Journal Logo


Bipolar Depression: A Historical Perspective of the Current Concept, with a Focus on Future Research

Martino, Diego J. MD, MSc, PhD; Valerio, Marina P. MD

Author Information
Harvard Review of Psychiatry: 9/10 2021 - Volume 29 - Issue 5 - p 351-360
doi: 10.1097/HRP.0000000000000309
  • Free


Bipolar depression is usually defined as a major depressive episode that occurs in the context of bipolar disorder—that is, in individuals with a prior history of hypo/manic episodes. It has been characterized as “a leading, unsolved, clinical and public-health challenge for contemporary psychiatry.”1 Depression is the most common polarity of illness-onset in both bipolar I and bipolar II disorders, and it is initially misdiagnosed as major depressive disorder (MDD) in as many as 40% of patients.2,3 The prevalence of bipolar depression has increased in recent years when considering different forms of subthreshold bipolarity, some of which (e.g., short-duration hypomanic episodes or those with insufficient symptoms) were included as “other specified bipolar and related disorder” in the fifth edition (DSM-5, 2013) of the Diagnostic and Statistical Manual of Mental Disorders.4 In contrast to manic episodes, for which multiple proven treatments are available, the U.S. Food and Drug Administration has approved only a few medications for managing depressive episodes in bipolar disorder (BD).5 Even after proper diagnosis and adequate maintenance treatment, depression is still associated with substantial morbidity, with different longitudinal studies showing that between 70% and 80% of the time spent ill corresponds to depressive symptomatology.6 Moreover, depressive burden has been associated with psychosocial impairment, decreased longevity, and increased risk of suicidality in people with BD.7–9

Our knowledge of bipolar depression relies on the assumption that it is a homogeneous clinical construct, positioned as the opposite pole of (hypo)mania in a unitary disease entity. In the last century, psychiatry has been undergoing a heated debate over whether there are different categorical types of depression or whether all forms of depressive episodes should be grouped along a continuum of severity. Paradoxically, the literature restricts this debate to MDD—with no attention to BD even though the criteria for major depressive episodes (MDEs) are the same for both disorders. The central thesis of this narrative review is that the current concept of bipolar depression (as defined in DSM-5) is likely not a homogeneous clinical construct and that this heterogeneity could be hindering our knowledge about pathophysiological, clinical, and therapeutic aspects of BD. In the first of our two main sections, the unitary and binary models of unipolar depression (currently MDD) are described, and we highlight, within this framework, the changes in the concept of bipolar depression throughout the twentieth century. In the section after that, we identify some limitations of the current concept of bipolar depression and propose an empirical approach to overcome those limitations.


The following review is not exhaustive but is intended, instead, to highlight some historical aspects of the classification of mood disorders that we consider relevant to clarifying the origin and limitations of the current concept of bipolar depression. To that end, articles published in peer-reviewed, English-language journals between 1900 and 2020 were retrieved from the online databases PubMed and PsycInfo using the terms bipolar or depression and history, historical, classification, review, subtypes, unitary, or binary. The reference lists of the studies identified for inclusion were also reviewed for further relevant reports and textbooks.

Unitary and Binary Models of Unipolar Depression

The classification of unipolar depression during the last century was marked by the controversy between unitary and binary models. Authors like Ballet, Kraepelin, Schneider, and Gillespie proposed binary models involving different categorical types of depression.10–13Melancholia (alternatively referred to by other authors as endogenous, autonomous, or psychotic depression) was characterized by motor and intellectual slowing, absence of emotional response to environmental changes, and an episodic course. By contrast, neurasthenia (alternatively referred to as psychogenic, reactive, or neurotic depression) was a more chronic condition that included pain, fatigue, anxiety, and emotional reactivity to environmental events. In the remainder of this historical review, if no further clarification is made, we will refer to the two subtypes of depression as melancholic/psychotic or non-melancholic/non-psychotic, respectively—which have an approximate correspondence with the current concept of MDE with or without melancholic/psychotic features. Contrarily, Mapother and Lewis14–16 stand out as the main subscribers of the unitary model of depression, according to which these depressive subtypes are not categorically different but represent, instead, a continuum of severity. Thus, for unitary models, compared to non-melancholic/non-psychotic depressions, melancholic/psychotic forms would be just a more severe form of the same disorder.

The debate between unitary and binary models of depression waned between the mid-1930s and 1950s. By 1952, when the first edition of the DSM was published, the binary position had prevailed. Those depressions with melancholic/psychotic features (i.e., “malignant symptoms” in DSM-I such as agitation, delusions, hallucinations, severe guilt feelings, intractable insomnia, severe psychomotor retardation, or profound retardation of thought) were classified within the section on psychotic disorders either as “manic-depressive reaction–depressive type” or “psychotic depressive reaction” (which differed by the absence of repeated episodes in the latter). Conversely, non-melancholic/non-psychotic depressions were classified as “depressive reaction” (i.e., “the anxiety is allayed, and hence partially relieved, by depression and self-depreciation”) within the section on psychoneurotic disorders.

From the mid-1950s to the 1970s, several studies used multivariate statistics, such as cluster analysis or factor analysis, to clarify the unitary or binary nature of depression.17,18 The results were inconclusive. Some authors, such as Eysenck,19 argued that the binary hypothesis was strongly supported. Others, such as Kendell, recognized the need to distinguish between two types of depression (called type A and type B in reference to melancholic/psychotic and non-melancholic/non-psychotic depressions, respectively), although this proposal could be framed in either the dimensional or categorical model.20 Interestingly, several of these studies suggested the possibility of adopting a trinitarian or even a more pluralistic stance.21–23 In fact, a milestone of those years was the introduction in 1957 of the first two antidepressant drugs,24 with the subsequent description of another subset of non-melancholic depressions called “atypical” because of their clinical features (hysterical/anxious features, and inverse vegetative symptoms).25–30

By the 1970s, the attempts to classify depression reached the point of being described as absurd; every possible combination had been proposed at least by some author.20 This diversity of viewpoints played into the hands of those critical of psychiatric nosography, which had been severely questioned both in terms of its reliability and its validity.31,32 Against this background, researchers at the Washington University, based on the validity criteria proposed by Robins and Guze,33 developed the Feighner criteria for a unitary operationalized definition of depression.34 These criteria were based on studies carried out in patients with melancholia and manic-depressive illness, as documented by Kendler and colleagues.35–37 In the 1970s, Akiskal and McKiney38,39 published two influential articles proposing a “common final pathway” hypothesis for all forms of depression, thereby reinforcing a unitary model. A few years later, the monolithic concept of major depressive disorder was described in the Research Diagnostic Criteria40 (RDC) and DSM-III (1980), derived with modest changes from Feighner’s criteria. Thus, a unitary model of unipolar depression based on an expert consensus has prevailed since the late 1970s. Moreover, the several depressive subtypes proposed for future research in the RDC were quickly relegated to a few specifiers: melancholia and psychotic features were included in DSM-III; atypical features, in DSM-IV (1994); and anxious distress and mixed features, more recently in DSM-5 (2013). The key changes in the conceptualization of MDD throughout the twentieth century are summarized in Figure 1.

Figure 1
Figure 1:
Timeline with key changes in the conceptualization of major depressive disorder and bipolar depression through the twentieth century. DSM, Diagnostic and Statistical Manual of Mental Disorders; MAOI, monoamine oxidase inhibitor; MDI, manic-depressive illness; RDC, Research Diagnostic Criteria.

Although Feighner’s criteria were presented as provisional, its construct for depression has remained practically unchanged for more than four decades. Moreover, despite exceptions that attempted to restore melancholia as a different categorical entity, the unitary concept of MDD has dominated clinical practice and research since 1980.46–48 There have nevertheless been strong concerns and criticism about the limitations of this framework in recent years. Briefly, it has been argued that these criteria fail to distinguish between a depressive disorder and intense normal sadness, with a consequent risk of overdiagnosis.49,50 This problem could help to explain the finding of a lifetime prevalence of MDD higher than 40% from prospective studies of healthy subjects aged 18 to 32 years.51 Moreover, the situation could have been exacerbated by the elimination of the bereavement exclusion in DSM-5.52,53 The consistency of the MDD criteria has also been criticized since individuals sharing the same diagnosis can show substantial variations in their symptom profiles.54,55 Furthermore, it has been argued that the heterogeneity of MDD could hinder pathophysiological research and the identification of more effective treatments.53,56–59 A 2018 study—one of the largest on the genetic architecture of the MDD, including 135,458 cases and 344,901 controls—found that “major depression is not a discrete entity at any level of analysis.”60 Likewise, the consistency and validity of the melancholic and atypical specifiers in the successive editions of DSM have also been criticized; hence, the widespread use of the specifiers could, over time, lead to invalid conclusions (e.g., that there are no clinical merits in making the distinction).61–63 Nevertheless, different classifications of depressive subtypes continue to be proposed (including melancholia, clinical depression, psychotic depression, pure depression, and neurotic depression, among others), leading to the paradox raised by Wakefield that MDD is one disorder (such as in DSM-5) but that it is also many disorders.52,64,65 An additional complication is that the initial field trials of the DSM-5 criteria showed a questionable interrater reliability for MDD.66 Therefore, it is fair to say that the validity of the MDD concept is an unresolved concern even now.

Development of the Concept of Bipolar Depression

As the father of modern psychiatry, Kraepelin subsumed all forms of melancholia and mania under the broad construct of manic-depressive insanity.12 The fundamental feature that distinguished this entity was recurrence rather than polarity, including both unipolar and bipolar forms. He contrasted these forms of melancholia with the so-called psychogenic depressions, a not necessarily recurrent phenomenon that had a foundation in response to stressful life events. This view of manic-depressive insanity dominated the scene during the first half of the twentieth century. As mentioned above, in DSM-I (1952), recurrent melancholic/psychotic depressions were considered “manic-depressive reactions” even if not accompanied by mania. Moreover, some of the first studies that proved the anti-recurrence efficacy of lithium included both patients with manic-depressive features and patients with recurrent unipolar melancholic/psychotic depressions.41,42

The broad Kraepelinian structure began to be challenged by the works of Leonhard (1957),43 who classified endogenous psychoses into “manic-depressive” (i.e., melancholic and manic episodes coexisted) and “unipolar recurrent depressive” psychoses. The nosological differentiation between bipolar and unipolar affective disorders was reinforced in the subsequent years by the studies of Perris and Angst (both 1966),44,45 who showed that both entities differed in relation to gender, pattern of symptoms, longitudinal course, and family aggregation. Winokur and colleagues67 published similar results a few years later (1969) in the United States. It is interesting to note the contrast between this situation and the unitary and binary models of unipolar depression mentioned above. While multiple subtypes of unipolar depression were described throughout the twentieth century, bipolar depression was always conceptualized as melancholic/psychotic and recurrent until 1980.68 This conceptualization was consistently followed in the work of Kraepelin, in studies such as those of Leonhard and Perris, and in the pioneering studies on lithium therapy.

The concept of bipolar depression changed substantially when the RDC and DSM-III (1980) classified MDD and BD as two different nosological entities. In fact, the RDC and DSM-III abandoned the traditional concept of bipolar depression as melancholic/psychotic and recurrent. Bipolar depression, similar to MDD, became a unitary concept that included any form of depressive experience, whether recurrent or not, that occurs in patients with mania. This change was made, however, without any empirical evidence showing that the subsets of bipolar depression (e.g., melancholic/psychotic and non-melancholic/non-psychotic) were equivalent in terms of clinical course, disability, family aggregation, response to treatment, or any other relevant diagnostic validator. The key changes in the conceptualization of depression in BD through the twentieth century are summarized in Figure 1.

Although the concept of bipolar depression from DSM-III has remained almost unchanged to now, the prevalence of this condition has increased over the years, as the definition of BD has expanded. The diagnosis of BD type II, included as a diagnostic entity in DSM-IV (1994), requires the lifetime experience of at least one episode of major depression. The same item is required for the diagnosis of some of the specified bipolar and related disorders (e.g., short-duration hypomanic episodes or those with insufficient hypomanic symptoms) included in DSM-5 (2013). Thus, any form of MDE, recurrent or not, would currently be considered “bipolar” if it occurs in a patient who has experienced subthreshold hypomania to mania. While, as outlined above, the validity of MDD has long been subject to various questions and criticisms, the validity of bipolar depression has basically escaped scrutiny. The following section lists some potential limitations that the current concept of major depression could imply for the advancement of our knowledge on BD.


An important risk of subsuming different forms of depressive experience—if these are not etiologically homogeneous—in a broad construct of bipolar depression is that this heterogeneous construct could lead to misconceptions about clinical, pathophysiological, and therapeutic aspects of BD.

For example, some recent studies seem to challenge the classical conception of BD, suggesting that its high heritability could be attributable only to mania and that familial transmission of mania and depression would be independent.69,70 Additionally, two studies in community samples of adolescents and adults showed that manic patients have a nearly equal proportion of comorbidity with MDE as with different anxiety disorders, providing additional support for the consideration of mania as an independent dimension of depression.71,72 These results agree with those from previous reviews (based on DSM diagnoses) suggesting that mania and depression could be two different illnesses.73,74 The etiological dissociation between MDE and mania suggested in these studies, however, has not been assessed for any specific subtype of bipolar depression. It would thus be possible to hypothesize that some depressive subtypes (i.e., melancholic/psychotic) were etiologically related to (hypo)mania, conforming to what has been historically called manic-depressive insanity. By contrast, many other non-melancholic/non-psychotic depressions that occur in people with a history of hypo/mania under conditions of chronic stress or in response to life adversities (particularly in subjects predisposed by high neuroticism or personality disorders) might not be etiologically related to BD. In other words, some non-melancholic MDE might be clinical phenomena comorbid to (hypo)mania, such as has been widely reported for schizophrenia, anxiety disorders or personality disorders, among others. This latter group would be a subsidiary of different types of depression that were originally described in the context of binary models of unipolar depression and then incorporated into the BD orbit after publication of DSM-III. In the context of the proposed hypothesis, it would be expected that the different subtypes of MDE differ in the familial aggregation of BD, which could be the focus of future studies.

If indeed some depressive episodes were not etiologically related to (hypo)mania, the current concept of bipolar depression might predispose to overdiagnosis of BD. This particularly concerns subjects with threshold and subthreshold hypomania, which together showed a prevalence of 20%–30% in prospective studies of healthy subjects and have a relatively low interrater reliability.66,75,76 In fact, the expansion of BD diagnostic criteria, including the hypomania threshold in DSM-IV (1994) and subthreshold in DSM-5 (2013), was done without considering the limitations in the validity and reliability of the MDE. If based on the limitations mentioned above, the MDD (and therefore MDE) is considered a not very valid and reliable nosological construct, it could be conceptually questionable to suppose that the antecedent of a few symptoms of hypomania makes it an acceptable diagnosis of subthreshold BD.4,75–77 Similarly, a few symptoms of concomitant hypomania in an MDE should also not make it a valid and reliable diagnosis of mixed depression, one that is frequently ascribed to BD.78,79 Therefore, these concepts should be revised if MDEs are not etiologically homogeneous in BD. In MDEs etiologically related to BD, subthreshold hypomania (both prior or concomitant) might be a useful predictor for early diagnosis. By contrast, in MDEs not etiologically related to BD, subthreshold hypomania might be only a marker of general psychopathology (i.e., in personality disorders, ADHD, substance use disorders, other impulse-control and conduct disorders) and predispose to overdiagnosis. Here again, empirical research would be necessary to determine whether hypomania (particularly subthreshold) is equally predictive of bipolarity in all depressive subtypes.

Another potential unintended consequence of the current broad concept could be the overestimation of the long-term morbidity and disability attributed to bipolar depression. Longitudinal studies show that 70%–80% of the time spent ill in BD involves depressive morbidity (mostly subsyndromal), which is often interpreted as the “natural history” of the disease.6 However, this interpretation does not consider several other factors overrepresented in BD patients that could both contribute to the subsyndromal depressive symptoms and impact on psychosocial functioning: substance-induced mood changes, side effects of some medications, comorbid medical conditions, mood fluctuations secondary to adverse life situations (e.g., job loss, family conflict, trauma), or personality disturbances (such as high neuroticism or personality disorders).80 Therefore, future studies controlling for the effect of these other variables would be necessary before concluding that all depressive symptoms and disability are inherent to BD.

Finally, these issues affect the interpretation and design of clinical trials. Enrolling patients using the current concept of bipolar depression could obscure the differences in efficacy of therapeutic interventions in specific subtypes of depression. There is some evidence of differential response to antidepressants and lithium between melancholic and non-melancholic MDD, but this type of research was infrequently conducted in bipolar depression.81,82 Melancholic features appeared as a predictor of good response in a pooled analysis of placebo-controlled studies on the efficacy of olanzapine for acute bipolar depression.83 Additionally, in a recent reanalysis of a previous controlled trial of lamotrigine for bipolar II depression, the odds of treatment response were higher for the melancholic subgroup than for their non-melancholic counterpart.84 On the other hand, the high depressive morbidity along the course of BD is usually attributed to limitations of current available treatments and problems of nonadherence.6 However, as mentioned above, part of this symptomatology could be accounted for by several conditions not necessarily sensitive to bipolar therapeutic algorithms. Therefore, future studies comparing therapeutic strategies within different bipolar depression subtypes could improve the detection of efficacy of the available treatments.


The potential limitations mentioned above must be contextualized in recent concerns about the impossibility of improving our knowledge with current nosology.85,86 Although there is evidence of structural and functional alterations in neuroimaging of BD, as well as changes in several cascades involving neurotrophic factors, cytokines, oxidative stress molecules, or mitochondrial function, it is difficult to establish which findings are causative or secondary to the disorder or treatment.87,88 Likewise, these neuroimaging and laboratory biomarkers tend to be somewhat inconsistent across studies and show overlap with those described in other neuropsychiatric disorders such as MDD, schizophrenia, and Alzheimer’s disease.87,88 Findings of the genome-wide association studies of BD also showed some inconsistency and high correlation with other mood and psychotic disorders.89,90 Overall, the last few decades show that the neurobiology and genetics of mental disorders do not adhere to the boundaries delineated by current psychiatric nosology. In order to overcome some of these questions, the National Institute of Mental Health proposed to replace the DSM diagnosis with another framework for research: the Research Domain Criteria.91 Although this approach could generate an interesting data matrix in the medium term, it received some criticism for moving excessively away from clinical research.86,92 Dimensional modeling has been proposed as another alternative way to overcome the limitations of the categorical definitions of BD.85,93 In this approach the symptoms commonly presented in mood disorders are grouped into psychopathological dimensions (i.e., activation, depression, and psychosis) or functional domains (i.e. activity, cognition, and emotion) rather than being clustered as clinical diagnoses. Furthermore, it has recently been hypothesized that both biomarkers and genetic substrate could be more closely related to these psychopathological dimensions than to the DSM diagnostic categories.90,94 However, these hypotheses could collide with the same limitations as the DSM categorical diagnoses, insofar as the psychopathological dimensions (i.e., depression) were also based on arbitrary criteria rather than on clinical research as we have reviewed here. In other words, if the current concept of unipolar or bipolar depression encompasses different forms of depressive episode not etiologically homogeneous, the proposed pathophysiological dimensions (e.g., depression) could be as fallible as the DSM categorical diagnoses.

An alternative approach to advance the clinical and research field would be to delineate a phenotype of bipolar depression based on empirical data. As shown throughout our review, the current concept of bipolar depression emerged from a consensus of experts in the late 1970s, subsuming in a broad construct any type of depressive episode. Although successive editions of the DSM conceptualized bipolar depression as a homogeneous clinical construct, there is no empirical evidence to consider this assumption as valid. Therefore, a critical first step would be to test the validity of the current concept of bipolar depression using the traditional criteria proposed by Robins and Guze.33 For this purpose, the different presumptive subtypes of bipolar depression would be compared by external validators such as clinical course, family aggregation, disability, response to treatment, or laboratory findings, among others. The potential subsets of bipolar depression are diverse. We use the concepts of specifier and subtype interchangeably here, since discussion of their nuances is beyond the scope of this manuscript. DSM-5 describes the same specifiers for MDD and bipolar depression, although other authors have proposed alternative models of unipolar depression subtypes that could also be considered for BD.64,65 As mentioned above, melancholia—which has been traditionally ascribed to manic-depressive insanity and whose core clinical features show some parallelism with those of mania (Text Box 1)95,96—would be an interesting target to be compared with other non-melancholic depressions in terms of external validators in BD. These comparisons could be conducted through research specifically designed for this purpose or by stratifying results from studies using the current concept of bipolar depression by subtype. Likewise, the analyses can be done by comparing specific subtypes to each other (e.g., melancholic vs. atypical vs. others) or by comparing one subtype to all other MDE subtypes combined (e.g., melancholic vs. non-melancholic or psychotic vs. non-psychotic).

Text Box 1

Parallelism of Core Features of Melancholia and Mania

Melancholia Mania
1. Core clinical features (Martino et al., 2019):95
– Decreased mood reactivity
– Motor retardation
– Retardation of speech and thought
2. Frequent mood-congruent overvalued (or delusional) ideas
3. Usually adequate premorbid adjustment
4. Phasic rather than chronic clinical course
5. Male:female relationship close to unity
6. Lithium treatment effectiveness reports
1. Core clinical features (Martino et al., 2020):96
– Increased mood reactivity
– Increased motor activity
– Increased speech and thought
2. Frequent mood-congruent overvalued (or delusional) ideas
3. Usually adequate premorbid adjustment
4. Phasic rather than chronic clinical course
5. Male:female ratio close to unity
6. Lithium treatment effectiveness reports

A first methodological challenge in this clinical research program would be to identify the subtypes of depression in the context of BD. The consensus is that a psychotic depression would be defined by the presence of delusions or hallucinations during the episode of clinical depression. The delimitation of the other depressive subtypes, however, could be more problematic. For example, the DSM specifier of melancholia has been criticized for its diagnostic overlap with MDE criteria.97 Likewise, a recent review showed discrepancies between the clinical items of the DSM-5 specifier and those derived from empirical data for the identification of melancholia.95 Therefore, even if melancholic and non-melancholic bipolar depression were categorically different, the use of the DSM-5 criteria could limit the ability to find differences in external validators between both conditions. Some alternative diagnostic tools based on clinical features alone (e.g., psychomotor disturbances in CORE system) or combined with illness correlates (e.g., Sydney Melancholia Prototype Index) have shown high discrimination of melancholic and non-melancholic episodes in both unipolar and bipolar depressions.48,98 Similar criticisms were made on the DSM’s atypical or mixed feature specifiers.63,99 So, the use of a polydiagnostic approach (including DSM-5 definitions and other alternatives) could be useful to control for the variability in identifying subtypes relating to differences among diagnostic schedules.

A second methodological challenge is how to weigh multiple validators of bipolarity—derived from studies comparing bipolar and unipolar depression—when comparing presumptively different bipolar depression subtypes.100,101 That is, if significant differences emerged between subtypes of bipolar depression in external validators, what would be the parameters that would indicate a closer etiological relationship with BD? The history of BD in first-degree relatives or (hypo)mania induced by antidepressants could be considered the most robust indicators that a subtype of depression is etiologically related to BD. Both would be more reliable in cases of mania than hypomania, and when they are not based exclusively on retrospective patient reports. Consideration should also be given, however, to the possibility that the switch induced by antidepressants depends more on the subject’s predisposition to mania than on the depression subtype. A reduction in the number of depressive recurrences during maintenance treatment with medications with a proven prophylactic efficacy for (hypo)mania (e.g., lithium) could also indicate an etiological link between a depressive subtype and BD. Furthermore, a phasic rather than chronic course and a similarity in gender distribution (male:female ratio), which are attributes traditionally ascribed to the manic-depressive insanity, could suggest a closer relationship of a depressive subtype with BD. Other frequently used bipolarity predictors such as clinical course (e.g., early age of onset, higher number of recurrences, brief episodes) or therapeutic response characteristics (e.g., lack of response or “wearing-off” when receiving antidepressants) should be used with caution since they might only be indicating underlying psychiatric comorbidities (e.g., anxiety or personality disorders). Therefore, comorbidities should be controlled before concluding that any of these validators are considered as indicators of bipolarity. Marked differences between MDE subtypes in external validators would challenge the current concept of bipolar depression. In such a case, the new, emerging concept of bipolar depression should then be subjected to more sophisticated neuroimaging/laboratory studies and genetic analysis. A summary of the proposed research program is shown in Figure 2.

Figure 2
Figure 2:
Clinical research program to validate the current concept of bipolar depression.

A recent JAMA Psychiatry editorial states: “Almost every clinician that has treated patients with major depressive disorder will boldly proclaim that depression is not a single illness with a common causative mechanism, pathophysiology, prognosis, or response to treatment.”102 As shown above, there is no reason to suppose that this heterogeneity is less of an issue in bipolar depression as it is currently defined, and each attentive physician would judge as much in his or her own clinical practice. By the same token, there is no reason to refrain from addressing this serious deficiency regarding our understanding of BD. Beyond the preeminence of DSM categories in our clinical practice and research programs, we need further clinical studies to develop diagnostic entities based on empirical data. Such diagnostic entities might be more suitable for subsequent validation by laboratory or neuroimaging biomarkers, and they could also lead to more specific and effective treatments.

Declaration of interest: The authors report no conflicts of interest. The authors alone are responsible for the content and writing of the article.

The authors would like to thank Drs. Gordon Parker and Nassir Ghaemi for their valuable comments, suggestions, and criticism on the content of this manuscript.


1. Baldessarini RJ, Vieta E, Calabrese JR, Tohen M, Bowden CL. Bipolar depression: overview and commentary. Harv Rev Psychiatry 2010;18:143–57.
2. Shen H, Zhang L, Xu C, Zhu J, Chen M, Fang Y. Analysis of misdiagnosis of bipolar disorder in an outpatient setting. Shanghai Arch Psychiatry 2018;30:93–101.
3. Stensland MD, Schultz JF, Frytak JR. Diagnosis of unipolar depression following initial identification of bipolar disorder: common and costly misdiagnosis. J Clin Psychiatry 2008;69:749–58.
4. Merikangas KR, Jin R, He JP, et al. Prevalence and correlates of bipolar spectrum disorder in the world mental health survey initiative. Arch Gen Psychiatry 2011;68:241–51.
5. Vieta E, Berk M, Schulze TG, et al. Bipolar disorders. Nat Rev Dis Primers 2018;4:18008.
6. Forte A, Baldessarini RJ, Tondo L, Vázquez GH, Pompili M, Girardi P. Long-term morbidity in bipolar-I, bipolar-II, and unipolar major depressive disorders. J Affect Disord 2015;178:71–8.
7. Martino DJ, Marengo E, Igoa A, et al. Neurocognitive and symptomatic predictors of functional outcome in bipolar disorders: a prospective 1 year follow-up study. J Affect Disord 2009;116:37–42.
8. Dickerson F, Origoni A, Schroeder J, et al. Natural cause mortality in persons with serious mental illness. Acta Psychiatr Scand 2018;137:371–9.
9. Baldessarini RJ, Tondo L, Pinna N, Nuñez GH, Vázquez GH. Suicidal risk factors in major affective disorders. Br J Psychiatry 2019;215:621–6.
10. Ballet G. Diagnostique différentiel, clinique et nosologique de la mélancholie. Bull Med 1911;25:157–60.
11. Kraepelin E. Psychiatrie: Ein Lehrbuch für Studirende und Artze. 8th ed.Leipzig: J.A. Barth, 1913.
12. Schneider K. Die Schichtung des emotionale, Lebens und der Aufbau der Depressianzustande. Z Ges Neurol Psychiat 1920;58:281.
13. Gillespie RD. The clinical differentiation of types of depression. Guy’s Hosp Rep 1929;3:306–44.
14. Mapother E. Discussion on manic-depressive psychosis. Br Med J 1926;2:872–9.
15. Lewis AJ. Melancholia: an historical review. J Mental Sci 1934;80:1–42.
16. Lewis AJ. Melancholia: prognostic studies and case material. J Mental Sci 1936;82:488–558.
17. Mendels J, Cochrane C. The nosology of depression: the endogenous-reactive concept. Am J Psychiatry 1968;124:suppl:1–11.
18. Nelson JC, Charney DS. The symptoms of major depressive illness. Am J Psychiatry 1981;138:1–13.
19. Eysenck HJ. The classification of depressive illness. Br J Psychiatry 1970;117:241–50.
20. Kendell RE. The classification of depressions: a review of contemporary confusion. Br J Psychiatry 1976;129:15–28.
21. Paykel ES. Classification of depressed patients: a cluster analysis derived grouping. Br J Psychiatry 1971;118:275–88.
22. Overall JE. The Brief Psychiatric Rating Scale in psychopharmacology research. Mod Trends Pharmacopsychiatry 1974;7:67–78.
23. Raskin A, Crook TH. The endogenous-neurotic distinction as a predictor of response to antidepressant drugs. Psychol Med 1976;6:59–70.
24. López-Muñoz F, Alamo C. Monoaminergic neurotransmission: the history of the discovery of antidepressants from 1950s until today. Curr Pharm Des 2009;15:1563–86.
25. West ED, Dally PJ. Effects of iproniazid on depressive syndromes. Br Med J 1959;1:1491–4.
26. Sargant W. Some newer drugs in the treatment of depression and their relation to other somatic treatments. Psychosomatics 1960;1:14–17.
27. Sargant W. Drugs in the treatment of depression. Br Med J 1961;1:225–7.
28. Klein DF, Davis JM. Diagnosis and drug treatment of psychiatric disorders. Baltimore, MD: Williams & Wilkins, 1969.
29. Pollitt JD, Young J. Anxiety state or masked depression? A study based on the action of monoamine oxidase inhibitors. Br J Psychiatry 1971;119:143–9.
30. Quitkin F, Rifkin A, Klein DF. Monoamine oxidase inhibitors. A review of antidepressant effectiveness. Arch Gen Psychiatry 1979;36:749–60.
31. Kendell RE, Cooper JE, Gourlay AJ, Copeland JR, Sharpe L, Gurland BJ. Diagnostic criteria of American and British psychiatrists. Arch Gen Psychiatry 1971;25:123–30.
32. Rosenhan DL. On being sane in insane places. Science 1973;179:250–8.
33. Robins E, Guze SB. Establishment of diagnostic validity in psychiatric illness: its aplicattion to schizophrenia. Am J Psychiatry 1970;126:107–11.
34. Feighner JP, Robins E, Guze SB, Woodruff RA Jr, Winokur G, Munoz R. Diagnostic criteria for use in psychiatric research. Arch Gen Psychiatry 1972;26:57–63.
35. Stone TT, Burris BC. Melancholia: clinical study of 50 selected cases. JAMA 1950;142:165–8.
36. Cassidy WL, Flanagan NB, Spellman M, Cohen ME. Clinical observations in manic-depressive disease: a quantitative study of one hundred manic-depressive patients and fifty medically sick controls. JAMA 1957;164:1535–46.
37. Kendler KS, Muñoz RA, Murphy G. The development of the Feighner criteria: a historical perspective. Am J Psychiatry 2010;167:134–42.
38. Akiskal HS, McKinney WT. Depressive disorders: toward a unified hypothesis: clinical, experimental, genetic, biochemical, and neurophysiological data are integrated. Science 1973;182:20–9.
39. Akiskal HS, McKinney WT. Overview of recent research in depression: integration of ten conceptual models into a comprehensive clinical frame. Arch Gen Psychiatry 1975;32:285–305.
40. Spitzer RL, Endicott J, Robins E. Research diagnostic criteria: rationale and reliability. Arch Gen Psychiatry 1978;35:773–82.
41. Baastrup PC, Schou M. Lithium as a prophylactic agent. Its effect against recurrent depressions and manic-depressive psychosis. Arch Gen Psychiatry 1967;16:162–72.
42. Schou M. Lithium in psychiatric therapy and prophylaxis. J Psychiatry Res 1968;6:67–95.
43. Leonhard K. Classication of endogenous psychoses. New York: Irvington, 1957.
44. Perris C. A study of bipolar (manic-depressive) and unipolar recurrent depressive psychoses. Acta Psychiatrica Scandinavica Suppl 1966;194:1–188.
45. Angst J. On the etiology and nosology of endogenous depressive psychoses. A genetic, sociologic, and clinical study. Monogr Gesamtgeb Neurol Psychiatr 1966;112:1–118.
46. Carroll BJ. The dexamethasone suppression test for melancholia. Br J Psychiatry 1982;140:292–304.
47. Fink M. Neuroendocrine aspects of the relief of melancholia by induced seizures (electroconvulsive treatment). Adv Biochem Psychopharmacol 1984;39:345–57.
48. Parker G, Hadzi-Pavlovic D, Boyce P, et al. Classifying depression by mental state signs. Br J Psychiatry 1990;157:55–65.
49. Wakefield JC, Schmitz MF. When does depression become a disorder? Using recurrence rates to evaluate the validity of proposed changes in major depression diagnostic thresholds. World Psychiatry 2013;12:44–52.
50. Parker G. Is depression overdiagnosed?Yes. BMJ 2007;335:328.
51. Moffitt TE, Caspi A, Taylor A, et al. How common are common mental disorders? Evidence that lifetime prevalence rates are doubled by prospective versus retrospective ascertainment. Psychol Med 2010;40:899–909.
52. Wakefield JC. The DSM-5 debate over the bereavement exclusion: psychiatric diagnosis and the future of empirically supported treatment. Clin Psychol Rev 2013;33:825–45.
53. Parker G. The DSM-5 classification of mood disorders: some fallacies and fault lines. Acta Psychiatr Scand 2014;129:404–9.
54. Lux V, Kendler KS. Deconstructing major depression: a validation study of the DSM-IV symptomatic criteria. Psychol Med 2010;40:1679–90.
55. Fried EI, Nesse RM. Depression is not a consistent syndrome: an investigation of unique symptom patterns in the STAR*D study. J Affect Disord 2015;172:96–102.
56. Shorter E. The doctrine of the two depressions in historical perspective. Acta Psychiatr Scand Suppl 2007;433:5–13.
57. Ghaemi SN, Vöhringer PA. The heterogeneity of depression: an old debate renewed. Acta Psychiatr Scand 2011;124:497.
58. Ghaemi SN. Why antidepressants are not antidepressants: STEP-BD, STAR*D, and the return of neurotic depression. Bipolar Disord 2008;10:957–68.
59. Østergaard SD, Jensen SOW, Bech P. The heterogeneity of the depressive syndrome: when numbers get serious. Acta Psychiatr Scand 2011;124:495–6.
60. Wray NR, Ripke S, Mattheisen M, et al; Major Depressive Disorder Working Group of the Psychiatric Genomics Consortium. Genome-wide association analyses identify 44 risk variants and refine the genetic architecture of major depression. Nat Genet 2018;50:668–81.
61. Maier W, Philipp M. Construct validity of the DSM-III and RDC classification of melancholia (endogenous depression). J Psychiatr Res 1986;20:289–99.
62. Young M, Scheftner W, Klerman G, Andreasen N, Hirschfeld R. The endogenous sub-type of depression: a study of its internal construct validity. Br J Psychiatry 1986;148:257–67.
63. Parker G, Roy K, Mitchell P, Wilhelm K, Malhi G, Hadzi-Pavlovic D. Atypical depression: a reappraisal. Am J Psychiatry 2002;159:1470–9.
64. Parker G. Classifying clinical depression: an operational proposal. Acta Psychiatr Scand 2011;123:314–6.
65. Ghaemi SN, Vöhringer PA, Vergne DE. The varieties of depressive experience: diagnosing mood disorders. Psychiatr Clin North Am 2012;35:73–86.
66. Freedman R, Lewis DA, Michels R, et al. The initial field trials of DSM-5: new blooms and old thorns. Am J Psychiatry 2013;170:1–5.
67. Winokur G, Clayton PJ, Reich T. Manic depressive illness. St. Louis: Mosby, 1969.
68. Carroll BJ, Feinberg M, Greden JF, et al. Diagnosis of endogenous depression. Comparison of clinical, research and neuroendocrine criteria. J Affect Disord 1980;2:177–194.
69. Merikangas KR, Cui L, Heaton L, et al. Independence of familial transmission of mania and depression: results of the NIMH family study of affective spectrum disorders. Mol Psychiatry 2014;19:214–9.
70. Vandeleur CL, Merikangas KR, Strippoli MP, Castelao E, Preisig M. Specificity of psychosis, mania and major depression in a contemporary family study. Mol Psychiatry 2014;19:209–13.
71. Merikangas KR, Cui L, Kattan G, Carlson GA, Youngstrom EA, Angst J. Mania with and without depression in a community sample of US adolescents. Arch Gen Psychiatry 2012;69:943–51.
72. Olfson M, Mojtabai R, Merikangas KR, et al. Reexamining associations between mania, depression, anxiety and substance use disorders: results from a prospective national cohort. Mol Psychiatry 2017;22:235–41.
73. Joffe RT, Young LT, MacQueen GM. A two-illness model of bipolar disorder. Bipolar Disord 1999;1:25–30.
74. Cuellar AK, Johnson SL, Winters R. Distinctions between bipolar and unipolar depression. Clin Psychol Rev 2005;25:307–39.
75. Angst J, Gamma A, Benazzi F, Ajdacic V, Eich D, Rössler W. Toward a re-definition of subthreshold bipolarity: epidemiology and proposed criteria for bipolar-II, minor bipolar disorders and hypomania. J Affect Disord 2003;73:133–46.
76. Zimmermann P, Brückl T, Nocon A, et al. Heterogeneity of DSM-IV major depressive disorder as a consequence of subthreshold bipolarity. Arch Gen Psychiatry 2009;66:1341–52.
77. Benazzi F, Akiskal HS. Refining the evaluation of bipolar II: beyond the strict SCID–CV guidelines for hypomania. J Affect Disord 2003;73:33–8.
78. Benazzi F. Depressive mixed state: testing different definitions. Psychiatry Clin Neurosci 2001;55:647–52.
79. Benazzi F. Depressive mixed state frequency: age/gender effects. Psychiatry Clin Neurosci 2002;56:537–43.
80. Schweitzer I, Maguire K, Ng Ch. Should bipolar disorder be viewed as manic disorder? Implications for bipolar depression. Bipolar Disord 2005;7:418–23.
81. Valerio MP, Szmulewicz AG, Martino DJ. A quantitative review on outcome-to-antidepressants in melancholic unipolar depression. Psychiatry Res 2018;265:100–10.
82. Valerio MP, Martino DJ. Differential response to lithium between melancholic and non-melancholic unipolar depression. Psychiatry Res 2018;269:183–4.
83. Tohen M, Katagiri H, Fujikoshi S, Kanba S. Efficacy of olanzapine monotherapy in acute bipolar depression: a pooled analysis of controlled studies. J Affect Disord 2013;149:196–201.
84. Peters EM, Bowen R, Balbuena L. Melancholic symptoms in bipolar II depression and responsiveness to lamotrigine in an exploratory pilot study. J Clin Psychopharmacol 2018;38:509–12.
85. Malhi GS, Irwin L, Hamilton A, et al. Modelling mood disorders: an ACE solution?Bipolar Disord 2018;(suppl 2):4–16.
86. Ghaemi SN. After the failure of DSM: clinical research on psychiatric diagnosis. World Psychiatry 2018;17:301–2.
87. Harrison PJ, Geddes JR, Tunbridge EM. The emerging neurobiology of bipolar disorder. Trends Neurosci 2018;41:18–30.
88. Teixeira AL, Barbosa IG, Machado-Vieira R, Rizzo LB, Wieck A, Bauer ME. Novel biomarkers for bipolar disorder. Expert Opin Med Diagn 2013;7:147–59.
89. Stahl EA, Breen G, Forstner AJ, et al; Bipolar Disorder Working Group of the Psychiatric Genomics Consortium. Genome-wide association study identifies 30 loci associated with bipolar disorder. Nat Genet 2019;51:793–803.
90. Charney AW, Mullins N, Park YJ, Xu J. On the diagnostic and neurobiological origins of bipolar disorder. Transl Psychiatry 2020;10:118.
91. National Institute of Mental Health. Research Domain Criteria (RDoC). 2013.
92. Malhi GS, Bell E. Is RDoC our Doc? The clinical translation of scientific inquiry. Bipolar Disord 2019;21:478–82.
93. Hickie IB. Evidence for separate inheritance of mania and depression challenges current concepts of bipolar mood disorder. Mol Psychiatry 2014;19:153–5.
94. Reynolds CF. Optimizing personalized management of depression: the importance of real-world contexts and the need for a new convergence paradigm in mental health. World Psychiatry 2020;19:266–8.
95. Martino DJ, Szmulewicz AG, Valerio MP, Parker G. Melancholia: an attempt at definition based on a review of empirical data. J Nerv Ment Dis 2019;207:792–8.
96. Martino DJ, Vlerio MP, Parker G. The structure of mania: an overview of factorial analysis studies. Eur Psychiatry 2020;63:e10.
97. Parker G, Bassett D, Outhred T, et al. Defining melancholia: a core mood disorder. Bipolar Disord 2017;19:235–7.
98. Parker G, McCraw S, Blanch B, Hadzi-Pavlovic D, Synnott H, Rees AM. Discriminating melancholic and non-melancholic depression by prototypic clinical features. J Affect Disord 2013;144:199–207.
99. Koukopoulos A, Sani G, Ghaemi SN. Mixed features of depression: why DSM-5 is wrong (and so was DSM-IV). Br J Psychiatry 2013;203:3–5.
100. 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;47:125–34.
101. Mitchell PB, Goodwin GM, Johnson GF, Hirschfeld RM. Diagnostic guidelines for bipolar depression: a probabilistic approach. Bipolar Disord 2008;10:144–52.
102. Sanacora G. Is this where we stand after decades of research to develop more personalized treatments for depression?JAMA Psychiatry 2020;77:E560–1.

bipolar disorder; depression; melancholia; nosology

© 2021 President and Fellows of Harvard College