In Part 3 of this 3-part series, Dr Dilsaver discusses dichotomization versus a continuum model and concerns that bipolar disorders are over-diagnosed.
[Editor's Note: This is Part 3 of a 3-part series. Click here for Part 1. Click here for Part 2]
Is the unipolar/bipolar dichotomy the most useful and conceptually sound way of viewing the disorders of mood? Or do the disorders, that we currently regard as unipolar and bipolar fall along a continuum? The continuum model is founded on the concept that these disorder are not distinct. This model reflects the view that at one end of the continuum there are major depressive episodes (MDEs) akin to those regarded to exist in the context of unipolar disorder and at the other extreme there are episodes of depression conforming to the concept of mixed hypomania and mixed episodes. Just as the term “continuum” implies, the model sets forth the tenet that there are depressive syndromes that are intermediate in form-and fall somewhere along the continuum.
The unipolar/bipolar dichotomy
Benazzi1 reviewed the literature on the unipolar/bipolar dichotomy. More than 100 articles met the criteria for inclusion in the review. He identified 10 reasons that suggest the dichotomy model is subject to question, only 2 of which are highlighted here.
First, he observed that opposite polarity symptoms are very common in the episodes of depression that occur in the course of “unipolar depression.” An alternative way of stating this point is that symptoms of hypomania are very common in MDEs among those classified as having unipolar disorder, if one only seeks to detect them. This is incongruent with the findings that 87.5% of 131 patients who met the criteria for MDE were in the midst of mixed depression, mixed hypomania, or a mixed episode.2
Second, Benazzi's statistical analysis indicated that there is not a bimodal distribution of the symptoms that distinguish unipolar depression from hypomania. This supports a spectrum concept, because if the two disorders were distinct, one would expect to find a bimodal distribution of the symptom complexes defining an MDE and hypomania.
In contrast, Benazzi’s statistical analysis suggests that unipolar depression and bipolar I disorder (BID) are categorically distinct. However, difficulty arises if one uses this finding to justify the view that unipolar and bipolar disorders are distinct phenomena. Should one choose to defend the unipolar/bipolar dichotomy on the basis of this finding, it follows that BID and bipolar II disorder (BIID) do not logically fall into the same category (ie, the accepted view that both are bipolar disorders must be rejected).
Whether the unipolar/bipolar dichotomy is valid can be determined using criteria that were set forth 4 decades ago by Robins and Guze.3 These criteria propose that diagnostic category be validated on the basis of the consideration of phenomenology, natural history of illness, treatment responsiveness, biological characteristics (laboratory findings), and genetic data. It is highly doubtful they had a spectrum phenomenon or the power of molecular genetics in mind. It will be quite interesting to see, given the growing interest in the concept of the bipolar spectrum, the role that the latter plays in addressing the validity of the unipolar/bipolar dichotomy over the next decade and, perhaps more realistically, the coming decades. In the interim, we have a growing body of clinical findings to rely on.
Concerns about the overdiagnosis of bipolar disorder
The data presented in this 3-part series support an extensive expansion of the scope of the bipolar spectrum. This expansion would have far-reaching consequences for clinical care. It is therefore important to thoughtfully acknowledge the concerns of those who hold that bipolar disorders are now being overdiagnosed.
Perhaps this concern has been expressed no less strongly by investigators at Brown University than elsewhere. For more than a decade, a group of researchers in the Department of Psychiatry and Human Behavior at Brown University have been working on the Rhode Island Methods to Improve Diagnostic Assessment and Services (MIDAS) project. The group has established a clinical laboratory in the context of an outpatient private practice in which a semistructured diagnostic interview is used to evaluate all patients.4 They have concluded that there is widespread overdiagnosis of bipolar disorders.
Seven hundred participants in the MIDAS project were asked whether they had ever received a diagnosis of bipolar disorder or manic-depressive illness by a health care professional; 20.7% of the sample reported having received such a diagnosis.5 On the basis of results of interviews using the Structured Clinical Interview for DSM-IV (SCID), the researchers found that only 3.7% of the patients met the criteria for bipolar I disorder; 5.9% of the patients met the criteria for BID; 3.0% of the patients met criteria for bipolar disorder, not otherwise specified (NOS); and 0.3% of the patients met the criteria for cyclothymia.4
There was a statistically significant difference between the number of subjects who previously received a diagnosis of a bipolar disorder and the number needed to warrant such a diagnosis based on the results of the SCID. Patients in the overdiagnosed group had a broad array of disorders including, but not limited to, panic disorder, social phobia, generalized anxiety disorder, posttraumatic stress disorder, substance use disorders, and borderline personality disorder.5,6
The MIDAS researchers used family history data in an effort to validate their findings. The morbid risk of bipolar disorder among first-degree relatives of patients who were verified as having a bipolar disorder compared with patients in the overdiagnosed group was 8% and 3.5%, respectively. The authors concluded that the data gleaned from the patients’ family histories validated their diagnostic practices.
Many of those who express concern over the possibility that bipolar disorders are now being overdiagnosed tacitly base their position on the assumption that the categories specified in DSM-IV, a document published 17 years ago, are definitive. In reality, the DSMs are an evolving series of documents, and the diagnostic criteria for a given disorder are subject to change over time based on scientific advances.
If one assumes that the DSM-IV criteria for the bipolar disorders capture the scope of bipolarity, DSM-IV is the authoritative source of information for diagnostic practices, and adherence to the algorithms set forth in instruments (such as SCID) designed to identify DSM-IV–defined cases constitutes ideal practice,4 it follows that one will conclude that the bipolar disorders are overdiagnosed.
These assumptions may be the foundation for the conclusions stemming from MIDAS and elsewhere that the bipolar disorders are being overdiagnosed. However, the adoption of this set of assumptions requires that one disregard data suggesting that the DSM-IV criteria are not ideal in capturing the full scope of bipolarity. For instance, they lead one to disregard the evidence that mixed depression is reasonably viewed as being a bipolar spectrum disorder.
Much of the concern that bipolar spectrum disorders are overdiagnosed stems from failing to appreciate that psychiatric nosology is subject to evolutionary change. A useful illustration is provided by a review of key epidemiological studies. On the basis of the data in the Epidemiologic Catchment Area (ECA) Survey, the lifetime prevalence of BID and BIID were estimated to be to 0.8% and 0.5%, respectively.7 Using the more refined methods of the National Comorbidity Survey-Replication (NCS-R), the lifetime prevalence of BID was 1.0% and that of BIID was 1.1%.8 This represents a 1.62-fold increase in the prevalence of BID and BIID.
Strict adherence to the DSM-IV criteria and inflexible reliance on DSM-IV–based diagnostic instruments may result in the underdiagnosis of bipolar disorders. This view is supported by research on subthreshold cases. Merikangas and colleague8 used the NCS-R database to ascertain the prevalence of these cases. Their findings indicate that when subthreshold cases are taken into account, the lifetime prevalence of bipolar spectrum disorders is 4.4%. Reanalyzing data from the ECA database, Judd and Akiskal9 estimated that the prevalence of disorders that fall along the bipolar disorders spectrum is at least 6.4% if subthreshold cases are included. This leads to the question of whether there is evidence that subthreshold cases are reasonably viewed as being bipolar spectrum phenomena.
Subthreshold cases are associated with significant morbidity, impairment, and increased utilization of health care services.9 These findings do not appear to be nonspecific consequences of psychopathology. Impairment mounts incrementally as hypomanic/manic symptoms accrue in the context of a MDE; 1 symptom is associated with less impairment than 2, and so on. Subthreshold cases would be included within the category of bipolar disorder NOS in DSM-IV. BID and BIID, which are now universally accepted as being valid diagnostic categories, potentially take into account only 32.3% (2.1%/6.4%) to 47.7% (2.1%/4.4%) of the bipolar spectrum if subthreshold cases are regarded to constitute bipolar spectrum disorders.8,9
It appears that the MIDAS researchers have based their work on the assumption that DSM-IV effectively captures the essence of bipolarity. Moreover, the researchers apply the principles governing the use of the SCID in routine clinical practice. Therefore, their scope of bipolarity encompasses cases of BID and BIID, as defined in the DSM-IV, along with some cases of bipolar disorder NOS. In this vein, it is important to remember that some reports suggest that SCID sensitivity for detecting cases of BIID is questionable.10-12
The findings that only 3.0% of the subjects in the MIDAS database meet the criteria for bipolar disorder NOS does not accord with the results of analyses using landmark epidemiological databases is interesting.7,8 These analyses indicate that the percentage of bipolar disorder NOS is greater than that of BID and BIID combined.8,9
The MIDAS researchers reported that 8% of their bipolar patients have a family history of bipolar disorder. In contrast, their data indicate that the prevalence of disorders among patients who are not classified as having a bipolar disorder is 3.5%. The latter finding is difficult to reconcile with the epidemiological findings pertaining to the prevalence of subthreshold cases in the general population. The base rate of bipolarity among the first-degree relatives of those not classified as having a bipolar disorder (ie, those categorized as being in the “overdiagnosed group”) in the MIDAS database is 54.9% (3.5%/6.4%) to 79.5% (3.5%/4.4%) of the adult population in the United States.7,8
Critical methodological points
There are 2 critical methodological points requiring discussion of relevance to both routine clinical work and research endeavors. The focus on the scope of the bipolar spectrum in this 3-part series has centered around the ascertainment of the proportion of all MDEs that meet the criteria for a recognized bipolar variant (eg, mixed episode, mixed hypomania, mixed depression). The thesis that the scope of the bipolar spectrum be broadened must be interpreted with this procedural element in mind.
One method of proceeding in determining the scope of the bipolar spectrum in a clinical database is to first identify those patients who meet the criteria for MDE. The second step is to determine whether a given patient meets criteria for mixed episode, mixed hypomania, or mixed depression. The third step is to calculate the fraction of all those who are first identified as being in the midst of an MDE who meet the criteria for any of these 3 states.
The MIDAS investigators have not proceeded in the manner described. They have not studied the prevalence of bipolar disorders among a series of patients whom they first recognize as being in the midst of an MDE. It is conceivable that the use of this strategy would lead them to find a higher prevalence of bipolar disorders among these patients than they have reported to exist among patients in the entirety of their database. Second, if a patient did not meet criterion A for hypomania/mania, the process of searching for other features characteristic of mixed depression was aborted. Thus, the MIDAS Project was not designed to determine the prevalence of mixed depression (M. Zimmerman, personal communication, May 28, 2010).
Mixed states can consist of the mingling of any and all combinations of the features that define a depressive episode and those of hypomania/mania. Consequently, the phenomenology of these states is marvelously diverse. Detecting both mixed hypomania and mixed depression requires the use of a structured interview format. The detection of mixed hypomania, particularly mixed hypomania with marked ultradian cycling, requires that clinicians be attuned to circadian changes in symptoms across days and in some cases weeks. Mixed hypomania can be treated with a combination of medications used in the treatment of mixed mania.
Mixed depression can include any and all features of hypomania/mania. There is a void in the literature as to how these states are best be treated. Alternative strategies that appear to be useful clinically but have not been subject to study in randomized trials were highlighted in part 2. Studies of the treatment of mixed depression with the goals of developing evidence-based treatment guidelines are under way abroad. This is only to be expected, because mixed depression has received minimal attention in the United States relative to Europe.
The vast majority of patients who present in the midst of an MDE are in mixed states. This perspective is supported by the extant literature. Skill in detecting mixed states in their manifold forms is basic to the practice of child, adolescent, and adult psychiatry.
References1. Benazzi F. Is there a continuity between bipolar and unipolar disorders? Psychother Psychosom. 2007;76:70-76.
2. Dilsaver SC, Akiskal HS. Does unipolar depression exist? J Affect Disord. In press.
3. Robins E, Guze SB. Establishment of diagnostic validity in psychiatric illness: its application to schizophrenia. Am J Psychiatry. 1970;126:983-987.
4. First MB, Spitzer RC, Gibbon M, Williams JB. Structured Clinical Interview for DSM-IV Axis I Disorders-Patients Version (SCID-I/P, Version 2.0). New York: Biometrics Research Department, New York State Psychiatric Institute; 1996.
5. Zimmerman M, Ruggero CJ, Chelminski I, Young D. Is bipolar disorder overdiagnosed? J Clin Psychiatry. 2008;69:935-940.
6. Zimmerman M, Ruggero CJ, Chelminski I. Psychiatric diagnoses in patients previously over-diagnosed with bipolar disorder. J Clin Psychiatry. 2010;71:26-31.
7. Regier DA, Farmer ME, Rae DS, et al. Comorbidity of mental disorders with alcohol and other drug abuse: results from the Epidemiologic Catchment Areas (ECA) Study. JAMA. 1990;264:2511-2518.
8. Merikangas KR, Akiskal HS, Angst J, et al. Lifetime and 12-month prevalence of bipolar spectrum disorder in the National Comorbidity Survey-Replication. Arch Gen Psychiatry. 2007;64:543-552.
9. Judd LL, Akiskal HS. The prevalence and disability of bipolar spectrum disorders in the US population: re-analysis o the ECA database taking into account subthreshold cases. J Affect Disord. 2006;73:123-131.
10. Akiskal HS, Benazzi F. Optimizing the detection of bipolar II in outpatient private practice: toward a systematization of clinical diagnostic wisdom. J Affect Disord. 2005;66:914-921.
11. Benazzi F, Akiskal HS. Refining the evaluation of bipolar II: beyond the SCID-CV guidelines for hypomania. J Affect Disord. 2003;73:33-38.
12. Benazzi F. Diagnosis of bipolar II disorder: a comparison of structured versus semi-structured interviews. Prog Neuropsychopharmac Biol Psychiatry. 2003;98:988-991.