high.diagnostic interviewing tapping criteria such as These high rates for bipolar disorders are not isolated

´ ´DSM-III and beyond. Two national studies under- instances, because Szadoczky et al. (1998) in Hun-taken in the USA have had a major impact on the gary have reported rates of 5% for bipolar spectrumrates of bipolar disorder which are cited in the disorders.literature. These are the Epidemiological Catchment In summary, the epidemiological literature fromStudy (ECA, Regier et al., 1988) and the National community studies in the USA and several EuropeanComorbidity Survey (NCS, Kessler et al., 1994). The countries strongly favors the inclusion of soft bipo-rates are, respectively, 1.2% and 1.6%. Another larity within the spectrum of bipolar disorders. Mostinfluential study (Weissman et al., 1996), compared importantly, this emerging literature considerablyrates in different countries reporting a cross-national broadens the bipolar spectrum from the conventionalrange of 0.3–1.5%. All of these rates were observed rate of 1% to at least 5%.

Table 3 4. Characterizing cyclothymia, hypomania andLifetime prevalence rates of bipolar disorder beyondAuthor (year /country) Rate (%)

4.1. CyclothymiaRegier et al. (1988) /USA 1.2Kessler et al. (1994) /USA 1.6

a Hypomania is critical for the definition of bipolarLewinsohn et al. (1995) /USA 5.7spectrum conditions below the threshold of mania.Weissman et al. (1996) /cross national 0.3–1.5

a´ ´Szadoczky et al. (1998) /Hungary 5.0 We will first document the phenomenology of hypo-aAngst (1998) /Switzerland 8.3 mania in the course of cyclothymia, because most

a Emerging new data. studies on hypomania derive from cohorts with either

S10 H.S. Akiskal et al. / Journal of Affective Disorders 59 (2000) S5 –S30

cyclothymia or recurrent hypomania. The increased that slightly under 10% of a mental health clinic’srates for bipolar spectrum conditions in community patients conformed to subsyndromal mood changesstudies are in line with classic concepts of manic- over extended periods of time. These were youngdepressive illness deriving from the work of such adults who presented clinically because of socialauthorities as Kraepelin (1921) and Kretschmer disruptions in their lives, such as romantic failure,(1936), who wrote about affective states which range financial extravagance, repeated change of line offrom the severest to the mildest, and which pass work or college studies, frequent geographicalwithout sharp boundary into the domain of personal moves, and polysubstance abuse. The underlyingpredisposition or temperament. Both described affective diathesis was validated on the basis ofcyclothymic individuals in whom low-grade affective phenomenological criteria that involved biphasicmanifestations of a subdepressive and hypomanic subsyndromal changes in energy, activity, mood, andnature oscillated over long periods of the life span. cognition, each phase typically lasting from 2 days toWhile in the classic, and especially German literature a week; some oscillated more in a depressive direc-(Schneider, 1959), cyclothymia refers to the entire tion, others more in a hypomanic direction, but inspectrum of manic-depressive manifestations, in its both directions at least at some point in their lives.current usage (Brieger and Marneros, 1997) With Italian collaboration involving 1010, 14–25cyclothymic disorder is restricted to a subthreshold, year old students (Placidi et al., 1998), high internalbipolar condition at the temperamental level. consistency and diagnostic specificity has been found

In some cyclothymes (Akiskal et al., 1979a), for six of the eight criteria for cyclothymic tempera-depressive or irritable moodiness predominates, in ment developed at the University of Tennesseeothers, trait hypomanic features (known as hyper- (Akiskal et al., 1979a). These revised criteria (listedthymic temperament) are more characteristic. These in Table 4) are thus more valid than their DSM-IVcould occur throughout life without progression to and ICD-10 counterparts from both clinical andmajor affective episodes, or represent predisposing psychometric standpoints. The subthreshold oscilla-or prodromal phases to more severe episodic illness; tion of hypomanic and subdepressive periods occur-upon recovery from these episodes, patients tend to ring in 6.3% of the population at large (Placidi et al.,return to their baseline temperament. Although large- 1998), represents a high-risk group predisposed toly neglected by the contemporary psychiatric estab- major affective episodes (Akiskal et al., 1977,lishment and clinical psychology, several large-scale 1985b).studies have examined cyclothymic disorder and its There has been much confusion in psychiatryvariants. These studies differ from the strictly epi- about the nature of predisposing traits to affectivedemiological studies just reviewed in that they focus disorder. Kraepelin (1921) spoke of personal pre-on clinical or student populations. disposition, and Kretschmer (1936) referred to them

The first of these studies on subthreshold bipolari- as affective temperaments. When affective oscilla-ty was conducted at the University of Tennessee, tions are extreme and are associated with veryMemphis, USA, where Akiskal et al. (1977) reported significant disruption and interpersonal conflict,

Table 4aValidated criteria for the cyclothymic

Biphasic mood swings — abrupt shifts from one phase to the other, each phase lasting for a few days at a time with infrequent euthymia. Atleast four of the following which constitute the habitual long-term baseline of the subject:

? Lethargy alternating with eutonia? Shaky self-esteem alternating between low self-confidence and overconfidence? Decreased verbal output alternating with talkativeness? Mental confusion alternating with sharpened and creative thinking? Unexplained tearfulness alternating with excessive punning and jocularity? Introverted self-absorption alternating with uninhibited people-seekinga Summarized from Akiskal et al. (1998).

H.S. Akiskal et al. / Journal of Affective Disorders 59 (2000) S5 –S30 S11

many cyclothymic individuals would also meet trum of bipolar disorders. Unfortunately, the criteriacriteria for so-called ‘borderline’ and other erratic for hypomanic episodes as described in DSM-IVpersonality disorders. Indeed, this was the case in the (1994) are insufficiently distinct from those forpast records of the patients studied at Tennessee mania. We agree with the stipulations that the(Akiskal et al., 1977, 1979a). The bipolar nature of characteristic features of hypomania must be ob-the disorder was confirmed by the propensity of the served by others and that psychotic symptoms shouldmore depressive cyclothymes to switch to hypomania be absent. Thus, major life disruptions are uncharac-and/or mania on antidepressants, as well as family teristic — indeed hypomania is sometimes adaptivehistory for bipolar disorders. Studies which have (Akiskal et al., 1977; Jamison et al., 1980). Tables 5started with borderline personality cohorts have also and 6 list the rich phenomenology of hypomaniafound high rates of cyclothymic (Levitt et al., 1990) observed, respectively, in the Memphis (Akiskal etand/or soft bipolar spectrum diagnoses (Deltito et al., 1979a,b) and Zurich (Wicki and Angst, 1991)al., in press). In a German study (Sab et al., 1993) studies — both of which go beyond the narrow rangewhich carefully rated ‘subaffective personality dis- of clinical presentations listed in DSM-IV and ICD-orders’, borderline and irritable-cyclothymic condi- 10.tions overlapped considerably. An authority on bor- The studies thus far reviewed on cyclothymia andderline conditions of the caliber of Stone (1988), has hypomania were conducted largely before the availa-declared irritable temperament to be the core under- bility of DSM-IV, and they are unanimous in validat-lying pathology in this personality type. It should not ing a duration for hypomania shorter than 4 days.come as a surprise that borderline personality has The Akiskal et al. (1979b) study specified a 2-daybeen found to be a predictor of pharmacological duration, and the Zurich study (Wicki and Angst,hypomania (Akiskal et al., 1985a; Levy et al., 1998). 1991) found a modal duration of 1–3 days. It is alsoThe adjective ‘borderline’ in many such instances apparent from these studies that recurrence of hypo-seems to refer to borderline manic-depressive psy- mania is a characteristic of disorders in the softerchosis (Akiskal et al., 1985a). spectrum. Therefore, the 4-day threshold of DSM-IV

Commonly used personality tests tend to misattri- is unjustified — and unnecessarily narrows down thebute subaffective mood changes to borderline per- range of bipolar spectrum disorders diagnosable insonality (O’Connell et al., 1991). clinical and epidemiological studies. Actually, a very

Another large study examining cyclothymia was large Italian clinical study on bipolar II patientsconducted in college students in Albany, New York (Cassano et al., 1992), which used a definition of(Depue et al., 1981; Klein et al., 1986). hypomanic duration of 2 days — whether it wasOperationalizing from the criteria developed in the episodic or part of cyclothymia — found that theseUniversity of Tennessee study — and modifying patients had rates of bipolar family history statistical-them on the basis of the classical psychiatric litera-ture and psychometric considerations — these au-

Table 5thors reported that 4–6% conformed to cyclothymic Signs and symptoms of a hypomanic episode based on a clinical

adisorder. Apart from the fact that cyclothymia was samplesignificantly higher among the offspring of bipolars

Three or more of the following, which must represent departureversus control subjects, the tendency of many of from patient’s habitual baseline for $ 2 daysthese cyclothymic students to develop depressive ? Cheerfulness and jocularity

? Gregariousness and people-seekingand/or suicidal states — as well as substance abuse? Heightened sexual drive and behaviour— during prospective observation — pointed to a? Talkativenessstrong bipolar diathesis.? Overconfidence and overoptimism? Disinhibition and carefree attitudes

4.2. Hypomania ? Hyposomnia? Eutonia and vitality? Over involvement in new projectsAccurate assessment of hypomania is critical forathe proper identification of the less-than-manic spec- Expanded from Akiskal et al. (1977, 1979a,b).

S12 H.S. Akiskal et al. / Journal of Affective Disorders 59 (2000) S5 –S30

Table 6 II patients (Cassano et al., 1992). The possibleThe most common manifestations of hypomania in a community inclusion of these ‘hyperthymic depressives’ withinastudy

the bipolar spectrum (Akiskal and Akiskal, 1988;Less sleep Akiskal and Pinto, 1999) — depending on theMore energy, strength threshold of the diagnosis of hyperthymic tempera-More self-confidence

ment — will shrink unipolarity by 10–20% (CassanoIncreased activities (including working more)et al., 1992). Current data are uncertain about theEnjoying work more than usual

More social activities (i.e., telephone calls, visiting other people) boundary of hyperthymic temperament and normalitySpending too much money (Akiskal et al., 1998); this temperament as currentlyMore plans and ideas measured may be considered abnormal only in theLess shy, less inhibited

presence of clinical depression. Further research inMore talkative than usualthis area will be important for both clinical andIncreased sex drive

Increased consumption: coffee, cigarettes, alcohol genetic investigations.Overly optimistic /euphoricIncreased laughter (making jokes, puns) 4.4. The question of axis IIThinking fast / sudden ideas

a Summarized from Angst (1998). There is an extensive literature based on clinicaland community samples which validates the defini-

ly indistinguishable from that of bipolar I disorder, tion of hypomania at a threshold lower than that setboth of which were significantly higher than that of in DSM-IV. In particular, hypomania emerges as amajor depressive disorders. condition with a duration threshold of 2 rather than 4

days; in recurrent brief hypomania, a duration of 14.3. Hyperthymia day would suffice. Furthermore, in a special sub-

group, hypomanic manifestations persist over aThere have also been studies that have focused on lifetime in a trait-like fashion: these hyperthymic

subthreshold lifelong hypomanic symptoms. Eckblad individuals have been excluded from official classifi-and Chapman (1986) studied college students at the cations, but their importance lies in the fact that theUniversity of Wisconsin: Six percent met lifetime depressive episodes which occur in such individualscriteria for hypomanic tendencies that, interestingly, are from a familial standpoint no different from thosein some cases were associated with mini-depressive with bipolar II disorder.dips. In the Pisa-San Diego collaborative study Finally, the concept of cyclothymic disorder as(Akiskal et al., 1998; Placidi et al., 1998), also defined in ICD-10 and DSM-IV would benefit fromconducted among students, 8% could be categorized better operationalization: what is crucial here is theas hyperthymic on the basis of the entire complement biphasic oscillation of subthreshold hypomanic andof seven persistent hypomanic traits (Akiskal, 1992). depressive manifestations over long periods of life-These psychometrically established traits are as time. Many cyclothymic individuals develop clinicalfollows: (1) Warm, people-seeking or extroverted; depression and therefore should be considered as a(2) cheerful, overoptimistic or exuberant; (3) uninhi- more complex form of bipolar II that might bebited, stimulus-seeking or promiscuous; (4) over- classified under ‘cyclothymic depression’ (Akiskal,involved and meddlesome; (5) vigorous, full of 1981, 1994; Akiskal and Pinto, 1999). Patients withplans, improvident or carried away by restless im- bipolar II disorder whose hypomanias are part of apulses; (6) overconfident, self-assured, boastful, cyclothymic temperamental background, presentbombastic or grandiose; (7) articulate and eloquent. with greater interpersonal disturbances and are at riskThese criteria are also validated on the basis of for being mislabeled ‘borderline’, ‘histrionic’ and/orfamily history in that clinically depressed patients ‘psychopathic’. From a therapeutic standpoint, muchwho met five or more of these criteria, had rates of is to be gained from considering such flamboyantfamilial bipolarity significantly higher than strictly individuals with erratic mood swings as sufferingunipolar patients without these temperamental attri- from a bipolar spectrum disorder. We submit thatbutes, and indistinguishable familially from bipolar patients presenting with fluctuating affective symp-

H.S. Akiskal et al. / Journal of Affective Disorders 59 (2000) S5 –S30 S13

toms, the diagnosis of a bipolar spectrum diagnoses reliability of the bipolar II diagnosis was low, allshould take precedence over that of a personality such diagnoses occurred in pedigrees with bipolardisorder within the dramatic cluster. disorder, suggesting that once hypomania was iden-

tified in association with major depressive disorder,it did carry great diagnostic specificity. What thismeans in practice is that the diagnosis of bipolar II

5. Bipolar II cannot be made in cross-section, but should be basedon repeated evaluations. The other and related meth-

5.1. Clinical diagnosis odological point was made by Dunner and Tay(1993), who found that clinicians specifically trained

This subtype refers to a common clinical situation to recognize bipolar II, far outperformed structuredwhere the patient presents with a major depressive instruments such as the SADS or the SCID in theepisode, and upon further inquiry, history for hypo- diagnosis of bipolar II disorder. Both methodologicmanic episodes is elicited. Accurate diagnostic points cohere with recommendations made by Akisk-subtyping then depends on the vagaries of the al et al. (1977) that the diagnosis of hypomaniapatient’s memory and how rigorously the clinician among cyclothymic subjects be based on repeatedpursues lead questions about hypomania — and, expert interview. Although these points go againstmost importantly whether relatives are interviewed. the usual tenets in the literature on structured inter-Otherwise, unless recorded in the patient’s past viewing, they are consistent in suggesting that thepsychiatric chart, the examining clinician may have proper identification of bipolar II requires a morelittle clue from the patient’s current depressive sophisticated approach in interviewing and diagnosis.mental state about the bipolar elements in the Another way to say this is that, because manypatient’s history. Accordingly, rates of bipolar II bipolar II patients have an underlying temperamentaldisorder were, until recently, underestimated. dysregulation, their clinical presentations are varied

The under-diagnosis of bipolar II disorder due to and inconsistent and often prove confusing in cross-under-reporting of hypoania is of critical importance section. That is, they could present with cross-sec-for both clinical practice and genetic investigations. tional features of atypical depression, and lifelongJohn Kelsoe, M.D., who heads the Bipolar Genetics history of anxiety states, bulimia, substance abuse,Program at the University of California at San Diego and personality disorder (Perugi et al., 1998; Benaz-(personal communication, March 31, 2000) summa- zi, 1999). Bipolar II is a complex diagnosis becauserized the diagnostic problem as follows: ‘‘The major it often encompasses many of these ‘comorbid’problem . . . in diagnosis is state dependent memory conditions; a prospective study has also demon-in patients. When they are high, all they remember strated that atypical depressions more often than notare past manias, when they are depressed, they only progress to bipolar spectrum disorders (Ebert et al.,recollect being depressed. I have seen many patients 1993). Faced with patients with atypical depressivefor whom we had thoroughly eliminated any history features as defined in DSM-IV, the clinician’s task isof hypomania from repeated interviews during the to identify a pattern of cyclic depressions withdepressed state, only to find that when they became distinct hypomanic periods as the core unifying orhypomanic on an antidepressant, they suddenly underlying diagnosis behind their varied comorbidremembered many past hypomanias. There is much manifestations.support for this in both animal and human studies of The question of bipolar II-anxious ‘comorbidity’ ismemory.’’ beyond the scope of the present review. Suffice it to

The foregoing considerations are of the utmost say that, though counterintuitive, it appears to have aimportance in the diagnostic workup of bipolar II stronger genetic basis than bipolar II without panicpatients. There have been two related methodological attacks (MacKinnon et al., 1998). The affectivedevelopments published in the recent literature with dysregulation of bipolar disorder obviously extendsrespect to this question. The first is by Rice et al. beyond elation and depression — to include, among(1986), who in the context of the NIMH collabora- others, such negative affective arousal states astive study of depression, reported that although the panic, irritability, and mood lability.

S14 H.S. Akiskal et al. / Journal of Affective Disorders 59 (2000) S5 –S30

5.2. The specificity of mood lability TEMPS or Temperament Evaluation of Memphis,Pisa, Paris, and San Diego (auto-evaluation version,

How should then a clinician optimally approach Akiskal et al., in press) should ultimately help in thethe diagnosis of bipolar II? Analyses from the NIMH proper identification of the soft bipolar spectrum.Collaborative Depression Study on unipolar patientswho switched to bipolar II may help in this regard. 5.3. Clinical prevalence of bipolar II in majorOf 559 patients with major depressive disorder at depressionentry, 48 converted to bipolar II during a prospectiveobservation period of 11 years (Akiskal et al., 1995). The diagnosis of bipolar II is crucial, not onlyWhat characterized these bipolar II converters at because of its therapeutic implications, but also forentry was early age at onset of first depression, prognostic reasons. Since Dunner et al. (1976)recurrent depression, high rates of divorce or sepa- pioneering report, the literature has supported theration, high rates of scholastic and/or job maladjust- risk for high suicidality in this group of patientsment, isolated ‘antisocial acts’, drug abuse — in (Rihmer and Pestality, 1999). Accordingly, it isbrief, a more tempestuous affective and life history. gratifying that a great deal of research has beenIn addition, the index depressive episode was further conducted on the clinical prevalence of bipolar IIcharacterized by such features as phobic anxiety, among patients presenting with major depressiveinterpersonal sensitivity, obsessive–compulsive disorder to various medical centres and clinicssymptoms, somatization (often with subpanic symp- worldwide. What emerges is that from 30–55% of alltoms), worse in evening, self-pity, demandingness, major depressions conform to the bipolar II or itssubjective or overt anger, jealousy, suspiciousness, variants: It is noteworthy that the data are not limitedand ideas of reference — again testifying to a broad to academic centers that specialize in mood disorders

´melange of ‘atypical’ depressive symptoms with (Akiskal and Mallya, 1987; Cassano et al., 1992), but‘borderline’ features. Temperamental attributes ob- include at least two large outpatient psychiatrictained at index interview proved decisive private practice settings (Koukopoulos et al., 1980;(sensitivity 5 91%) in identifying those who Benazzi, 1997b). Nor are these high rates limited toswitched from depression to hypomania: these attri- psychiatric settings; at least one such report hasbutes consisted of trait ‘mood lability’, ‘energy come from a family practice clinic (Manning et al.,activity’, and ‘daydreaming’ — all characteristic of 1997). Moreover, according to Simpson et al.Kretschmer’s (1936) description of the cyclothymic (1993), bipolar II may represent the most commontemperament; mood lability was the most specific phenotype of bipolar disorder.predictor (specificity 5 86%) of which depressions The French EPIDEP study (Hantouche et al.,will prospectively change to bipolar. This study 1998) based on a clinical sample from a variety oftestifies to the fact that bipolar II disorder is a hospital and clinical settings — private and public,complex affective disorder with biographical in- inpatient and outpatient, academic and general psy-stability — deriving more often than not from an chiatric sector — have provided the most compellingintense temperamental dysregulation. Mood lability data on the high prevalence of bipolar II among— with rapid shifts, often in a depressive polarity — major depressive patients. The overarching purposewas the hallmark of ‘unipolar’ patients who of this study was to assist practicing psychiatrists toswitched to bipolar II. The foregoing characteristics recognize bipolarity in all of its varieties. Preparatoryrevealed in a prospective study on a large clinical to the ambitious aim of obtaining national data oncohort in five university centers provide a pattern the full spectrum of bipolar disorders, 40 roundrecognition which clinicians can use in their diagnos- tables were conducted in six regions of Francetic evaluation for ascertaining bipolar II disorder and involving 650 psychiatrists. This culminated in aits variants. Unfortunately, our formal diagnostic Paris symposium attended by senior French profes-systems (e.g., ICD-10 and DSM-IV) are symptom- sors of psychiatry, directors of ambulatory clinicsoriented and do not consider extreme temperamental and hospitals, as well as other opinion leaders. Thesedispositions in clinical evaluation; also regrettably training activities over a period of 30 months ad-such patients often get labeled ‘borderline’. The dressed the gaps between the classic (Baillarger,

H.S. Akiskal et al. / Journal of Affective Disorders 59 (2000) S5 –S30 S15

1854; Falret, 1854; Kraepelin, 1921) and the current ICD-10 have denied distinct bipolar status to theseliterature on the one hand, and current diagnostic patients. This is a regrettable decision, a morepractice based on ICD-10 and DSM-IV systems on extensive literature existed before the publication ofthe other. New data in support of the classic litera- DSM-IV, strongly in favour of including such pa-ture was extensively discussed, especially regarding tients within the rubric of bipolar disorders (Bunneythe high prevalence of depressive conditions with et al., 1972; Akiskal et al., 1979b, 1983; Strober andmild excitement (Akiskal et al., 1979a,b; Akiskal, Carlson, 1982; Wehr and Goodwin, 1987; Sultzer1983; Egeland, 1983; Akiskal and Akiskal, 1988; and Cummings, 1989) — and new reports (AltshulerCassano et al., 1992; Simpson et al., 1993), as well et al., 1995; Menchon et al., 1993; Benazzi, 1997a;as the high prevalence of temperamental dysregula- Post et al., 1997) have been published since then.tion in many patients with cyclic or bipolar II Mood stabilizers do not seem to fully preventdepressives (Akiskal et al., 1979a, 1995; Cassano et antidepressant-associated switches (Bottlender et al.,al., 1992). The foregoing activities culminated in the 1998), though an adequate level of a mood stabilizerestablishment of a co-ordinating body consisting of might be protective (Jann et al., 1982). Antidepres-Drs Hantouche, Akiskal, plus Allillaire, Azorin, sant-mobilized hypomanic episodes tend to be some-Bourgeois and Sechter from different regions of what milder (Stoll et al., 1994), less likely to occurFrance, and the construction of a semi-structured with SSRIs compared with tricyclics (Peet, 1994;interview schedule modified from DSM-IV and Bottlender et al., 1998), and more euphoric withincorporating various rating scales and the French monoaminoxidase inhibitors than tricyclics which aretranslation of temperamental attributes from an early likely to induce more dysphoric hypomania (Him-version of the TEMPS-A (Hantouche and Akiskal, melhoch et al., 1991). However, during prospective1997). The major finding of this study, reported on observation, nearly all adult patients with antidepres-the first 250 patients evaluated nationally (Hantouche sant-associated hypomanic episodes progress monthset al., 1998) indicated that at index interview 22% of or years later to bipolar states with spontaneousmajor depressive patients could be diagnosed as hypomania or mania (Akiskal et al., 1983); this alsobipolar II based on past history of hypomania; a is true for adolescent depressives (Strober and Car-month later, upon re-interview, 40% of patients were lson, 1982). Table 7 summarizes the most sensitivediagnosed as bipolar II on the basis of more in-depth and specific parameters in the prospective predictionevaluation and collateral information from significant of bipolar outcome — and pharmacologically oc-others, as well as observed hypomania by the casioned hypomania tops the list.clinician. What has been reviewed thus far, indicates that

The question is often raised whether patients with bipolar II is a prevalent condition accounting forbipolar II disorder represent an autonomous type of one-third to one-half of all major depressive statesbipolar illness or a transitory condition between encountered in clinical practice. Those with pharma-unipolar and full-blown bipolar disorder with mania. cologically mobilized hypomania seem to represent aIt is beyond the scope of this paper to address this variant of the bipolar II pattern that can be provision-complex issue, but suffice it to say that in patients ally termed bipolar III. The data listed in Table 7with at least a 5-year history of affective illness, further suggest that depressions with bipolar familythose with the bipolar II diagnosis represent a stable should be closely observed for eventual bipolarcondition that rarely progresses to bipolar I disorder transformation. Observed hypomania on antidepres-(Coryell et al., 1995). sants may represent the first gross indication for such

transformation. Systematic interviewing may often5.4. The question of pharmacological hypomania reveal spontaneous hypomanic excursions buried in

the past history — which had been consideredAnother burning question, vital for private prac- ‘normal’ mood fluctuation. It is noteworthy, too, that

tice, pertains to hypomania that becomes first mani- antidepressant-associated hypomania is not limited tofest upon pharmacological challenge with antidepres- major depressions. It could occur in dysthymicsants. Based on Lewis and Winokur (1982), Angst patients (Rosenthal et al., 1981; Rihmer, 1990), as(1985), and Kupfer et al. (1988), both DSM-IV and well as social phobic, obsessive–compulsive, and

S16 H.S. Akiskal et al. / Journal of Affective Disorders 59 (2000) S5 –S30

Table 7 other conditions, whose link to bipolarity is tenuousDiagnostic performance of variables significantly associated with (Akiskal and Pinto, 1999): these include atypical andabipolar outcome

seasonal depressions without discernible hypomanicVariable Sensitivity Specificity states, which have periodicity and abrupt onset and

(%) (%) offset; other patients may present with episodicPharmacological hypomania 32 100 obsessive–compulsive symptoms, periodic states ofBipolar family history 56 98 irritability, or acute suicidal crises in the absence ofLoaded pedigrees 32 95

clear-cut affective symptoms; there are also patientsHypersomnic-retarded depression 59 88with cyclic episodic neurasthenic or sleep complaintsPsychotic depression 42 85

Postpartum onset 58 84 or those with severe brief recurrent depressions.Onset of depression before age 26 71 68 Lastly, McElroy et al. (1996) have drawn attention

a to the possible bipolar nature of some impulse-riddenSummarized from Akiskal et al. (1983).behaviors such as those in the realm of aggression

other anxiety states (reviewed in Himmelhoch, 1998 control, gambling and paraphilias. All of theseand Perugi et al., 1999). In brief, the depressive conditions require further study before a definite linkphase of bipolar II patients can, in a significant to bipolarity can be claimed.minority of patients, be replaced by subthresholddepressive, socially anxious or obsessive inhibitions.These clinical observations, which require greater 6. Rapid cyclingresearch validation, are quite important in clinicalcase management. Patients with rapid-cycling disorder as defined in

DSM-IV and ICD-10 present a minimum of four5.5. Soft bipolarity beyond bipolar II episodes per year, i.e., mania /hypomania and major

depression (Maj et al., 1994). They are most likely toThe foregoing discussion has considered what — arise from a bipolar II base (Coryell et al., 1992) —

on the basis of rigorous data — can be categorized and thus present with at least four alternating depres-within a broad spectrum of bipolarity short of full- sive or hypomanic episodes per year. ‘Alternating’ isblown mania and extending into the realm of tem- the correct verb, because such patients do notperament. It is beyond the scope of this paper to typically have respite from affective episodes duringconsider the differential diagnosis of adult attention- the rapid-cycling phase of their illness.deficit hyperactivity disorder (ADHD) and bipolar II. Rapid-cycling patients lie along a spectrum basedSuffice it to say that history for childhood hyperac- on the duration of episodes which, by definition,tivity is more common in adult bipolars than unipo- must meet the symptom severity thresholds forlars (Winokur et al., 1993), and that increased sexual mania /hypomania and depression. Rapid ( $ 4/drive, grandiosity and psychosis are uncharacteristic year), ultra-rapid ( $ 4/month), and ultradianof ADHD (Weller et al., 1995). Nor have we ( $ within a day) cycling patterns can be recognizedconsidered in requisite depth mood, alcohol and clinically; they are distinguished from cyclothymicstimulant abuse comorbidity (Regier et al., 1988; disorder which pursues a subthreshold course as farWinokur et al., 1998; Sonne and Brady, 1999), as symptoms.which should be carefully evaluated for bipolar In rapid cycling, bipolar illness takes on a rollerspectrum diagnoses. Regrettably, DSM-IV conven- coaster course for both patient, family, and thetions tend to favor diagnosis of alcohol and sub- physician. Rapid cycling appears to be on the risestance abuse at the expense of bipolar disorders. (Wolpert et al., 1990). Fortunately, this conditionWhen in doubt, faced with a patient who exhibits which is reported to occur in 13–56% of bipolarjoint problems of substance abuse and unrelenting patients (reviewed in Kilzieh and Akiskal, 1999),mood swings, it would often prove clinically advan- appears to be a transient phase in the course oftageous to err in favor of bipolar spectrum diagnoses bipolar disorder, rather than a distinct subtypefor which specific treatment options exist. (Coryell et al., 1992); in 2–4 years, most rapid

In this review, we have also omitted discussion of cycling observed during naturalistic follow-up will

H.S. Akiskal et al. / Journal of Affective Disorders 59 (2000) S5 –S30 S17

return to a less cycling pattern. The higher rates of psychopathological state. However, Kraepelin ob-rapid-cycling are reported from research institutions served that full affective episodes with such admix-which specialize in refractory bipolar disorders; the tures did commonly occur during the course ofprevalence is less than 20% in most studies. Risk manic depressive illness. He described depressivefactors discussed in the literature — but not unanim- admixtures occurring during mania, as well as hypo-ously agreed upon — include female sex, manic intrusions into full depressive episodes. Hiscyclothymic temperament, borderline hypothyroid- categorization included at least six types, of whichism, and possibly excessive use of antidepressants depressive or anxious mania, and agitated or excited(Koukopoulos et al., 1980; Wehr et al., 1988; Bauer depression are the most prevalent in current clinicalet al., 1994b). One recent meta-analysis (Tondo and practice. Although this broad definition of mixedBaldessarini, 1998) found inconsistent association states is well accepted in European psychiatrywith female sex. Such discrepancies are likely due to (Berner et al., 1992), it is not fully reflected inmethodologic differences in case definition and ICD-10 (1992); the narrowest definition is that ofinclusion. DSM-IV (1992) which requires fullfledged manic

Rapid cycling is typically a post hoc diagnosis. and syndromal depressive manifestations.This may be another explanation for some of the Mixed states represent a new focus of clinicaldiscrepancies in the literature regarding risk factors. research in mood disorders. There is no terminologi-Retrospective systematic analysis of large samples cal uniformity in the literature, and there is amay minimize biased observations. Thus, Perugi et regrettable tendency to use such terms as ‘mixedal. (2000) have shown that bipolar illness with state’, ‘mixed mania’, ‘depression during mania’,depression as the episode at onset, are significantly and ‘dysphoric mania’ interchangeably. Dilsaver etmore likely than manic and mixed state onsets to al. (1999) have recently described different phe-develop rapid cycling, suicidal behavior, and psy- nomenological subtypes within the larger manicchotic symptoms; mixed onsets, too, had high rates population. We will not attempt to review thisof suicide attempts, but differed from depressive literature which is still inconclusive. In this report weonsets in having significantly more chronicity yet will briefly discuss the depressive mixed statesnegligible rates of rapid cycling. The authors con- (major depressions with few hypomanic symptoms)cluded that because those with depressive onset had which, though of great clinical significance, remainreceived significantly higher rates of antidepressant under-studied; and then focus on dysphoric maniatreatment, the findings supported the hypothesis that (Post et al., 1989), the most studied form of mixedantidepressants may have played a role in the state in the literature and referring to manic con-induction of rapid cycling. Such a conclusion was ditions with such dysphoric features as irritability,also reached by Akiskal et al. (1985b) in the juvenile anxious depression, and hostile–aggressive–paranoidand young adult offspring of bipolar probands. These admixtures.considerations suggest that rapid cycling and mixed Hypomanic symptoms such as racing and gran-states represent distinct patterns in the course of diose thoughts, sexual arousal, and psychomotorbipolar disorder. acceleration have been described in major depressive

episodes in contemporary psychiatry — therebytestifying to Kraepelin’s diagnostic acumen — yet

7. Bipolar mixed states the number of studies reporting on ‘bipolar depres-sive mixed states’ are too few (Akiskal and Mallya,

Mixed states, defined as simultaneous mixtures of 1987; Koukopoulos and Koukopoulos, 1999; Perugidepressive and hypomanic symptoms, represented a et al., 1997). Unfortunately, such studies have notmajor line of evidence for Kraepelin’s concept commanded sufficient interest in official nosologiclinking mania and depressive illness that was fully systems, nor in the clinical literature. This is aenunciated in the 1899, sixth edition of his handbook clinical tragedy because these are the very ‘unipolar’(Kraepelin, 1921, english translation). As a brief depressive patients who are likely to do poorlytransitional phase from depressive to manic episodes on antidepressants and require mood stabilizers,or vice versa, they do not represent a specific antipsychotics, or electroconvulsive therapy.

S18 H.S. Akiskal et al. / Journal of Affective Disorders 59 (2000) S5 –S30

Table 8Koukopoulos and Koukopoulos (1999) have recentlyRates of mixed states in representative studieswritten a superb clinical article on agitated depres-Study Patients (N) %sion as a mixed state. This is an instance where

clinical acumen has outpaced the conventional sci- Winokur et al. (1969) 61 16entific literature. Kotin and Goodwin (1972) 20 65

Himmelhoch et al. (1976) 84 31We will be focusing the remainder of this sectionAkiskal and Puzantian (1979) 60 25to mixed states conceived as dysphoric mania. TheNunn (1979) 112 36

literature from Kraepelin until the last decade of the Secunda et al. (1985) 18 4420th century is sparse and has been masterfully Prien et al. (1988) 103 67reviewed by McElroy et al. (1992). Alcohol abuse Post et al. (1989) 48 46

Dell’Osso et al. (1991) 108 45and neuropsychiatric conditions are common inMcElroy et al. (1995) 71 40mixed states (Himmelhoch et al., 1986). MixedCassidy et al. (1998) 273 14

states have been best characterized in female inpati- Akiskal et al. (1998) 104 37ents (Dell’Osso et al., 1991; Perugi et al., 1997; Dilsaver et al. (1999) 105 40Akiskal et al., 1998), often arising from a course of

Total 1167 43illness with more depressive than manic episodes andwith a tendency to repeat over time (Perugi et al.,2000). Family history is more often depressive than interviews were conducted derived from the DSM-IVmanic (Dell’Osso et al., 1991), and suicidality is a schema for mixed state, but with suspension of thedistinct risk (Dilsaver et al., 1993; Strakowski et al., arbitrary DSM-IV threshold of full syndromal de-1996; Goldberg et al., 1998). Confusion and psychot- pression. Patients were also extensively tested psy-ic features, including mood incongruence, are also chometrically, including the French version of theimportant clinical characteristics in cross section TEMPS (Hantouche and Akiskal, 1997). Because(Dell’Osso et al., 1993; Perugi et al., 1997). An patients were entered into the study on the basis ofaverage of at least 40% of all patients with bipolar meeting full criteria for index manic episodes, thedisorder give evidence at one point or another of rates for strictly defined DSM-IV mixed states werehaving a mixed state (Table 8); these rates vary, low, 6.7%. But using a cutoff of two or moredepending on the criteria used (narrow or broad) and depressive symptoms, 37% could be characterized aswhether the setting is a community hospital versus a dysphoric manic. As expected, these patients scoredtertiary care or specialized bipolar unit. From a more than 10 on the modified Hamilton-D Scale.diagnostic standpoint, the most important advance in Depressed mood and suicidal thoughts had the bestour understanding of mixed states has come from predictive diagnostic value for mixed mania. Anstudies during the past decade, and which indicate important finding of this study was that mixed manicthat the DSM-IV threshold for syndromal depression patients, compared with those with pure mania, had aduring mania is too restrictive in the diagnosis of higher percentage of depressive temperamental traits.mixed mania. A large literature from both European Such data argue that mixed mania can be definedand US centres (Bauer et al., 1994a; McElroy et al., categorically by two or more depressive symptoms,1995; Perugi et al., 1997; Swann et al., 1997; psychometrically on the basis of HAM-D . 10, orAkiskal et al., 1998) suggest that few depressive dimensionally on the basis of depressive tem-symptoms would suffice in validating the clinical peramental traits similar to long-standing dysthymia.diagnosis of mixed mania. The latter had also been observed in the Pisa-San

The most convincing data comes from the French Diego collaborative study (Perugi et al., 1997)EPIMAN study which was conducted in four centres supporting, hypothetically, the possibility that mixedin France, and involving over 100 patients (Akiskal mania represents mania arising from a depressiveet al., 1998). Like its companion EPIDEP, this study temperamental baseline — i.e., a mixed state conce-involved extensive training for French clinicians and ived as a reversal of temperament to its oppositeacademic opinion leaders about the emerging litera- polarity (Akiskal, 1992).ture on dysphoric mania. Semistructured diagnostic There are no studies of mixed mania examining

H.S. Akiskal et al. / Journal of Affective Disorders 59 (2000) S5 –S30 S19

prospectively the list of discriminatory depressive ‘schizobipolar’ subtype may well represent an evensymptoms developed by McElroy et al. (1992). more severe course variant of bipolar disorder (VanBased on the literature (Bauer et al., 1994a; Cassidy Eerdewegh et al., 1987; Marneros, 1999). What inet al., 1997, 1998; Akiskal et al., 1998), a constella- practice distinguishes this disorder from more classiction of largely emotional–cognitive symptoms ap- mania is the presence of mood-incongruent featurespear as the most discriminatory (Table 9) and worthy during mood-free intervals.of further investigation. McElroy et al. (1992) pro- Differential diagnosis of manic and schizophrenicposed a cut-off of $ 3, Akiskal et al. (1998) $ 2, patients does not present much difficulty in patientsand Swann et al. (1997) $ 1, depressive symptoms who have had previous bouts of illness. The coursein the midst of mania for the diagnosis of mixed and distinctive clinical features of the two disordersstate. These are not mere nosologic nuances, because would, in most instances, lead to appropriate diag-even one depressive symptom during mania seems to nostic assignment. The differential diagnosis is morepredict low response to lithium and good response to problematic in adolescent patients. Particularlydivalproex (Swann et al., 1997). problematic are such symptoms as ‘loose associa-

tions’ and ‘flatness of affect’ (Akiskal, 1994). Re-garding ‘looseness’, if this disturbance in the stream

8. Psychotic forms of mania of thought occurs in a manic psychosis, it would beassociated with pressure of speech, distractibility and

The literature amply testifies to the occurrence of expansive mood, thereby clinching the diagnosis; bypsychotic symptoms, including mood-incongruent contrast, in schizophrenia ‘looseness’ occurs in thefeatures, during manic episodes (Carlson and Good- context of derailment of thought, perseveration, andwin, 1973; Taylor and Abrams, 1973; Pope and restricted affect, again pointing to a diagnosis awayLipinski, 1978; Akiskal and Puzantian, 1979). These from a manic psychosis. As far as ‘flatness’, in anhave been incorporated into the DSM-IV and ICD-10 affective psychosis, one would observe severe de-diagnostic schemas. Manic patients with such ex- pression and slowed thinking; whereas, in a schizo-treme psychotic manifestations appear as more se- phrenic psychosis, ‘flatness’ will be associated withvere versions of bipolar disorder. Discussion of inappropriate affect and poverty of thought content.schizoaffective disorders, which are quite heteroge- In brief, in the differential diagnosis of affective andneous, is beyond the scope of the present review. schizophrenic psychoses, rather than depending onSuffice it to say that the ‘schizomanic’ or pathognomonic signs, the clinician must be guided

by a pattern of signs and symptoms which arecharacteristic of one rather than the other psychosisTable 9

Depressive symptoms to be evaluated in supporting a diagnosis of (Andreasen and Akiskal, 1983).adysphoric mania There has been recent research in possible genetic

overlap between certain bipolar and schizophrenicDiscriminatory? Depressed mood disorders (reviewed in Berrettini, 2000). Whether? Irritability these data argue for a continuum between the two? Mood lability groups of disorders or a specific clinical subtype with? Anhedonia

shared oliogenic diathesis is presently unresolved.? Hopelessness /helplessnessThis question is beyond the scope of the present? Suicidal ideation and/or attempt

? Guilt review.? Fatigue

Nonspecific9. Life charting of patients’ course? Agitation

? Insomnia? Weight changes Many clinical settings that specialize in thea evaluation and treatment of bipolar disorder utilizeBased on McElroy et al. (1992), Bauer et al. (1994a,b),

Cassidy et al. (1998), Akiskal et al. (1998). some form of life charting — a graphical method for

S20 H.S. Akiskal et al. / Journal of Affective Disorders 59 (2000) S5 –S30

detailing the course of manic-depressive illness that challenged the conservative figures of 1% commonlywas championed by Kraepelin (1921). A more cited in the literature for bipolar disorder. If onesophisticated version has been adopted and further were to include bipolar spectrum conditions, thedeveloped at the National Institute of Mental Health rates jump to 5% or even higher. Clinical studies(Leverich and Post, 1998) as a research tool to detail indicate that bipolar disorders may be nearly asthe polarity and severity of episodes and their common as unipolar disorders. The new proposedlongitudinal course. Significant biological (e.g., spectrum is in accordance with Kraepelin’s positionmenarche, menopause, somatic disorders and their which proposed a manic depressive condition that attreatment) and psychosocial events (e.g., marriage, the one extreme verged on the psychotic, and at thebereavement, geographical move, job loss) can be other extreme merged with affective temperaments.easily highlighted. Most importantly, the treatments The diagnostic categories which are established inand their impact on the course of the illness can be this broad spectrum include:documented. We submit that clinicians — and pa-tients! — can easily learn this technique or a • Bipolar I for manic episodes with or withoutsimplified variant of it, adjusted to their specific major depression: the illness can take an extreme-clinical needs, thereby optimizing record-keeping of ly psychotic form, including schizobipolar var-the course of illness and its progress under different iants.therapeutic conditions. The benefits of life charting • Bipolar II refers to patients with recurrent majorare summarized in Table 10, and the methodology depressions associated with spontaneous hypo-provided in greater detail in Appendix A. mania, and representing the most common pheno-

type of bipolar disorder; current data indicate thatthe modal duration of hypomanic episodes is two

10. Conclusions days. Recurrent brief hypomanias, with excitationas short as one day, when complicated by major

Emerging data from several epidemiological depression, should also be classified as a variantstudies conducted both in the USA and abroad have of bipolar II.

Table 10aThe benefits of life charting

Document prior course Patient as active partner Applicability for clinical research

Assess partial treatment Directs future clinical trials Uniform, systematicresponses (sequential trials and rational longitudinal assessments

polypharmacy) across patients and sites

Continuity between Opportunity for education Useful for evaluating detailsretrospective and prospective Target psychotherapy of drug responsivenessassessments

Psychosocial precipitants Increased compliance Severity, duration, andpatterning of affectivedisturbances are mapped

Seasonal variation Detection of early warning system

Tolerance patterns Medicalization of illness Illness variables quantifiedbased on daily prospectiveratings

Manic switches or cycle Destigmatization Precise structure of illnessacceleration (e.g., TCA, Portable history thus available for subsequentMAOI, SSRI) Enables consultations definitions of ill and well

states (arbitrary cutoffs notneeded)

a Based on Leverich and Post (1998).

H.S. Akiskal et al. / Journal of Affective Disorders 59 (2000) S5 –S30 S21

• Another variant of the bipolar II pattern can be will to measure bipolarity broadly across differenttermed ‘cyclothymic depression’. These are major pattens of alcohol and substance use, abuse, anddepressive episodes superimposed on cyclothymic addiction.mood swings. The foregoing conclusions, based on an extensive

• As for clinically depressed patients who ex- literature, challenge several conventions in our for-perience hypomanias during antidepressant treat- mal classificatory system (i.e., ICD-10 and DSM-IV)ment (sometimes referred to as bipolar III), the regarding duration thresholds and multiple barriers inevidence is nearly unanimous in supporting bipo- the diagnosis of various bipolar subtypes. We submitlar status for these patients. that the enlargement of classic bipolar disorders to

include a spectrum of conditions ranging fromPatients within the spectrum, especially when psychotic to ambulatory forms will greatly enhance

recurrence is high and the interepisodic period is not both clinical practice and research endeavors. Lastly,free of affective manifestations, may meet criteria for some form of life charting of patients‘ course willpersonality disorders. This is particularly true for further enhance the documentation of the naturalbipolar II disorder arising from a cyclothymic progression of bipolar disorder — a colourfulbaseline. In view of their extreme mood lability, graphic representation of episodes of all severity andespecially when pursuing subacute or chronic course, residual or subthreshold symptoms, as well as stres-these patients are often misclassified as borderline sors and treatments provided.personality disorder. Treatment considerations indi-cate that in the presence of prominent affective

1symptoms of bipolar nature, a bipolar spectrum Appendix A. Charting patients’ coursediagnosis should have precedence to Axis II per-sonality disorders. Actually, mood lability has been In his pioneering work, Emil Kraepelin (1921)prospectively validated as a sensitive and specific instituted a systematic and detailed approach topredictor of bipolar II outcome. patient care by recording acutely and longitudinally

Rapid-cycling represents a transient phase in the each patient’s manic and depressive episodes using acourse of bipolar — especially bipolar II — disorder, life chart graph. This description of the course ofand occurring in up to 20% of patients. It is distinct bipolar disorder generated the most comprehensivefrom mixed states, which are presently best char- and still extraordinarily relevant clinical observationsacterized as dysphoric-mania. Mixed states occur in that continue to have far-reaching implications toan average of 40% of bipolar patients, and do not date. Moreover, as documented by Kraepelin, inneed to have the full constellation of depressive and some patients the illness can progress from: (1)manic symptoms; two or more depressive symptoms isolated, intermittent episodes; (2) to more rapidappear sufficient in imparting mixed state status to recurrences with regular or irregular patterning; (3)manic patients. to a pattern of rhythmic, continuous cycling; and, (4)

Alcohol and substance abuse is highly comorbid ultimately, to one of ultradian and chaotic frequen-across the entire spectrum of bipolar disorders. cies and patterns (Kramlinger and Post, 1995) asAlthough the present review did not consider with schematized in Fig. 1 and delineated in Fig. 2.the requisite depth the differential diagnosis of Life charts make no a priori assumptions aboutbipolar from alcohol and/or substance abuse dis- patients’ course of illness but collect informationorders, the latter have been given an exaggerated about the retrospective and prospective course ofpreferential status in DSM-IV. Thus, so-called al- illness in a systematic and continuous fashion bycohol- or substance-induced mood disorders should using functional impairment as a measure of episodenot be diagnosed before prospective follow-up; it isnot uncommon at all for mood swings to persist

1The appendix was kindly provided by Gabrielle S. Leverich,following detoxification, suggesting that these dis-M.S.W. from the Biological Psychiatry Branch of the National

orders may have much in common with bipolar Institute of Mental Health. It is based on ongoing protocols beingspectrum conditions. This is an area of fertile clinical pursued as part of the Stanley Foundation Bipolar programresearch in search of investigators with the requisite (Leverich and Post, 1998).

S22 H.S. Akiskal et al. / Journal of Affective Disorders 59 (2000) S5 –S30

severity that can be corroborated by family members, Mild hypomania involves a distinct increase offriends, diaries, calendars and so on, and are further energy with either no impairment or an enhancedvalidated through pertinent records such as hospital ability to function. A period of hypomania in theand physician notes. The NIMH Life Chart Meth- mild range only would still be considered andodology (NIMH-LCME, Leverich and Post, 1998) counted as an episode because of the tendency by thethus provides a longitudinal description of the var- patient and family to minimize or ‘overlook’ theseiegated course of bipolar illness and its response to periods. Moderate mania is clearly noticeable to thetreatment in the different phases of illness progres- patient’s environment and causes significant disrup-sion. tions in behavior and productivity of the patient.

As illustrated in Fig. 1, retrospective and prospec- Severe mania includes out-of-control activity, bizarretive course of illness is charted with manic episodes behaviors, often no sleep for days, and possibleabove and depressive episodes below a date line that psychosis; the patient is unable to function in anyalso signifies baseline or euthymic mood. goal-oriented activities and requires close supervi-

Time domain for retrospective assessment is by sion or hospitalization. A dysphoric, depressive, ormonth while the prospective (i.e., current) ratings are anxious mania at any severity level is denoted bydone on a daily basis. Daily prospective ratings are cross-hatching. Hospitalizations for mania or depres-instituted to promote better monitoring of acute and sion are shaded in for easy recognition of thelongitudinal treatment response and to achieve better patient’s most difficult phases of illness.symptom control through dose adjustments and In daily prospective ratings manic and depressivetreatment augmentation. symptoms of moderate severity are further distin-

Assessment of episode severity uses criteria that guished. Low moderate signifies functioning withare based on the degree of functional incapacitation some difficulty, while high moderate includes mucharising from mood disturbances. This categorization difficulty in functioning.of episode severity by functional impairment helps Treatments, including medications and psycho-uncover affective episodes that otherwise might have therapy, are charted directly above the depressivebeen missed. For example, patients might not re- and manic ratings (Fig. 1) to better elucidate andmember a period of depressed mood but associated evaluate both acute and longterm responsiveness tomissed days of work or plummeting academic grades single or multiple treatments and their dose adjust-as a result of a mood change will be recalled with ments. This graphic depiction helps elucidate detailsgreater certainty and accuracy. of treatment response that is not always obvious

Retrospective life-charting employs three levels from chart review or office notes.of episode severity as a conservative measure of While life charting facilitates acute and long-termretrospective recall. Mild depression indicates a evaluation of pharmacotherapies, it also allows thedistinct alteration in mood from normal but is not assessment of psychosocial stressors, seasonality,associated with functional impairment in the pa- endocrine determinants such as menarche, childbirth,tient’s usual social or occupational roles. Moderate menopause, and other potential precipitating factorsdepression signifies a distinct increase in mood such as anniversaries of significant events. Recordingsymptoms and patients have significant difficulties in of stressors in the life event section, intercalatedtheir usual roles but are able to function with extra below the depressive phases (Fig. 1), forms the basiceffort. In severe depressive episodes the patient is outline of the patient’s psychosocial history, providesessentially unable to function outside or inside the a useful framework for inquiry about episodes, andhome, requires supportive or protective care, or is assists in the estimation of the patient’s stresshospitalized. Only depressive episodes of moderate reactivity to repeated stressors (i.e., matching events)or greater severity are integrated into a formal and novel occurrences in the longitudinal course ofepisode count (unless contiguous to a moderate or the illness. An event can be rated for its negative orsevere episode, thereby adding to the duration factor positive impact (from 2 4 to 1 4).of that episode); but the recording of mild depression Patients are encouraged to begin daily prospectivecan contribute to a better estimation of subsyndromal ratings at the time of first treatment contact, whilesymptomatology and completeness of remission. they construct their own retrospective life chart in

H.S.

Akiskal

etal.

/Journal

ofA

ffectiveD

isorders59

(2000)S5

–S30

S23

Fig. 1. Schema for graphing course of affective illness.

S24H

.S.A

kiskalet

al./

Journalof

Affective

Disorders

59(2000)

S5–

S30

Fig. 2. Phases in illness evolution and tretment response in a bipolar II female.

H.S. Akiskal et al. / Journal of Affective Disorders 59 (2000) S5 –S30 S25

Fig. 3. NIMH 2 LCM prosective self-ratings (‘‘My Chart’’).

S26 H.S. Akiskal et al. / Journal of Affective Disorders 59 (2000) S5 –S30

more detail when less ill or euthymic to facilitate hours of sleep are omitted and a check box for therecall. Wherever possible, the life chart is reworked presence of psychosis is included.and expanded in collaboration with the clinician as In two studies Denicoff et al. (1997, in press)part of ongoing clinical care and history taking. found preliminary evidence of the reliability and

Prospective ratings can be accomplished in six validity of the prospective life chart technique in thatsteps: LCM depression ratings were highly correlated with

Hamilton depression ratings (r 5 0.86), or Inventory1. Patients are instructed to first rate their mood on of Depressive Symptomatology (r 5 0.79). LCM

the mood analogue scale (where 0 is most de- mania ratings were correlated with the Young Maniapressed ever, 50 ‘balanced’, and 100 is most Rating Scale (r 5 0.61, r 5 0.66), and both wereactivated or manic ever). highly correlated with the Global Assessment Scale

2. Assess how much their mood has affected their (r 5 0.81, r 5 0.73).ability to function in their usual daily social and As in accurate daily monitoring of urine or bloodoccupational roles. If rapid and distinct mood glucose in diabetes, an accurate life charting of theswitches occur within a single day (ultradian illness may lead to the best acute and long-termcycling), this is depicted graphically as in Fig. 1. treatment decisions, the best chance for minimizing

3. Record the number of switches /day as well as the the impact of the illness on physiology, biochemis-highest and lowest mood ratings for the day. This try, and behavior, and the most effective approachtype of rapid, dramatic, and distinct mood fluctua- for developing optimal treatment paradigms andtions can be between depression and either algorithms for patients with bipolar illness. The lifeeuphoric or dysphoric mania allowing the distinc- chart delineates the diverse and often-times progres-tion between ultra-ultra rapid cycling and sive course of the illness, elucidates subcategoriesdysphoric mania. within the illness pattern, and helps address complex

4. Hours of sleep of the previous night (daytime treatment issues (e.g., development of treatmentnaps are not included) are entered and rounded to resistance, cycle acceleration, and possible switchesthe nearest hour. Menses can be tracked on an in polarity). It would appear, therefore, that a carefuladditional date line at the bottom of the rating longitudinal mapping of the course of illness andform by circling the relevant days for the month. response to treatment is important to the optimal

5. The patient then charts the number of tablets assessment and treatment of the bipolar patient. Usetaken of each medication. of this or related techniques are highly recommended

6. Life events (and their impact), severity of side for detailed tracking of the illness in clinical settings.effects, and other symptoms can then be indi- Clinicians may wish to adopt a simplified version ofcated. graphing a patient’s course and treatment — modi-

fied to their specific clinical setting.In this fashion, a full daily prospective rating takes

few minutes, and is easy to accomplish after a littlepractice. Establishing a nightly routine of ratings Referencesprior to brushing one’s teeth or taking one’s medica-tions is encouraged. Most patients like the process Akiskal, H.S., 1981. Subaffective disorders: Dysthymic,

cyclothymic, and bipolar II disorders in the ‘‘borderline’’and come to greatly value their active participation inrealm. Psychiatr. Clin. North Am. 4, 25–46.the management of their own illness. The patient can

Akiskal, H.S., 1983. The bipolar spectrum: new concepts inthus be in possession of his or her life chart as a classification and diagnosis. In: Grinspoon, L. (Ed.). Psychiatry‘portable psychiatric history’, which is invaluable in Update: The American Psychiatric Association Annual Review,case of a consultation or transition to a new physi- Vol. II. American Psychiatry Press, Washington, DC, pp. 271–

292.cian.Akiskal, H.S., 1992. Delineating irritable-choleric and hyper-Fig. 3 presents a sample rating of a patient (self)

thymic temperaments as variants of cyclothymia. J. Person.rated prospective NIMH-LCME. The prospective Disord. 6, 326–342.clinician LCM rating form is similar to the self-rated Akiskal, H.S., 1994. Dysthymic and cyclothymic depressions:version except the 100-mm mood analogue scale and therapeutic considerations. J. Clin. Psychiat. 55, 46–52.

H.S. Akiskal et al. / Journal of Affective Disorders 59 (2000) S5 –S30 S27

Akiskal, H.S., 1996. The prevalent clinical spectrum of bipolar Andreasen, N.C., Akiskal, H.S., 1983. The specificity ofdisorders: beyond DSM-IV. J. Clin. Psychopharmacol. 17 Bleulerian and Schneiderian symptoms: A critical re-evalua-(Suppl 3), 117–122. tion. Psychiatr. Clin. North Am. 6, 41–54.

Angst, J., 1966/1973. The etiology and nosology of endogenousAkiskal, H.S., 1999. Bipolarity: Beyond Classic Mania. Psychiat-depressive psychoses. Foreign Psychiatry 2.ric Clinics of North America 22, 512–703.

Angst, J., Frey, R., Lohmeyer, B., Zerbin-Rudin, E., 1980. BipolarAkiskal, H.S., Djenderedjian, A.H., Rosenthal, R.H., Khani, M.K.,manic-depressive psychoses: results of a genetic investigation.1977. Cyclothymic disorder: Validating criteria for inclusion inHuman Genet. 55, 237–254.the bipolar affective group. Am. J. Psychiatry 134, 1227–1233.

Angst, J., 1985. Switch from depression to mania — a recordAkiskal, H.S., Khani, M.K., Scott-Strauss, A., 1979a. Cyclothymicstudy over decades between 1920 and 1982. Psychopathologytemperamental disorders. Psychiatr. Clin. North Am. 2, 527–18, 140–154.554.

Angst, J., Merikangas, K., 1997. The depressive spectrum: diag-Akiskal, H.S., Rosenthal, R.H., Rosenthal, T.L., Kashgarian, M.,nostic classification and course. J. Affect. Disord. 45, 31–40.Khani, M.K., Puzantian,V.R., 1979b. Differentiation of primary

Angst, J., 1998. The emerging epidemiology of hypomania andaffective illness from situational, symptomatic, and secondarybipolar II disorder. J. Affect. Disord. 50, 143–151.depressions. Arch. Gen. Psychiatry 36, 635–643.

`Baillarger, J., 1854. De la folie a double forme. Ann Med PsycholAkiskal, H.S., Puzantian, V.R., 1979. Psychotic forms of depres-(Paris) 6, 367–391.sion and mania. Psychiatr. Clin. North Am. 2, 419–439.

Bauer, M.S., Whybrow, P.C., Gyulai, L., Gonnel, J., Yeh, H.S.,Akiskal, H.S., Walker, P.W., Puzantian, V.R., King, D., Rosenthal,1994a. Testing definitions of dysphoric mania and hypomania:T.L., Dranon, M., 1983. Bipolar outcome in the course ofprevalence, clinical characteristics and inter-episode stability. J.depressive illness: phenomenologic, familial, and pharmacolog-Affect. Disord. 32, 201–211.ic predictors. J. Affect. Disord. 5, 115–128.

Bauer, M.S., Calabrese, J., Dunner, D.L., Post, R., Whybrow, P.C.,Akiskal, H.S., Chen, S.E., Davis, G.C., Puzantian, V.R., Kas-Gyulai, L. et al., 1994b. Multisite data reanalysis of the validityhgarian, M., Bolinger, J.M., 1985a. Borderline: an adjective inof rapid cycling as a course modifier for bipolar disorder insearch of a noun. J. Clin. Psychiatry 46, 41–48.DSM-IV. Am. J. Psychiatry 151, 506–515.Akiskal, H.S., Downs, I., Jordan, P., Watson, S., Daugherty, D.,

Benazzi, F., 1997a. Antidepressant-associated hypomania in out-Pruitt, D.B., 1985b. Affective disorders in referred children andpatient depression: A 203-case study in private practice. J.younger siblings of manic depressives: mode of onset andAffect. Disord. 46, 73–77.prospective course. Arch. Gen. Psychiatry 42, 996–1003.

Benazzi, F., 1997b. Prevalence of bipolar II disorder in outpatientAkiskal, H.S., Mallya, G., 1987. Criteria for the ‘soft’ bipolar

depression: A 203-case study in private practice. J. Affect.spectrum: treatment implications. Psychopharmacol. Bull. 23,

Disord. 43, 163–166.68–73.

Benazzi, F., 1999. Prevalence of bipolar II disorder in atypicalAkiskal, H.S., Akiskal, K., 1988. Re-assessing the prevalence of

depression. Eur. Arch. Psychiatr. Clin. Neurosci. 249, 62–65.bipolar disorders: Clinical significance and artistic creativity.

Berner, P., Gabriel, E., Katsching, H., Kieffer, W., Koehier, K.,Psychiatrie et Psychobiologie 3, 29s–36s.

Lenz, G. et al., 1992. Diagnostic Criteria For FunctionalAkiskal, H.S., Maser, J.D., Zeller, P., Endicott, J., Coryell, W.,

Psychoses, 2nd Edition. Cambridge Press, UK.Keller, M., 1995. Switching from ‘unipolar’ to ‘bipolar II’: an

Berrettini, W.H., 2000. Susceptibility loci for bipolar disorder:11-year prospective study of clinical and temperamental pre-

overlap with inherited vulnerability to schizophrenia. Biol.dictors in 559 patients. Arch. Gen. Psychiatry 52, 114–123.

Psychiatry 47, 245–251.Akiskal, H.S., Placidi, G.F., Signoretta, S., Ligouri, A., Gervasi, Bertelsen, A., Llaovald, B., Hauge, M., 1977. A Danish twin study

R., Maremmani, I. et al., 1998. TEMPS-I: delineating the most of manic-depressive disorders. Br. J. Psychiatry 130, 330–351.discriminant traits of cyclothymic, depressive, irritable and Bottlender, R., Rudolf, D., Strauss, A., Moller, H.J., 1998.hyperthymic temperaments in a nonpatient population. J. Antidepressant-associated maniform states in acute treatmentAffect. Disord. 51, 7–19. of patients with bipolar I depression. Eur. Arch. Psychiatr.

Akiskal, H.S., Pinto, O., 1999. The evolving bipolar spectrum. Clin. Neurosci. 248, 296–300.Prototypes I, II, III, and IV. Psychiatr. Clin. North Am. 22, Bowden, C.L., 1978. Lithium-responsive depression. Compr.517–534. Psychiatr. 19, 224–231.

Akiskal, H.S., Perugi, G., Hantouche, E., Haykal, R., Manning, S., Brieger, P., Marneros, A., 1997. Dysthymia and cyclothymia:Connor P., in press. The affective temperament scales of historical origins and contemporary development. J. Affect.Memphis, Pisa, Paris and San Diego: progress towards a Disord. 45, 117–126.self-rated auto-questionnaire version (TEMP-A). J Affect Dis- Bunney, Jr. W.E., Goodwin, F.K., Murphy, D.L., House, K.M.,ord. Gordon, E.K., 1972. The ‘switch process’ in manic-depressive

Altshuler, L.L., Post, R.M., Leverich, G.S., Mikalauskas, K., illness. II: relationship to catecholamines, REM sleep, andRosoff, A., Ackerman, L., 1995. Antidepressant-induced mania drugs. Arch. Gen. Psychiatry 27, 304–309.and cycle acceleration: a controversy revisited. Am. J. Psychi- Carlson, G G.A., Goodwin, F.K., 1973. The stages of mania: Aatry 152–8, 1130–1138. longitudinal analysis of the manic episode. Arch. Gen. Psychi-

American Psychiatric Association, 1994. Diagnostic and Statistical atry 28, 221–228.Manual of Mental Disorders, 4th Edition. APA Press, Washing- Cassano, G.B., Akiskal, H.S., Savino, M., Musetti, L., Perugi, G.,ton DC. Soriani, A., 1992. Proposed subtypes of bipolar II and related

S28 H.S. Akiskal et al. / Journal of Affective Disorders 59 (2000) S5 –S30

disorders: With hypomanic episodes (or cyclothymia) and with Eckblad, M., Chapman, U., 1986. Development and validation ofhyperthymic temperament. J. Affect. Disord. 26, 127–140. a scale for hypomanic personality. J. Abnorm. Psychol. 95,

Cassidy, F., Murry, E., Forest, K., Carroll, B.J., 1997. The 214–222.performance of DSM-III-R major depression criteria in the Egeland, J.A., 1983. Bipolarity: the iceberg of affective disorders?diagnosis of bipolar mixed states. J. Affect. Disord. 46, 79–81. Compr. Psychiatry 24, 337–344.

Cassidy, F., Murry, E., Forest, K., Carroll, B.J., 1998. Signs and Endicott, N.A., 1989. Psychophysiological correlates of ‘bipolari-symptoms of mania in pure and mixed episodes. J. Affect. ty’. J. Affect. Disord. 17, 47–56.Disord. 50, 187–201. Falret, J.P., 1854. Memoire sur la folie circulaire. Bull. Acad. Natl.

Coryell, W., Endicott, J., Keller, M., 1992. Rapidly cycling Med. 19, 382–415.affective disorder: Demographics, diagnosis, family history and Fieve R.R . Dunner, D.L., 1975. Unipolar and bipolar affectivecourse. Arch. Gen. Psychiatry 49, 126–131. states. In: Flach, F.F., Draghi, S.S. (Eds.), The Nature and

Coryell, W., Endicott, J., Maser, J.D., Keller, M.B., Leon, A.C., Treatment of Depression. John Wiley & Sons, New York, pp.Akiskal, H.S., 1995. Long-term stability of polarity distinctions 145–160.in the affective disorders. Am. J. Psychiatry 152, 385–390. Gershon, E.S., Hamovit, J., Guroff, J.J. et al., 1982. A family

Dell’Osso, L., Placidi, G.F., Nassi, R., Freer, P., Cassano, G.B., study of schizoaffective, bipolar I, bipolar II, unipolar andAkiskal, H.S., 1991. The manic-depressive mixed state: famili- control probands. Arch. Gen. Psychiatry 39, 1157–1167.al, temperamental and psychopathologic characteristics in 108 Goldberg, J.F., Garno, J.L., Leon, A.C., Kocsis, J.H., Portera, L.,female inpatients. Eur. Arch. Psychiatry Clin. Neurosci. 240, 1998. Association of recurrent suicidal ideation with nonremis-234–239. sion from acute mixed mania. Am. J. Psychiatry 155, 1753–

Dell’Osso, L., Akiskal, H.S., Freer, P., Barberi, M., Placidi, G.F., 1755.Cassano, G.B., 1993. Psychotic and nonpsychotic bipolar Goodwin, F.K., Jamison, K.R., 1990. Manic-depressive Illness.mixed states: Comparisons with manic and schizo affective Oxford University Press, New York.

´disorders. Eur. Arch. Psychiatry Clin. Neurosci. 243, 75–81. Hantouche, E.G., Akiskal, H.S., 1997. Outils devaluation cliniqueDeltito, J., Martin, L., Riefkohl, J., Austria, B., Kissilenko, A., des temperaments affectifs (Clinical assessment of affective

´Corless, P., Morse, C., in press. Do patients with borderline temperaments). Encephale 23 (sp 1), 27–34.personality disorder belong to the bipolar spectrum? J Affect Hantouche, E.G., Akiskal, H.S., Lancernon, S., Allilaire, J.F.,

ˆDisord. Sechter, D., Azorin, J.M., Bourgeois, M., Fraud, J.P., Chatenet-ˆDenicoff, K.D., Smith-Jackson, E.E., Disney, E.R., Suddath, R.L., Duchene, L., 1998. Systematic clinical methodology for val-

Leverich, G.S., Post, R.M., 1997. Preliminary evidence of the idating bipolar-II disorder: Data in mid-stream from a Frenchreliability and validity of the prospective life-chart methodolo- national multisite study (EPIDEP). J. Affect. Disord. 50, 163–gy (LCM-p). J. Psychiatr. Res. 31, 593–603. 173.

Denicoff, K.D., Leverich, G.S., Nolen, N.A., Rush, A.J., McElroy, Himmelhoch, J.M., Mulla, D., Neil, J.F., Detre, T.P., Kupfer, D.J.,S.L., Keck, P.E., Suppes, T., Altshuler, I.L., Kupka, R., Frye, 1976. Incidence and significance of mixed affective states in aM.A., Hatef, J., Brotman, M.X., Post, R.M., Psychol Med., in bipolar population. Arch. Gen. Psychiatry 33, 1062–1066.press. Himmelhoch, J.M., Thase, M.E., Mallinger, A.G., Houck, P.,

Depue, R.A., Slater, J.F., Wolfsetter-Kausch, H., 1981. A be- 1991. Tranylcypromine versus imipramine in anergic bipolarhavioral paradigm for identifying persons at risk for bipolar depression. Am. J. Psychiatry 48, 910–916.disorder: a conceptual framework and five validation studies. J. Himmelhoch, J.M., 1998. Social anxiety, hypomania and theAbnorm. Psychol. 90, 381–437. bipolar spectrum: Data, theory and clinical issues. J. Affect.

Dilsaver, S.C., Swann, A.C., Shoaib, A.M., Bowers, T.C., Halle, Disord. 50, 203–213.M.T., 1993. Depressive mania associated with nonresponse to Hirschfeld, R.M.A., Keller, M.B., Panico, S., Arons, B.S. et al.,antimanic agents. Am. J. Psychiatry 150, 1548–1551. 1997. The National Depressive and Manic-Depressive Associa-

Dilsaver, S.C., Chen, R., Shoaib, A.M., Swann, A.C., 1999. tion consensus statement on the undertreatment of depression.Phenomenology of mania: evidence for distinct depressed, J. Am. Med. Assoc. 277, 333–340.dysphoric, and euphoric presentations. Am. J. Psychiatry 156, Jamison, K.R., Gerner, R.H., Hammen, C., Padesky, C., 1980.426–430. Clouds and silver linings: positive experiences associated with

Dunner, D.L., Gershon, E.S., Goodwin, F.K., 1976. Heritable primary affective disorders. Am. J. Psychiatry 137, 198–202.factors in the severity of affective illness. Biol. Psychiatry 11, Jann, M.W., Bitar, A.H., Rao, A., 1982. Lithium prophylaxis of31–42. tricyclic antidepressant-induced mania in bipolar patients. Am.

Dunner, D.L., Patrick, V., Fieve, R.R., 1977. Rapid cycling manic J. Psychiatry 139, 683–684.depressive patients. Compr. Psychiatry 18, 561–566. Judd, L.L., Akiskal, H.S., Paulus, M.P., 1997. The role and clinical

Dunner, D.L., Tay, K.L., 1993. Diagnostic reliability of the history significance of subsyndromal depressive symptoms in unipolarof hypomania in bipolar II patients with major depression. major depressions. J. Affect. Disord. 45, 5–18.Compr. Psychiatry 34, 303–307. Kessler, R.C., McGonagle, K.A., Zhao, S., Nelson, C.B., Hughes,

Ebert, D., Barocka, A., Kalb, R., Ott, G., 1993. Atypical depres- M., Eschleman, S. et al., 1994. Lifetime and 12-month preval-sion as a bipolar spectrum disease: evidence from a longi- ence of DSM-III-R psychiatric disorders in the United States.tudinal study: the early course of atypical depression. Psychiat- Results from the National Comorbidity Survey. Arch. Gen.ria Danubina 5, 133–136. Psychiatry 51, 8–19.

H.S. Akiskal et al. / Journal of Affective Disorders 59 (2000) S5 –S30 S29

Kilzieh, N., Akiskal, H.S., 1999. Rapid-cycling bipolar disorder: practice setting: The high prevalence of bipolar IL and relatedAn overview of recent research and clinical experience. disorders in a cohort followed longitudinally. Compr. Psychi-Psychiatr. Clin. North Am. 22, 585–607. atry 38, 102–108.

Klein, D.N., Depue, R.A., Slater, J.F., 1986. Inventory identifica- Marneros, A. (Ed.), 1999. Handbuch der Unipolaren undtion of cyclothymia. IX. Validation in offspring of bipolar I Bipolaren Erkrankungen. Georg Thieme Verlag, Stuttgart.patients. Arch. Gen. Psychiatry 43, 441–445. McElroy, S.L., Keck, P.E., Pope, H.G., Hudson, J.I., Faedda, G.L.,

Klerman, G.L., 1981. The spectrum of mania. Compr. Psychiatry Swann, A.C., 1992. Clinical and research implications of the22, 11–20. diagnosis of dysphoric or mixed mania or hypomania. Am. J.

Psychiatry 149, 1633–1644.Kotin, J., Goodwin, F.K., 1972. Depression during mania: clinicalobservations and theoretical implications. Am. J. Psychiatry McElroy, S.L., Strakowski, S.M., Keck, P.E., Tugrul, K.L., West,129, 679–686. S.A., Lonczak, H.S., 1995. Differences and similarities in

mixed and pure mania. Compr. Psychiatry 36, 184–194.Koukopoulos, A., Reginaldi, D., Laddomada, P., Floris, G., Serra,G., Tondo, L., 1980. Course of the manic-depressive cycle and McElroy, S.L., Pope, Jr. H.G., Keck, Jr. P.E., Hudson, J.I.,changes caused by treatment. Pharmakopsychiatr. 13, 156–167. Phillips, K.A., Strakowski, S.M., 1996. Are impulse-control

disorders related to bipolar disorder. Compr. Psychiatry 37,Koukopoulos, A., Koukopoulos, A., 1999. Agitated depression as299–340.a mixed state and the problem of melancholia. Psych. Clin.

North Am. 22, 547–564. Menchon, J.M., Gasto, C., Vallejo, J. et al., 1993. Rate andsignificance of hypomanic switches in unipolar melancholicKraepelin, E., 1899/1921. Manic-depressive insanity anddepression. Eur. Psychiatry 8, 125–129.paranoia. ES Livingstone, Edinburgh.

Kretschmer, E., 1936. Physique and Character. Kegan, Paul, Mendels, J., 1976. Lithium in the treatment of depression. Am. J.Trench, Trubner, London. Psychiatry 133, 373–378.

Kramlinger, K.G., Post, R.M., 1995. Ultra rapid and ultradian Nunn, C.M.H., 1979. Mixed affective states and the naturalcycling in bipolar affective illness. Br. J. Psychiatry 167 history of manic-depressive psychosis. Br. J. Psychiatry 134,(95/31), 1–10. 153–160.

Kupfer, D.J., Carpenter, L.L., Frank, E., 1988. Is bipolar II a O’Connell, C.A., Mayo, J.A., Sciutto, M.S., 1991. PDQ-R per-unique disorder? Compr. Psychiatry 29, 228–236. sonality disorders in bipolar patients. J. Affect. Disord. 23,

217–221.Kupfer, D.J., Pickar, D., Himmelhoch, J.M., Detre, T.P., 1975. Arethere two types of unipolar depression? Arch. Gen. Psychiatry Peet, M., 1994. Induction of mania with selective serotonin32, 866–871. re-uptake inhibitors and tricyclic antidepressants. Br. J. Psychi-

atry 164, 549–550.Leverich, G.S., Post, R.M., 1998. Life charting of affectivedisorders. CNS Spectrums 3, 21–37. Perugi, G., Akiskal, H.S., Micheli, C., Musetti, L., Paiano, A.,

Quilici, C., Rossi, L., Cassano, G.B., 1997. Clinical subtypes ofLevitt, A.J., Joffe, R.T., Ennis, J., MacDonald, C., Kutcher, S.P.,bipolar mixed states: Validating a broader European definition1990. The prevalence of cyclothymia in borderline personalityin 143 cases. J. Affect. Disord. 43, 169–180.disorder. J. Clin. Psychiatry 51, 335–339.

Levy, D., Kimhi, R., Barak, Y., Viv, A., Elizur, A., 1998. Perugi, G., Akiskal, H.S., Lattanzi, L., Cecconi, D., Mastrocinque,Antidepressant-associated mania: A study of anxiety disorders C., Patronelli, A. et al., 1998. The high prevalence of softpatients. Psychopharmacol. 136, 243–246. bipolar (II) features in atypical depression. Compr. Psychiatry

39, 63–71.Lewinsohn, P.M., Klein, D.L., Seeley, J.R., 1995. Bipolar dis-orders in a community sample of older adolescents: prevalence, Perugi, G., Akiskal, H.S., Ramacciotti, S., Nassini, S., Toni, C. etphenomenology, comorbidity, and course. J. Am. Acad. Child al., 1999. Depressive comorbidity of panic, social phobic, andAdolesc. Psychiatry 34, 454–463. obsessive–compulsive disorders reexamined: Is there a bipolar

II connection? J. Psychiatr. Res. 33, 53–61.Lewis, J.L., Winokur, G., 1982. The induction of mania. A naturalhistory study with controls. Arch. Gen. Psychiatry 39, 303– Perugi, G., Micheli, C., Akiskal, H.S., Madaro, D., Socci, C.,306. Quilici, C. et al., 2000. Polarity of the first episode, clinical

characteristics, and course of manic depressive illness: ALish, J.D., Dime-Meenan, S., Whybrow, P.C., Price, R.A. et al.,systematic retrospective investigation of 320 bipolar I patients.1994. The National Depressive and Manic-Depressive Associa-Compr. Psychiatry 41, 13–18.tion (DMDA) survey of bipolar members. J. Affect. Disord. 31,

281–294. Placidi, G.F., Signoretta, S., Liguori, A., Gervasi, R., Maremanni,I., Akiskal, H.S., 1998. The Semi-Structured Affective Tem-MacKinnon, D., Xu, J., McMahon, F.J., Simpson, S.G., Stine, C.,perament Interview (TEMPS-I): Reliability and psychometricMcLnnis, M.G. et al., 1998. Bipolar disorder and panicproperties in 1010 14–26 year students. J. Affect. Disord. 47,disorder in families: an analysis of chromosome 18 data. Am.1–10.J. Psychiatry 55, 829–831.

Pope, H.G., Lipinski, J.F., 1978. Diagnosis in schizophrenia andMaj, M., Magliano, L., Pirozzi, R., Marasco, C., Guarneri, M.,manic-depressive illness: A reassessment of the specificity of1994. Validity of rapid cycling as a course specifier for bipolar‘schizophrenic’ symptoms in the light of current research.disorder. Am. J. Psychiatry 151, 1015–1019.Arch. Gen. Psychiatry 35, 811–828.Manning, J.S., Haykal, R.F., Connor, P.D., Akiskal, H.S., 1997.

On the nature of depressive and anxious states in a family Post, R.M., Rubinow, D.R., Uhde, T.W., Roy-Byrne, P.P.,

S30 H.S. Akiskal et al. / Journal of Affective Disorders 59 (2000) S5 –S30

´ ´Lmnnoila, M., Rosoff, A. et al., 1989. Dysphoric mania: Szadoczky, E., Papp, Z., Vitrai, I., Rihmer, Z., Furedi, J., 1998.clinical and biological correlates. Arch. Gen. Psychiatry 46, The prevalence of major depressive and bipolar disorder in353–358. Hungary. J. Affect. Disord. 50, 155–162.

Taylor, M.A., Abrams, R., 1973. The phenomenology of mania: APost, R.M., Denicoff, K.D., Leverich, G.S., Frye, M.A., 1997.new look at some old patients. Arch. Gen. Psychiatry 29,Drug-induced switching in bipolar disorder. CNS Drugs 8,520–522.352–365.

Prien, R.F., Himmelhoch, J.M., Kupfer, D.J., 1988. Treatment of Taylor, M.A., Abrams, R., 1980. Reassessing the bipolar–unipolarmixed mania. J. Affect. Disord. 15, 9–15. dichotomy. J. Affect. Disord. 2, 195–217.

Regier, D.A., Boyd, J.H., Burke, Jr. J.D., Rae, D.S., Myers, J.K., Tondo, L., Baldessarini, L.U., 1998. Rapid cycling in women andKramer, M. et al., 1988. One-month prevalence of mental men with bipolar manic-depressive disorders. Am. J. Psychi-disorders in the United States. Based on five epidemiologic atry 155, 1434–1436.catchment area sites. Arch. Gen. Psychiatry 45, 977–986. Tsuang, M.T., Faraone, S.V., Fleming, J.A., 1985. Familial trans-

Rice, J.P., McDonald-Scott, P., Endicott, J., Coryell, W., Grove, mission of major affective disorders, is there evidence support-W.M., Keller, M.B. et al., 1986. The stability of diagnosis with ing the distinction between unipolar and bipolar disorders? Br.an application to bipolar II disorder. Psychiatry Res. 19, 285– J. Psychiatry 146, 268–271.296. Van Eerdewegh, M.M., Van Eerdewegh, P., Coryell, W., Clayton,

Rihmer, Z., 1990. Dysthymia: a clinician’s perspective, in. In: P.J. et al., 1987. Schizo-affective disorders: bipolar–unipolarBurton, S.W., Akiskal, H.S. (Eds.), Dysthymic Disorder. Royal subtyping: natural history variables: a discriminant analysisCollege of Psychiatrists, London, Gaskell, pp. 112–125. approach. J. Affect. Disord. 12, 223–232.

Rihmer, Z., Pestality, P., 1999. Bipolar II disorder and suicidal Wehr, T.A., Goodwin, F.K., 1987. Can antidepressants causebehavior. Psychiatr. Clin. North Am. 22, 667–673. mania and worsen the course of affective illness? Am. J.

Rosenthal, T.L., Akiskal, H.S., Scott-Strauss, A., Rosenthal, R.H., Psychiatry 144, 1403–1411.David, M., 1981. Familial and developmental factors in Wehr, T.A., Sack, D.A., Rosenthal, N.E., Cowdry, R.W., 1988.characterological depressions. J. Affect. Disord. 3, 183–192. Rapid cycling affective disorder: contributing factors and

Sab, H., Herpertz, S., Steinmeyer, E.M., 1993. Subaffective treatment responses in 51 patients. Am. J. Psychiatry 145,personality disorders. Int. Clin. Psychopharmacol. 1 (Suppl 1), 179–184.39–46. Weissman, M.M., Myers, J.K., 1978. Affective disorders in a US

Schneider, K., 1959. Clinical Psychopathology. Grune and Strat- urban community: the use of research diagnostic criteria in anton, New York. epidemiological survey. Arch. Gen. Psychiatry 35, 1304–1311.

Secunda, S.K., Katz, M.M., Swann, A.C., Koslow, S.H., Maas, Weissman, M.M., Bland, R.C., Canino, G.J., Faravelli, C. et al.,J.W., Chuang, S. et al., 1985. Mania: Diagnosis, state measure- 1996. Cross-national epidemiology of major depression andment and prediction of treatment response. J. Affect. Disord. 8, bipolar disorder. J. Am. Med. Assoc. 276, 293–299.113–121. Weller, E.B., Weller, R.A., Fristad, M.A., 1995. Bipolar disorder

Simpson, S.G., Folstein, S.E., Meyers, D.A. et al., 1993. Bipolar in children: misdiagnosis, underdiagnosis, and future direc-II: The most common bipolar phenotype? Am. J. Psychiatry tions. J. Am. Acad. Child Adolesc. Psychiatry 34, 709–714.150, 901–903. Wicki, W., Angst, J., 1991. The Zurich study. X. Hypomania in a

Sonne, S.C., Brady, K.T., 1999. Substance abuse and bipolar 28- to 30-year-old cohort. Euro Arch. Psychiatr. Clin. Neuro-comorbidity. Psychiatr. Clin. North Am. 22, 609–627. sci. 40, 339–348.

Stone, M.H., 1988. Toward a psychobiological theory of border- Winokur, G., 1980. Is there a common genetic factor in bipolarline personality disorder: Is irritability the red thread that runs and unipolar affective disorder? Compr. Psychiatry 21, 460–through borderline conditions? Dissociation 1, 2–15. 468.

Stoll, A.L., Mayer, P.V., Kolbrener, M., Goldstein, E. et al., 1994. Winokur, G., Clayton, P.J., Reich, T., 1969. Manic depressiveAntidepressant-associated mania: A controlled comparison illness. CV Mosby, St. Louis.with spontaneous mania. Am. J. Psychiatry 151, 1642–1645. Winokur, G., Turvey, C., Akiskal, H., Coryell, W., Solomon, D.,

Strakowski, S.M., McElroy, S.L., Keck, Jr. P.E., West, S.A., 1996. Leon, A., Mueller, T. et al., 1998. Alcoholism and drug abuseSuicidality among patients with mixed and manic bipolar in three groups — bipolar I, unipolars and their acquaintances.disorder. Am. J. Psychiatry 153, 674–676. J. Affect. Disord. 50, 81–89.

Strober, M., Carlson, G., 1982. Bipolar illness in adolescents with Winokur, G., Coryell, W., Endicott, J., Akiskal, H.S., 1993.major depression. Arch. Gen. Psychiatry 39, 549–555. Further distinctions between manic-depressive illness (bipolar)

Sultzer, D.L., Cummings, J.L., 1989. Drug-induced mania? Causa- and primary depressive disorder (unipolar). Am. J. Psychiatrytive agents, clinical characteristics and management. A re- 150, 1176–1181.trospective analysis of the literature. Med. Toxicol. Adv. Drug Wolpert, E.A., Goldberg, J.F., Harrow, M., 1990. Rapid cycling inExper. 4, 127–143. unipolar and bipolar affective disorders. Am. J. Psychiatry 147,

Swann, A.C., Bowden, C.L., Morris, D., Calabrese, J.R., Petty, F., 725–728.Small, J., Dilsaver, S.C., Davis, J.M., 1997. Depression during World Health Organization, 1992. The ICD-10 classification ofmania. Treatment response to lithium or divalproex. Arch. Gen. mental and behavioral disorders. WHO, Geneva Switzerland.Psychiatry 54, 37–42.