Copyright © Blacknell Munksgaard 2003 BIPOLAR DISORDERS
Bipolar Disorders 2003: 5: 436^42
Review Article
The bipolar spectrum: a clinical perspective Katzow JJ, Hsu DJ, Ghaemi SN. The bipolar spectrum: a clinical perspective. Bipolar Disord 2003: 5: A},6-AA1. © Blackwell Munksgaard, 2003 The relative misdiagnosis and underdiagnosis of bipolar disorder is due in part to the 'soft' symptoms ofbipolarity that characterize patients with non-classical bipolar disorder. While no agreement has been reached on the term for this group of patients, the most common classification used is 'bipolar spectrum', which shifts the emphasis in diagnosis away from polarity and toward other diagnostic validators. In order to recognize and properly treat patients with bipolar disorder, clinicians should focus on careful evaluation of patients with mixed anxiety/depressive symptoms or impulsivity conditions (substance abuse, borderline personality, bulimia, and attention deficit disorder). Furthermore, in the treatment of bipolar disorder, clinicians should also recognize that antidepressants can have a negative effect on patients by increasing the likelihood of more severe rapid cycling. While antidepressants may be useful in particularly difficult cases, emphasis should be placed on mood stabilizers for treatment of the bipolar spectrum.
One of the major debates in the treatment of bipolar disorder revolves around the use of antidepressants. One side believes that antidepressants are necessary and the danger is either exaggerated or mitigated by the potential benefits, while the other side views antidepressants as potentially dangerous. This debate is further inflamed when looking at non-classical forms of bipolar disorder, grouped under the term bipolar spectrum disorder. Based on clinical experience, the efficacy of antidepressant treatment of bipolar disorder likely lies between the two extremist views. While antidepressants are problematic in the treatment of bipolar disorder, and should thus be minimized while mood stabilizers are maximized, many bipolar patients with bipolar spectrum disorder may need antidepressants along with mood stabilizer treatment. In this paper, we will use our qualitative clinical experience (especially that of the first author who has been practicing psychopharmacology in the Washington DC area since the late 1960s) to inform a discussion of the concept of the bipolar spectrum from the perspective of the practicing clinician. A focus on clinical trials and randomized data, while useful, is not the purpose of this paper, and will be found in the other articles in this special issue. Many of our observations. 436
Jacob J Katzow^, Douglas J Hsu" and S Nassir Ghaemi" ''George Washington University, Washington, DC, USA, '^Bipolar Disorder Research Program, Cambridge Hospital, Cambridge, MA and Harvard Medical Sohooi, Boston, MA, USA
Key words; antidepressants - bipolar disorder bipolar spectrum - diagnosis - mood stabilizers treatment Received 4 October 2002, revised and accepted for pubiication 17 February 2003 Corresponding author: S Nassir Ghaemi, MD, Cambridge Hcspitai, 1493 Cambridge St., Cambridge, MA 02139, USA. Fax: 617 665 1623. e-mail:
[email protected]
gleaned from intensive patient contact, have not been examined in randomized clinical trials. We hope this clinically oriented paper will help guide researchers to topics that merit such rigorous study, as well as serve as a practical interpretation of these complex issues for the immediate needs of clinicians. The evolution of psychopharmacology practice since the 1960s
From the 1960s to the 1980s, the only proven treatment for bipolar disorder that had received regulatory approval was lithium. Lithium, however, was a generic drug and thus the pharmaceutical industry put a weak effort into marketing lithium and diagnosing bipolar disorder. In contrast, non-generic antipsychotic agents were heavily studied and marketed for the treatment of schizophrenia. It was perhaps not a coincidence that, 25 years ago, researchers identified that schizophrenia was overdiagnosed and bipolar disorder underdiagnosed (1, 2). Schizophrenia, however, was still not a condition seen in outpatient practices and diagnostic habits for depressive and anxiety disorders were generally amorphous until the arrival of the serotonin reuptake inhibitors (SRIs) (3).
The bipolar spectrum: a clinical perspective
Towards the end of the 1980s, a watershed event in the history of modern psychiatric medicine was the arrival of fluoxetine on pharmacy shelves. Prior to fluoxetine, primary care physicians did not play a major role in the treatment of anxiety or depressive disorders. In addition, the eflicacy of older antidepressants, such as tricyclics, was minimized because of their increased side effects as compared with SRIs. With the introduction of the newest generation of SRIs, however, the psychopharmacologic treatment of anxiety and mood disorders skyrocketed and primary care physicians were soon in the vanguard of treatment (3). Roughly around this time, the practice patterns in the United States were also moving steadily towards a managed care environment where primary care physicians were taking on more of the responsibilities of specialists. This intertwining of the newer antidepressants and historical reversion from specialists back to primary care physicians still influences the present practice of psychopharmacology. The type of patients seen by psychopharmacologists today typifies the effect of the abovementioned phenomena. Many patients coming to see a psychopharmacologist will have tried a number of antidepressants, previously provided by another doctor, that have been found to be ineffective. A significant number of these patients, while eventually being diagnosed with some form of bipolar disorder, have been misdiagnosed with unipolar depression or other conditions (4). The ratio of bipolar to unipolar depression
The Epidemiologic Catchment Area (ECA) study reported that mania and hypomania occur in 1.2% of the population over a lifetime, which is roughly one-fourth the prevalence of major depression (5). Since this study, some researchers in bipolar disorder have raised concerns over the accuracy of this 1:4 ratio of unipolar major depression to bipolar disorder. A follow-up study, done by Anthony and colleagues, on the diagnostic validity of the ECA study found poor interrater agreement (kappa values) for Axis I psychiatric diagnoses (the highest kappa value was 0.35, although conventionally acceptable kappa values for epidemiological studies are greater than 0.50). The confusion regarding the ratio of diagnosis between unipolar and bipolar depression is highlighted in other analyses. Goodwin and Jamison's comprehensive review of the topic led to an estimate of a 2:1 ratio of unipolar to bipolar disorder (6), and an epidemiologic study among the Amish reported a ratio of 1:1 (7). Furthermore, while the limita-
tions in methodology still pertain, it is of interest that a reanalysis of the ECA data to include all patients in the bipolar spectrum reported a lifetime prevalence rate of 6.4% (8). We think it is important to emphasize that the ratio of unipolar to bipolar depression is probably about two to one, and not 5-10 to 1. Certainly, one cannot simply assume, as has been the case previously, that bipolar disorder is highly uncommon relative to unipolar depression. The mixed anxiety/depression picture
In our experience, the most common type of patient seen by psychopharmacologists present with various admixtures of anxiety and depression. A significant percentage of this group of patients do not respond well to antidepressants, and ultimately prove to have bipolar illness. Unless the patient has been previously hospitalized with a manic diagnosis, proper diagnosis is very difficult based on the patient's history. It is from this anxiety/depressive group of patients that a subset of patients are eventually diagnosed and treated properly as having bipolar disorder. This problem with diagnosis is due to the ego-dystonic nature of anxiety-depressive symptoms - patient's complain of and desire treatment for such symptoms. But grandiose, euphoric, irritable, or angry emotions are frequently not thought of as pathological, and thus a patient's lack of insight on the manic side of the spectrum can make a proper bipolar diagnosis difficult. Moreover, patients do not have more insight when they are euthymic, rather they are more insightful of past rather than current episodes (9). Studies have shown that insight in bipolar disorder is more impaired than in unipolar major depressive disorders, and equally impaired as compared to schizophrenia (10, 11). Thus, the reliance on patient insight in self-reports on mania most likely contributes to the underdiagnosis of bipolar disorder. Impulsivity disorders
In addition to mixed anxiety/depressive patients, the other group of patients from which a subset of bipolar spectrum diagnoses can be made consists of patients with impulsivity disorders including- substance abuse, borderline personality, bulimia, and attention deficit disorder (ADD). Space precludes a careful discussion of each of these presentations, but a few points need to be emphasized. The impulsivity disorders have many symptoms in common with bipolar disorder. More importantly, patients diagnosed with impulsivity 437
Katzow et al.
The bipolar spectrum
tional forms of bipolar illness may exist. In these forms, spontaneous mania or hypomania do not exist, making the condition diflicult to recognize in a clinical setting. A good way of understanding the bipolar spectrum concept is to use an analogy from general medicine, such as hypertension, that can also be viewed along a continuum. After first ruling out other specific causes, such as kidney disease, we are left with essential hypertension. While it is likely that in the future other specific causes will be discovered, the spectrum from normal to high blood pressure will still exist. The line between normal and abnormal is determined empirically (in the case of blood pressure, currently 140/90 mm Hg), where research has determined that the tradeoffs become better to treat the with anti-hypertensives than to withhold treatment. This line, however, is arbitrary in that the line will change over time as we learn more about hypertension or as new medications are introduced. The above analogy demonstrates how the idea of a spectrum already exists within medicine. It may be that psychiatric disorders may be best understood with the concept of spectra/continua. While present diagnostic frameworks, such as in the DSM-IV, are helpful for researchers to achieve consistency in diagnosis, from a clinical perspective they are often artificial in their diagnostic categories. Instead of specific categories, the idea of a spectrum better fits the facts seen in clinical practice.
The diagnosis of bipolar disorder is further complicated if we are ready to diagnose patients with non-classical bipolar features. From a practical perspective, patients with non-classical symptoms tend to be diagnosed under the current large heterogeneous label of 'major depression', by which clinicians in effect mean unipolar depression. Many of these patients, however, are unresponsive to standard antidepressants, and ultimately prove to have evidence of bipolarity, responding to medications used for classical bipolar illness, like standard mood stabilizers, novel anticonvulsants, or atypical neuroleptic agents. There is no term on which a consensus has developed for this group of patients with nonclassical bipolar features, but 'bipolar spectrum' is probably the most common label used. Bipolarity is a disorder that is characterized by unstable mood and behavior, and any recurrent cycling psychiatric disturbance can be evaluated with regard to being a symptom or form of bipolar disorder. Thus, the hypothesis of the bipolar spectrum is that outside of classical manic-depressive illness, or type I bipolar disorder, less conven-
'Soft' symptoms of bipolarity have been studied for over 2 decades (12-14). A review of six studies carried out since 1978 suggests that broadening the bipolar diagnostic criteria to include other aspects of the bipolar spectrum, such as hypomania and cyclothymia, yields a higher prevalence range (3-8.8%) than previously expected (15). While Baldessarini has emphasized the potential pitfalls for research if the bipolar diagnostic spectrum is broadened (16), this paper outlines several advantages to the concept for the practicing clinician. If we are willing to accept the bipolar spectrum concept, the question arises of how broad the spectrum should be. Clinical data and experience suggest that the continuum is quite large, ranging from bipolar I disorder to cyclothymia (14). One approach, developed by Akiskal and Klerman, is to split the spectrum into many subgroups (types II, III, IV, V, VI, or more) based on specific characteristics of each subgroup (17). Another approach, suggested by Ghaemi and Goodwin, lumps all the non-type I or type II subgroups into one generic label (bipolar spectrum disorder) (18).
disorders who present with common symptoms can have bipolar disorder cither instead of or in conjunction with impulsive conditions. Symptom assessment is often limited and difficult to disentangle, and thus careful attention to other diagnostic validators such as family history, course, and treatment response can help identify the bipolar subset of patients who present with these impulsive features. Antidepressant response, in particular, can be informative. For instance, we frequently observe that the level of substance abuse increases with antidepressant use in patients who are later determined to have bipolar illness. Furthermore, children with ADD symptoms whose parents have bipolar disorder should be carefully assessed, especially as the preadolescent presentation of manic symptoms may be significantly different than the adult criteria used in DSM-IV diagnoses. Persons with borderline personality traits should also be carefully assessed for bipolar criteria and a bipolar diagnosis should be considered. In another situation, patients with bulimia are often so devastated by the condition that their resulting depression often overshadows other manic symptoms or bipolar history. In general, a characteristic of bipolar disorder is impulsive behavior, and thus other diagnoses, including but not limited to a bipolar diagnosis, should be considered in patients with impulsive disorders.
438
The bipolar spectrum: a clinical perspective Mild Moderate Severe
Mania
Hypomania Cylothymia
Subthreshold hypomania
Psychotic
Euthymia Recurrent depression
Fig. 1. The bipolar spectrum.
From the standpoint of the practicing clinician, these categorical labels may be less useful than the simple idea of a smooth continuum, as shown in the figure, ranging from bipolar I disorder to psychotic depression (Fig. 1). Patients can cycle from any two points on the spectrum including (very importantly) cycling just within the depressive range. The Kraepelinian concept of 'manic-depressive, depressed' has not translated well into the concept of bipolar depression because clinicians are forgetting that manic depressive illness can present with just various grades of depression, i.e. as cycling within the depressive pole of the illness. Unfortunately, the term 'bipolar' implies cycling between mania and depression, whereas the key to the illness may be the cycling rather than the occurrence of a manic pole. Shifting away from an emphasis on polarity
The bipolar spectrum concept shifts the emphasis in diagnosis of bipolar disorder away from polarity (e.g. presence of spontaneous manic episodes) and toward other diagnostic validators (course, family history, and antidepressant outcomes). As Goodwin and Jamison emphasized (6), an excessive focus on polarity, as seen in the diagnostic schema of DSM-III and IV, obscures the relationship between bipolar and recurrent forms of unipolar depression leading to a lack of attention to soft symptoms such as family history or antidepressantinduced hypomania. DSM-IV describes hypomania as a major diagnosis that requires absence of 'significant social or occupational dysfunction', and thus hypomania differs from mania based on function rather than symptoms. This creates difficulty in proper diagnosis for clinicians because the term 'significant' in the DSM-IV is vague, making identification of hypomania unreliable (19). Patient histories are often so sparse and untrustworthy that it is almost impossible for the clinician to confidently identify hypomanic episodes. The concept of the bipolar spectrum shifts the clinician's attention to more objective and accurately ob-
tained data, such as family history or response to antidepressant medications. The specific soft signs of the bipolar spectrum have been described in detail elsewhere (12). Clinically, we find the following list of soft symptoms useful: hyperthymic personality, seasonality or light sensitivity, having a high from staying up all night, times of diminished need for sleep accompanied by high energy, atypicality (increased appetite or sleep), any admixture of manic and depressed symptoms, lower age of onset, severity of depression, melancholic features, postpartum reactions, any recurrent out-of-control, impulsive, or reckless behavior, or a family history of any of the above. While Ghaemi and Goodwin have previously suggested specific diagnostic guidelines for bipolar spectrum disorder (18), there is not yet consensus as to which specific symptom or how many symptoms need to be present to diagnose a bipolar spectrum condition. Common sense, however, suggests that the more such signs are present, the more likely the diagnosis of a bipolar spectrum condition. Involving families
While a specialist usually examines patients alone, involving family members in the clinical process can aid in proper diagnosis and treatment. The patient may believe that the symptoms of mania are not unusual: a manic emotion such as irritability can be rationalized as a legitimate reaction to real situations. Family members, however, can often report behavioral symptoms of mania better than the patient can. A study on the prodromal symptoms of mania and depression demonstrated that families reported behavioral symptoms of mania more than twice as frequently as patients (47% vs. 22%) (20). These results are in comparison to findings on unipolar depression, where families and patients reported symptoms at similar rates. Thus, obtaining information on symptoms from family members or another third-party (therapist, nurse, social worker) offsets the concealing effects of patient's impaired insight. 439
Katzow et al. Antidepressant treatment
As highlighted by this special issue, the use of antidepressants in bipolar disorder is controversial. There is some evidence that antidepressants may worsen the long-term course of bipolar illness. Our clinical experience is that negative response to antidepressants is an important feature of" patients with bipolar spectrum conditions. One observed effect to antidepressant treatment in patients with bipolar spectrum illness is an increase in cycling frequency and in the amplitude of highs and lows. In contrast, another common effect of antidepressant use in this group of patients is increased cycling but with milder depressive swings (mitigated depression). This is where the patient, after beginning antidepressant therapy, has an increase in cycling without the most severe part of the depressive cycle. This clinical observation is supported by a study which showed that while antidepressants increased the number of mood episodes per year, the percent of time ill with depression was marginally decreased in patients with type II bipolar disorder (4). In these patients, antidepressants are a mixed blessing, with some benefit but no marked overall improvement. Tolerance, or the 'wear-off' effect, is an important aspect of antidepressant outcomes in patients with bipolar spectrum conditions (21). For example, antidepressant 1 demonstrates strong initial effectiveness, but over time loses its efficacy. The patient is then switched to antidepressant 2, which does not work quite as well as 1 and also wears off more quickly. The patient responds even worse when switched to antidepressant 3. Technically, these patients might be acutely responsive to the antidepressants; however, the antidepressants fail to prevent future mood episodes. A related feature of antidepressant response in these patients is quicker onset of benefit with antidepressants. Although we observe and hear of such an effect from other clinicians, we are unaware of any empirical evidence on this topic. Unlike most patients with unipolar depression, we sometimes observe rapid antidepressant response on the order of hours to days in patients with bipolar disorder. In our experience, this initial good news often heralds later bad news, as these patients seem more likely to develop increased mood cycling on antidepressants, or experience tolerance with eventual depressive relapse. Principies of treatment for the bipolar spectrum
The impact of the antidepressant dilemma can often be minimized with aggressive use of mood 440
stabilizers. We define 'aggressive' mood stabilizer use as routine polypharmacy with two or three mood-stabilizing agents soon after insufficient response with a single mood stabilizer. We would also hope that clinicians and patients make a concerted effort, even in the face of side effects, to keep mood stabilizers in the treatment mix. Too frequently, clinicians will try mood stabilizers for only brief periods, often stopping them because of side effects or apparent inefficacy. On the other hand, many more antidepressants will have been tried for much longer durations of treatment. We would certainly encourage the opposite pattern - brief trials of antidepressants, and multiple long trials of stabilizers. A naturalistic outcome study indicates that with aggressive use of mood stabilizers only 19% of bipolar patients required long-term antidepressant treatment (22). While we do not advocate a prohibition on antidepressant treatment for patients with soft bipolar disorders, we suggest applying the lessons learned from the treatment of classic bipolar disorder, such as emphasizing mood stabilizers over antidepressants to control cycling. Patients with soft bipolar illnesses usually have already tried many antidepressants that resulted in poor outcomes. Regardless of whether or not classic bipolar disorder is present, when features of bipolarity are present, shifting the emphasis from antidepressants to mood stabilizers often proves beneficial. In these patients, it is important to recognize the possibility that by inducing cycling, antidepressants can act as mood destabilizers, serving to counteract the potential benefits of mood stabilizers. When considering the use of antidepressants in treating bipolar depressed patients, we tend to limit their application to cases of severe depression that are unresponsive to multiple mood stabilizers or to situations where rapid control of depressive symptoms is imperative, such as in marked suicidality. In general, a proper mood stabilizer trial should occur in the absence of concurrent antidepressant use; otherwise one cannot be sure that lack of response is due to inefficacy of the mood stabilizer as opposed to destabilizing effects of the antidepressant. In contrast, many patients take antidepressants so continuously that they never receive even one mood stabilizer monotherapy trial. We also recommend using atypical neuroleptics in cases of anxiety or agitation, and novel anticonvulsants as potential mood stabilizers in patients who cannot tolerate or accept lithium, valproate, or carbamazepine.
The bipolar spectrum: a clinical perspective The skeptic's view
We are very aware that this paper will generate skepticism in many quarters. We are not unaware of some important potential criticisms of ideas in this paper, which we would like to discuss here. The ideas expressed here about the bipolar spectrum concept are not meant to be definitive. We do not provide evidence intended to prove the bipolar spectrum concept from the perspective of empirical research. While necessary, such research is still in its relative infancy. Instead, we wished to provide a venue for a description of the contours and relevance of the bipolar spectrum concept from the point of view of clinical experience. It may also be suggested that our perspective will lead to an overdiagnosis of bipolar disorder and that we may be unaware of important political and economic factors that might be driving such overdiagnosis. For instance, it might be argued that interest in bipolar disorder has risen with the marketing of mood-stabilizing medications by pharmaceutical companies, and thus the increased use of the bipolar diagnosis may represent a kind of manufactured need, much like advertising for clothes, rather than a scientitic reality. Such a perspective may indeed have some relevance for some psychiatric disorders (3). However, we do not think that bipolar disorder is one of them. The diagnosis of bipolar disorder is over a century old, and when Kraepelin proposed his very broad approach to manic-depressive illness, pharmaceutical companies did not exist. Further, perhaps the most important treatment for bipolar illness, lithium, has long been generic and very inexpensive. Obviously, there is always a tradeoff in diagnosing anything in medicine; anything can be overdone and there are always risks of false positives (an example is the controversy regarding the age at which to recommend mammography). Frankly, in our opinion, the risks of underdiagnosing bipolar disorder (false negatives) are more serious than overdiagnosing it (false positives), though this is obviously not a rationale for indiscriminate diagnosis of it. Summary
We feel that bipolar disorder, a condition of unstable mood and behavior, is common, but very hard to diagnose. The misdiagnosis and underdiagnosis of bipolar conditions is in part because of the 'soft' symptoms of bipolar spectrum disorder. The bipolar spectrum - although ranging from severe mania on the one end to psychotic depression on the other - focuses our
attention on the less classical presentations of bipolarity. To recognize and therefore treat bipolarity, clinicians should focus on patients who present with two types of conditions- anxiety/depressive illnesses and impulsivity disorders (substance abuse, borderline personality, bulimia, and ADD). For proper treatment of bipolar disorder, it is important to recognize that antidepressants can have a detrimental effect on a patient's condition. This destabilization of bipolar disorder is often manifested by a worsening condition and/or increased cycling. Mood stabilizers, which are generally underutilized, should be emphasized in treatment. To adequately test the efficacy of a mood stabilizer, monotherapy should at some point be attempted. It is important, however, to recognize that while mood stabilizers should be emphasized over antidepressants, effective treatment for difficult cases may require both types of medications. References 1. Pope HG, Jr, Lipinski JF. Diagnosis in schizophrenia and manic-depressive illness. Arch Gen Psychiatry 1978; 35: 811-828. 2. Kendell RE, Brockington IF. The identification of disease entities and the relationship between schizophrenic and affective psychoses. Br J Psychiatry 1980; 137: 324-331. 3. Healy D. The Antidepressant Era. Cambridge, MA: Harvard University Press, 1998. 4. Ghaemi SN, Ko JY, Goodwin FK. The bipolar spectrum and the antidepressant view of the world. J Psychiatr Pract 2001; 7: 287-297. 5. Regier DA, Kaelber CT. The epidemiologic catchment area (ECA) program: studying the prevalence and incidence of psychopathology. In: Tsuang MT, Tohen M, Zahner GEP eds. Textbook in Psychiatric Epidemiology. New York: John Wiley, 1995: 133-157. 6. Goodwin FK, Jamison KR. Manic Depressive Illness. New York: Oxford University Press, 1990. 7. Weissman MM, Leaf PJ, Tischler GL et al. Affective disorders in five United States communities [published erratum appears in Psychol Med 1988 Aug; 18(3): following 792], Psychol Med 1988; 18: 141-53. 8. Judd L, Akiskal H. The prevalence and disability ofbipolar spectrum disorders in the US population: re-analysis of the ECA database taking into account subthreshold cases. J Affect Disord 2003; 73: 123-31. 9. Cuesta M, Peralta V, Zarzuela A. Reappraising insight in psychosis. Multi-scale longitudinal study. Br J Psychiatry 2000; 177: 233-40. 10. Amador XA, Flaum M, Andreasen NC et al. Awareness of illness in schizophrenia and schizoaffective and mood disorders. Arch Gen Psychiatry 1994; 51: 826-836. 11. Ghaemi SN, Stoll AL, Pope HG. Lack of insight in bipolar disorder: the acute manic episode. J Nerv Ment Dis 1995; 183: 464-467. 12. Akiskal HS. The prevalent clinical spectrum of bipolar disorders: beyond DSM-IV. J Clin Psychopharmacol 1996; 16 (Suppl. 1): 4S-14S.
441
Katzow et al. 13. Akiskal HS, Djenderedjian AH, Rosenthal RH, Khani MK. Cyclothymie disorder: validating criteria for inclusion in the bipolar affective group. Am J Psychiatry 1977; 134: 1227-1233. 14. Akiskal HS, Pinto O. The evolving bipolar spectrum. Prototypes I, II, III, and IV. Psychiatr Clin North Am 1999; 22: 517-534, vii. 15. Angst J. The emerging epidemiology of hypomania and bipolar II disorder. J Affect Disord 1998; 50: 143-151. 16. Baldessarini R. A plea for the integrity of the bipolar concept. Bipolar Disord 2000; 2: 3-7. 17. Klerman GL. The spectrum of mania. Comp Psychiatry 1981; 22: 11-20. 18. Ghaemi SN, Ko JY, Goodwin FK. 'Cade's disease' and beyond: Misdiagnosis, antidepressant use, and a proposed definition for bipolar spectrum disorder. Can J Psychiatry 2002; 47: 125-134.
442
19. Gershon ES, Guroff JJ. Information from relatives. Diagnosis of affective disorders. Arch Gen Psychiatry 1984; 41: 173-180. 20. Keitner Gl, Solomon DA, Ryan CE et al. Prodromal and residual symptoms in bipolar I disorder. Comp Psychiatry 1996; 37: 362-367. 21. Ko J, Ghaemi SN, Kontos N, Baldassano C, Goodwin F. Antidepressant wear off and outcomes in bipolar versus unipolar depression. American Psychiatric Assocation Annual Meeting. Philadelphia, PA: American Psychiatric Assocation, 2002. 22. Ghaemi SN, Goodwin FK. Long-term naturalistic treatment of depressive symptoms in bipolar illness with divalproex vs. lithium in the setting of minimal antidepressant use. J Affect Disord 2001; 65: 281-287.