Consensus and evidence in bipolar disorder: Part I, Consensus

CP: Eleven professional associations over the last two decades have published guidelines for the treatment of bipolar disorder (G.B. Parker and colleagues, Acta Psychiatrica Scandinavica 135(6):515-526, 2017; Editor), and of the many recommendations therein, only two have achieved consensus, and neither pertains to the treatment of bipolar depression. One recommends treating mania with antimanic agents, and the other cautions against treating bipolar depression with antidepressants to avoid risk of antidepressant-induced mania. What does that tell us about bipolar disorder, or the construct of bipolar disorder as currently understood by a preponderance of titular psychiatric experts and opinion-makers?

Quidnunc: I believe that it may tell us what your question implies—that there may be something incomplete or incorrect about our model of bipolar disorder, and about our understanding of bipolar depression, in particular. I believe that it has to do with something incomplete or incorrect about how we understand bipolar mania. During the 1960s and increasingly during the 1970s and 1980s, the defining features of mania were psychomotor activation, reduced need for sleep, and mood elevation. Anger and irritability were understood as depressive symptoms almost exclusively, and when observed, they conduced to ruling out dysphoric mania or mixed states and diagnosing depression.  Accordingly, persons with dysphoric hypomania or mania were being diagnosed as having bipolar depression and treated with antidepressants. Antidepressants with mood stabilizers may be effective for bipolar depression, but evidence is strong that antidepressants exacerbate hypomania and mania. Persons with a missed diagnosis of dysphoric hypomania or dysphoric mania therefore were not responding well to treatment, and many were considered to have treatment-resistant depressive syndromes. When hypomania or mania had been missed entirely and a diagnosis of (unipolar) depression had been made, those persons had been diagnosed as having treatment-resistant major depressive disorder. I suspect that a substantial number of clinicians were oblivious of mania when it failed to present in a classic manner–that is, when no apparent mood-elevation was detected.

 I believe that diagnosis was further “shuffled” as the supply of a new, putatively highly effective treatment increased (lithium, approved in the U.S. in 1970 for treatment of DSM manic-depressive disorder). The demand for treatment increased, and the population of treatment candidates became increasingly heterogeneous, incorporating, as it did, persons with diverse comorbid disorders; persons with symptoms that impaired rather than enhanced productive functioning (the representationally biased depiction of a “typical” patient with mania); and those who experienced both manic and depressive symptoms within the same episodes. We were (and still may be) unready for heterogeneity within a group not explicitly defined by rage, anger, and, in the less symptomatic, irritability, accompanying and connoting mania. Response rates have fallen pari passu with increasing heterogeneity in persons with bipolar disorder.

CP: Upon what evidence do you base your interpretation?

Quidnunc: Studies by U.S. and European researchers have reported error rates of up to 80 percent when samples of patients initially diagnosed with unipolar or bipolar major depressive episodes—bipolar disorder, depressive phase (patients with bipolar disorder, diagnosed with a first episode or a relapse of major depressive episode)—were systematically reevaluated and found to have clinically significant mania spectrum symptoms, prompting revision of diagnoses to depressive mixed state, bipolar disorder, mixed subtype, or dysphoric hypomania or dysphoric mania. Their original diagnoses of major depressive episode were revised to accommodate dysphoria, of which anger, rage, and irritability are common and conspicuous features.

Before an initial misdiagnosis of “pure depression” is corrected to a diagnosis of depressive mixed state or dysphoric mania or hypomania, a “vicious cycle” is initiated—while receiving regimens for depression, manic symptoms misdiagnosed as depressive symptoms go untreated but are expected to improve; instead, they worsen.  When a neuroleptic or mood stabilizer are coadministered with the theretofore ineffective antidepressant, they improve, and the coadmistered drug is claimed to have operated as an antidepressant augmentation or pharmacodynamic additive. What it has really provided is antimanic activity. Similar studies, of patients deemed “treatment-resistant” have been found to represent a refined group of such persons, and account for some of the highest “missed diagnosis” rates within the range. Treated with antidepressants, persons with mania or hypomania become more symptomatic; co-treated with neuroleptics and/or mood stabilizers, they improve; both are generalizations, of course, but strongly evidenced ones.


CP: We have been discussing worsening of extant mania. What about antidepressant-induced new-onset mania in persons with bipolar disorder?

Quidnunc: What is less consistently evidence-based, and in fact, happens to be the second of two recommendations about which consensus was reached in the study by Parker and colleagues (supra; editor) is avoidance of antidepressants in persons with a history of mania, out of concern that they can induce new-onset mania in persons who are in remission. In my opinion, the evidence adduced in support of that is equivocal and does not justify its current status as one of two consensualized recommendations in psychiatry pertaining to the treatment of bipolar disorder.

CP: What has been the response of dysphoric hypomanic or dysphoric manic patients when they have received antidepressant monotherapy for depression?

Quidnunc: Antidepressants are documented to worsen mania spectrum symptoms in persons with bipolar I disorder (those who meet criteria for manic episode) , so if their manic episode is dysphoric, anger and irritability worsen, and if diagnosed as depressive symptoms, their failure to respond is interpreted as treatment-resistance. This occurred with the patients in studies of unipolar major depression as well—they were interpreted to belong to a subgroup of persons with treatment-resistant major depression, rather than one of the 20- to 80-percent of persons with bipolar disorder experiencing a first episode of mania in which dysphoria was a prominent symptom.

  “Poor long-term outcome” is the “fate,” as some see it, of persons with bipolar I depression who receive antidepressants. When first formulated, that fate prompted treatment with atypical antipsychotic monotherapy or monotherapy with mood stabilizers at higher dosages, the rationale, that they had failed to respond because of insufficient dosage of mood-stabilizers, which were averred early on to have antidepressant as well as antimanic properties. Hence, the enhanced emphasis on use of neuroleptics, such as olanzapine, quetiapine and more recently, lurasidone, as monotherapy for bipolar depression. It seems logical, if ironic, that when treatments with antimanic properties eventuated in improvement of manic or hypomanic symptoms that had been misdiagnosed as depressive symptoms, that improvement was averred to represent improvement of treatment-resistant bipolar depression.

CP: What about benefit of antidepressants for “true” bipolar depression? Does risk outweigh benefit?

Quidnunc: Because antidepressants worsen extant mania, the initial step should be diagnostic reevaluation of persons thought to have treatment-resistant depression, to rule out current manic features and personal history of hypomania or mania.

This presupposes that there is benefit in treating bipolar depression with antidepressants, and that presupposition, according to the best available evidence, is correct: Both the older literature and more recent studies with newer-generation drugs consistently report acute and prophylactic effectiveness of both older- and newer-generation antidepressants for treatment of major depressive disorder and for reducing relapse or recurrence of major depression in patients with remitted bipolar depression.

The prospective studies published by Laurie Altshuler and colleagues, of Robert Post’s group at NIMH, of remitted patients with bipolar disorder, is representative of that body of evidence in two respects: first, it reports less than half the number of relapses in patients who were taking antidepressants (compared with those who had tapered and stopped them), and reported no difference in the numbers of patients with recurrences of mania  (both having taken mood stabilizing medication throughout). 

The literature is replete with studies reporting acute and prophylactic antidepressant activity of antidepressants in persons with bipolar disorder who are adjunctively taking mood stabilizers. Longer terms of treatment have continued to be associated with statistically significant superior response and remission rates.

Because the studies in evidence are shorter than naturalistic durations of need, long-term follow-up is not yet authentically  long-term. Yet, those studies consistently report that antidepressant coadministration with mood stabilizers continues to prove superior to regimens of mood stabilizers without antidepressants. Still, the consensus of the field is that antidepressants are less effective than they should be for persons who receive pharmacotherapy with antidepressants. This impression, of insufficient response of depression in persons with bipolar disorder, has not been systematically documented, but withal is scrupulously broached when the need for better treatments for bipolar depression comes up. Whether the decrement in quality of life over time is incommensurate with “physiologic” demoralization related to age-related constraints or losses (rather than to symptomatic depression that may respond to treatment better than that currently available) remains obscure. If it is, there may be no reason to invoke differential response to antidepressants. The issue arises, however, as a feature that distinguishes bipolar from unipolar depression in the context of discussing benefit and risk of antidepressants. Nevertheless, it seems that the issue does not come up as a characteristic of antidepressants when discussing benefit and risk of antidepressants for unipolar depression (and there are as many or more examples of studies of unipolar depression, wherein long-term outcome or quality of life turned out to be less superior, relative to placebo, than had been hoped).

Back in the 1970s and 1980s, the emphasis was on monotherapy with lithium or valproate at high enough dosages or blood concentrations. Results had been less robust than had been hoped, which is to say, less robust than the results in best responders to coadministration of antidepressants and neuroleptics and/or mood-stabilizers.

The issue, as it pertains to diagnosis, is that we still have a representation bias of persons with bipolar disorder as typically euphorious when they are manic and sad when they are depressed. That is not invariably so: Dysphoria differs from sadness. Dysphoric patients are irritable, sometimes episodically enraged; they are tense, intolerant of frustration, of criticism; they typically have sleep disturbances. We may have been diagnosing many of these persons with major depressive episodes (both those in persons with prior mania and those with a first episode of mania or hypomania) and treating them with antidepressants, rather than diagnosing them with mixed states or dysphoric hypomania or dysphoric mania, and adding an atypical antipsychotic or mood stabilizing antimanic agent to their antidepressants.

CP: Patients with a missed diagnosis of dysphoric hypomania or dysphoric mania represent an example of one of the rare cases in which mania, with so many observable (rather than subjective) manifestations, is missed, and contributes to the small percentage by which the sensitivity of a diagnosis of mania is less than one hundred percent. It adversely affects the specificity of the diagnosis of mania as well, in that manic symptoms, which are not absent in these patients, are thought to be, and they are not diagnosed with mania.

Quidnunc: Yes. This is one of the few instances that reduce the sensitivity and specificity of a very conspicuous diagnosis—one that can even be made through the telephone when anyone who observes the patient and describes what they see to a physician on the line, and the sensitivity. The specificity of diagnosis is almost identical to that resulting from direct observation of the patient.

The frequency of missed diagnosis and consequent inadequate responses to regimens intended for persons with major depressive episodes rather than exacerbation of dysphoric hypomania has contributed, I believe, to a lack of consensus about what is therapeutically optimal and most conducive to favorable long-term outcome. Lack of consensus with respect to treatment of depression in the context of bipolar disorder has been a “wake-up call” of need for diagnostic correction.

Another issue broached by diagnostic error is whether the association between antidepressants and mania is causal. It is an important issue for several reasons, not the least of which is that it is one of the two issues about which consensus exists, with respect to bipolar disorder. (In the study we have been discussing, the other is that acute mania should be treated with a mood stabilizer, a neuroleptic, or, as has been reported, coadministered mood stabilizer and neuroleptic.) Among the earliest reports in the modern literature, coadministered fluoxetine and olanzapine prove more effective than olanzapine monotherapy.) Evidence that antidepressants exacerbate mania has been reported more consistently than have reports that compellingly suggest that antidepressants are causally associated with new-onset mania evidence Someone who has a mixed state but has been diagnosed with a depressive disorder will, according to a consistent evidence base, evince worsening of mania spectrum symptoms (such as irritability, psychomotor activation, sleep disturbance (especially reduced need for sleep), impulsivity, reduced frustration tolerance—collectively, evolution of hypomania to mania). Worsening of preexisting manic symptoms is supported by a consistent body of evidence, while induction of hypomania or mania in persons who are euthymic or have a “pure” depressive syndrome, absent hypomanic or manic symptoms, is based on anecdotage and observational studies. Moreover, abrupt withdrawal of antidepressants also has been implicated in inducing mania de novo.  I believe that the distinction between worsening of extant mania and induction of mania de novo is less scrupulously documented or appreciated than it should be, and that contributes to a confirmation bias that lends credence to a causative association between antidepressants and new-onset mania. I believe that some, if not a preponderance, of patients who are considered to have mania induced by antidepressant exposure may instead  be showing an exacerbative relationship after mild or subtle mania symptoms are missed.

 I believe that antidepressant exposure is associated with mania, but not necessarily causally, and therefore that the association need not be considered a risk that should be considered against the benefit of treating bipolar depression with antidepressants. I do believe that the evidence supports a causal association between antidepressant exposure and worsening of mania present before antidepressant exposure begins, however, and that that should be considered in the calculus of risk-benefit associated with antidepressant treatment of bipolar depression. That risk can be subtracted by coadministration of drugs that have antimanic activity.

I do not believe that antidepressant-induced mania is a necessary or causal interpretation of the findings observed in every case wherein mania and antidepressants are associated.

I believe that current evidence best supports the conclusion that  antidepressant exposure is sometimes a non-causal association of mania; in my opinion, no compelling evidence conclusively demonstrates a causal relationship between antidepressant exposure and new-onset mania. Moreover, I believe that our evidence base suggests that antidepressants convey significant benefit for treatment of bipolar depression, and are not associated with increased risk of mania spectrum symptoms when coadministered with mood-stabilizing agents that have antimanic activity.

CP:   Does hat bring us back to why consensus about bipolar disorder in psychiatry is confined to treatment recommendations for mania?

Quidnunc: It does, because that was the rationale for one of the two treatment recommendations about which consensus was reached in the study by Parker and colleagues (supra; editor).

Clinicians typically succeed in diagnosing mania when it is present, and typically do not diagnose it when it is not.  I believe that that is because they do so on the basis of objectively observed symptoms, such as psychomotor activation, reduced sleep duration, rapid, pressured speech, increased interpersonal interactivity, and impulsivity. Euphoriousness and rage are typically expressed by observable, objectively ratable behavior rather than by subjective self-report. The observability of manic symptoms conduces to high sensitivity and specificity of diagnosis. (Sensitivity refers to the percentage of persons with mania who are diagnosed by standard diagnostic criteria as having mania; specificity is defined as the percentage of persons who do not have mania, and do not meet standard diagnostic criteria for mania.) Likewise, mania ratings show high inter-rater reliability for presence and severity of manic symptoms that correspond to diagnostic criteria. What works and what doesn’t are accurately and dependably documented in our evidence base of the manic phase of bipolar disorder. Consensus about what constitutes optimal practice is predictably easier to attain when sensitivity and specificity of assessment during clinical trials are high. As with diagnosis, remission of mania is assessed with high sensitivity and specificity. Results of clinical trials of antimanic agents may therefore be less likely to prove falsely positive or falsely negative than may clinical trials of antidepressants. Short onsets and offsets of robust drug effects conduce to consensus.    

  Depressive disorders, with their greater dependence on subjective symptoms, are less sensitive and specific with respect to diagnosis and response to treatment. Moreover, assessment of response and outcome in clinical trials of antidepressants are less generalizable to patients encountered in practice. Antidepressant response in practice is more variable than antimanic response, and some of that variability is likely attributable to greater subjectivity in assessment of depression. Variability of response does not favor consensus, and that, too, may explain why the study we have been considering found consensus only for treatment recommendations in the manic phase of bipolar disorder.

Symptoms of depression, such as anhedonia, sadness , self-reported feelings of hopelessness, helplessness, low self-esteem, suicidality and imminence of suicide risk, outnumber that that is objectively observed, such as sleep disturbance, appetite change, impaired attention, and fatigue. The opposite is true of mania. Diagnosis of depressive syndromes depends more upon self-reported subjectivities than on objective observables. Moreover, though just an impression, it seems to me that more symptoms of hypomania or mania are in practice considered manifestations of depression (particularly irritability, rage, and emotional over-reactivity) than are symptoms of depression considered manifestations of hypomania or mania. “Depression,” self-reported, and (in my experience) not uncommonly taken at face value during assessment, may more likely represent dysphoric hypomania or mania, with dysphoria devolving from mania-related sleep deprivation, prolonged hyper-exertion, substance exposure, and/or discontinuation effects of some prescription medications.


CP: To what extent might the lack of consensus about treatment recommendations reflect change in the conceptual model from which the construct of bipolar disorder derives?

Quidnunc: They are related. The modern era has been dominated by a “continuum model,” in which bipolar depression and mania represent clinically “opposite” states of mood that represent extremes of a normative range of physiologic mood cycling. There are problems with that model that do not fit empirical findings. According to the continuum model, the points at which normal “high” is deemed hypomania or mania and normal “low” depression is arbitrary and, as with “essential” hypertension and idiopathic entities representing extremities of normal, continuously distributed traits, with a threshold of “disorder” defined by the severity of their acute manifestations and long-term complications. Endocrine and metabolic diseases afford examples wherein pathology is defined as an extremity of physiologic function—idiopathic thyroid hormone and growth hormone disorders are so defined by virtue of manifesting “bipolar” pathology beyond antipodal thresholds of normal function.

Hypothetical models of pathogenesis posit significant interindividual variability with respect to severity and periodicity of the phases of bipolar disorder, and six decades of investigation have documented significant contributions of both genetic and non-genetic factors to the variance of its occurrence.

Phases of bipolar disorder typically have been regarded as variants of normal states, with mania and depression representing exaggerated manifestations of normative mood —the opposite tails of a continuous distribution of normative stability and reactivity of mood, diurnal cycling of sleep, motor activity, and arousal—akin to models of idiopathic “essential” hypertension, hypo- and hyperthyroidism, or hypo- and hypersecretion of growth hormone, with normal and abnormal defined arbitrarily, by virtue of presence and severity of complications late in their courses contributing to the arbitrary thresholds of disease. The model is consistent with polygenic inheritance, incidence rates consistent with expectation from genetics, and juvenile and adult syndromes. Bipolar disorder, a construct, and analogous externally validated diseases differed in a predictable way: dosages of thyroid hormones and growth hormone were continuous in transition from therapeutic for deficiency and excessive for exacerbation. The points on the distribution at which the thresholds for depressive and manic phases lie have not been formulated, although the phases have been extensively described and codified in the form of diagnostic criteria. This “continuum model” of bipolar disorder afforded rationales for prescribing antidepressants for bipolar depression, and provided a “priming bias” for a causal role with respect to induction or exacerbation of mania proportional to their bioavailabilities.

A related problem clinically was assessment of effects at the low end of the spectrum. Diagnostic confusion arose from the less severe or atypical presentations—milder forms of mania, in particular, wherein tense, irritable, sleep-disordered patients (those with dysphoria), deviating from the stereotyped elation or euphoriousness of mania, were more likely to be diagnosed with bipolar depression and treated with antidepressants. When unaccompanied by antimanic mood stabilizers, antidepressants are documented to worsen hypomania and mania. Whether they may induce them in a euthymic person is less certain, although, as you mention, consensus in our field is that they do.

CP: You believe, then, that antidepressant treatment meets an evidence-based benefit-risk standard as therapeutic for bipolar depression.

Quidnunc: When coadministered with drugs that have antimanic (mood-stabilizing) activity, yes. That is what a preponderance of evidence suggests.

Leave a Reply

Your email address will not be published. Required fields are marked *