Polarised views about bipolar disorder(s): A critique of the 2020 College guidelines for mood disorders

Abstract

The 2020 College guidelines for mood disorders banish bipolar II disorder – despite its formal status in Diagnostic and Statistical Manual of Mental Disorders and International Classification of Diseases manuals for more than two decades – and argue that there is no need to partition bipolar disorder into separate sub-types. Their single-entity model is seemingly based on opinion rather than any support from referenced scientific studies. The author challenges the Committee’s model of there being only one bipolar disorder and argues that it presents several clinical management risks, particularly of ‘over-treatment’.

The very ink with which history is written is merely fluid prejudice. (Mark Twain)

Two ideas are always needed: one to kill the other. (Georges Braque)

In the 2020 College guidelines for mood disorders (Malhi et al., 2021), bipolar II disorder is effectively banished. The report stated that the Committee ‘… no longer distinguishes bipolar I and bipolar II disorder. It is our basic premise that all syndromes characterised by symptoms of mania are best described simply as bipolar disorder’ (p. 11). The Committee was further critical of the International Classification of Diseases (ICD) and Diagnostic and Statistical Manual of Mental Disorders (DSM), noting, particularly with regard to Diagnostic and Statistical Manual of Mental Disorders, 5th edition (DSM-5), that its ‘partitioning [of] bipolar disorder …. is arbitrary and does not meaningfully inform management’ (p. 11).

Bipolar II disorder achieved formal status in the 1994 Diagnostic and Statistical Manual of Mental Disorders, 4th edition (DSM-IV) and continued to be positioned as distinct from bipolar I disorder in DSM-5. It was formally classified in the ICD, 10th Revision (ICD-10) system (adopted for use from 1994) and remains as a separate bipolar condition in ICD, 11th Revision (ICD-11). Thus, it has had formal DSM and ICD status for over a quarter of a century.

Its declared non-existence by the Committee is therefore a provocative stance and worthy of close attention in light of its implications, particularly with regard to management. One looks for a logic. Could the decision simply reflect semantics in defining mania? No; the authors state that (in their view) ‘mania is used to describe all manic symptoms of varying severity in which there are no concurrent depressive (mixed) symptoms’ (p. 70) and that they do not use the term ‘hypomania’ as it is ‘a product of the arbitrary subtyping of bipolar disorder’. No argument is offered for this stance, and it is therefore polemically based.

In this critique, let me first note the Committee’s overall operative process. It is stated in the report (p. 8) that evidence-based recommendations were prioritised, where possible, and if there was insufficient or no evidence, the Committee proceeded with consensus-based recommendations, a common and understandable model.

In a 2019 monograph focusing on bipolar II disorder, the Committee chair (Malhi, 2019) prefaced his chapter with

I do not view bipolar II (BP II) as a discrete subtype of bipolar disorder … instead, I position it as a pragmatic construct … [and] I am not persuaded that it has a distinct underlying pathophysiology and, as such, its treatment does not need to be necessarily unique.

This ex cathedra statement is effectively captured in the Committee’s stated views about classifying and managing the bipolar disorders.

I seek to challenge their two postulates. First, I argue that bipolar II is a discrete entity and is therefore separate from bipolar I. Second, I argue that a singular management model for managing all the bipolar disorders is a Procrustean one with attendant clinical risks.

So, first, is there only one bipolar disorder simply varying by severity (a unitary model), or does evidence of two principal separate categorical types (a binary model) exist? In providing their answer to this question, we would expect the Committee to provide evidence that supports their unitary model and/or rejects the binary model.

If a binary model is valid then, as Kendell (1989) argued, bimodality (evidencing a ‘point of rarity’) in symptom scores should be evident. No such studies are considered by the Committee. In an earlier study (Parker et al., 2016), 1081 patients with a bipolar disorder completed a 46-item self-report bipolar screening measure. A mixture analysis of item scores showed a distinct bimodal distribution, supporting a binary model. Those subjects analytically assigned to the putative bipolar I category were more likely to report psychotic manic episodes, to have been hospitalised in a hypo/manic episode, to have distinctly longer hypo/manic episodes and to have a higher rate of bipolar disorder in their families.

In a second study (Parker et al., 2020a), 165 bipolar patients completed a 96-item measure of hypomanic/manic symptoms. Latent class analyses were undertaken, which again determined that a two-class solution provided the best fit, with those assigned to Class I (putative bipolar I) being significantly more likely to report psychotic symptoms, grandiosity/overvalued ideas, and impaired judgement and to have been hospitalised. Mixture analyses showed a bimodal pattern with a distinct point of rarity, again arguing for there being two principal bipolar disorder ‘types’ and rejecting a unitary model (or single condition). Such studies provide empirical evidence rejecting a single bipolar disorder entity simply varying by severity and proceed beyond any personal view about modelling the bipolar disorders.

As noted, the Committee viewed the DSM-5 classification as ‘arbitrary’ but did not explain what they meant. ‘Arbitrary’ implies decisions based on personal whim or random choice rather than being systematic and based on reason. Against this assertion, I suggest that the DSM criteria for the bipolar disorders do reflect a distinct ‘system’ – albeit one open to some criticism about some illogical nuances, which will now be considered.

DSM-5 criteria for both manic (bipolar I) and hypomanic (bipolar II) episodes have similar requirements for Criterion A as well as seven identical symptom B criteria and identical imposed cut-off numbers, suggesting that DSM-5 seeks to first define ‘a’ bipolar condition irrespective of the sub-type. DSM-5 then differentiates bipolar I from bipolar II disorder by several criteria (i.e. psychotic status, hospitalisation, impairment and duration), with each being somewhat problematic. For instance, Criterion E for hypomanic episodes states that ‘if there are psychotic features, the episode is, by definition, manic’ as opposed to hypomanic (p. 125), and Criterion B for bipolar II disorder states ‘there has never been a manic episode’ (p. 134). Logically, such criteria stipulate that, to be diagnosed with a bipolar II disorder, a patient must never have experienced psychotic features during a period of mood elevation. However, the specifiers then listed for bipolar II disorder (pp. 134–135) include having (1) mood-congruent psychotic features or (2) mood-incongruent psychotic features. Presuming that the incongruence reflects an undetected error of logic, then the presence of psychotic features automatically assigns a bipolar I DSM-5 diagnosis to a hypo/manic patient.

Bipolar I status can, however, also be obtained in the absence of any psychotic features if two other (seemingly problematic) criteria are met. The first is if the patient is hospitalised. This is a quaint criterion, as I am unaware of any other psychiatric or medical conditions defined by hospitalisation and which may more reflect service nuances rather than ‘psychopathology’. The second is the presence of ‘marked impairment’, as against a bipolar II hypomanic episode being ‘not severe enough to cause marked impairment’. This criterion is both dimensional and subjective and, having no clear or operational criteria, risks misclassification. Furthermore, it assumes that function in hypo/manic states is always ‘impaired’. As referenced in a report by the AREDOC (Assessment, Revision and Evaluation of DSM and other Operational Criteria) group (Parker et al., 2020b), a distinct percentage of individuals experience genuinely improved functioning in both hypomanic and manic states. This reality led the AREDOC group to recommend, in short, that any such criterion be weighted to a ‘distinct change in functioning rather than impairment being required of necessity’.

As part of Criterion A for hypo/manic episodes, DSM-5 imposes a minimum duration of 7 or more days for mania and 4 or more days for hypomania. These criteria were preserved from earlier DSM editions and were derived from clinical consensus rather than empirical evidence. This was underpinned by a concern about the bipolar disorders being over-diagnosed; here, the Committee’s use of ‘arbitrary’ is cogent. Numerous studies (e.g. Angst, 1988; Bauer et al., 2011; Judd et al., 2003) have shown, however, that for those meeting DSM symptom criteria for a bipolar disorder, symptoms are similar across subsets of those (1) meeting and (2) not meeting such duration criteria. Imposing the DSM duration criteria risks excluding those with a true bipolar condition, and as a result, the most common bipolar diagnosis in America for those not meeting the duration criteria is a diagnosis of ‘bipolar disorder, not otherwise specified (NOS)’.

Considering such contentious differentiating criteria, I would argue that an alternate, more parsimonious diagnostic model should be employed – one in which a diagnosis of bipolar I disorder is assigned to those who experience psychotic manic states (with such patients also commonly experiencing psychotic depressive episodes) and a diagnosis of bipolar II disorder is assigned to those who have categorical hypomanic states but have never been psychotic at any phase of their illness. Such a model reflects both relevant empirical studies and historical descriptions of manic-depressive illness.

Why is this diagnostic issue of such importance? If there are separate bipolar sub-types, then they are likely to have differing clinical features, causes and responses to differing treatments. A simple analogy is diabetes. Managing type I diabetes may prioritise insulin, while managing type II diabetes may prioritise weight loss and attention to diet.

The Committee’s model risks a Procrustean treatment approach where management is identical in those with psychotic and non-psychotic mood swings, with multiple attendant risks and particularly risks of ‘over-treatment’ for those not having a bipolar I condition. The issue of management is central to this article.

While the Committee’s guidelines address differing illness phases, let me focus only on the maintenance phase medication recommendations. Here, the Committee’s preferred monotherapy is lithium, with alternative or second-line options being valproate, quetiapine and asenapine. Lamotrigine is not in that prioritised list, although it is later considered as another option in nuanced non-maintenance situations.

The potential utility of lamotrigine has effectively gone under the radar as a mood stabiliser for several decades. That history is worthy of some detailing, as it may have contributed to the limited support for it provided by the Committee as a maintenance medication. Abstracting from an earlier review (Parker and McCraw, 2015), the five early pivotal trials had a limited target condition (bipolar depression as against maintenance), trial lengths were too brief, the lamotrigine dose was commonly too low and most of the five studies were weighted to subjects with a bipolar I rather than a bipolar II condition. These key pivotal studies were interpreted by their authors (often the same authors) as variably demonstrating that lamotrigine was equivalent to placebo or superior to placebo, with the first and earlier interpretation dominating and leading to a common summary view that lamotrigine was a ‘failed drug’ for managing bipolar disorder. Our review then summarised findings from several subsequent maintenance studies of those with a bipolar I disorder where lithium and lamotrigine were compared. Most suggested equivalent efficacy, while one study argued that lithium delayed time to manic (but not depressive) episodes and lamotrigine had more ‘robust’ benefit against depression. We referenced one meta-analysis (Smith et al., 2007) of data from 14 randomised controlled trials (concluding that lithium and lamotrigine were of comparable efficacy and were both superior to valproate) and another meta-analysis (Soares-Weiser et al., 2007) of 45 relapse prevention trials (concluding that lithium, lamotrigine and valproate were of comparable benefit).

Thus, while I personally judge lithium as distinctly superior to lamotrigine in the maintenance phase of bipolar I disorder, there is a base of evidence in support of lamotrigine for managing this sub-type, despite the Committee failing to list it in their recommended monotherapy list.

The evidence base in relation to bipolar II disorder maintenance therapy is essentially non-existent, and thus management options currently weight clinical opinion. In our earlier review (Parker and McCraw, 2015), we noted that, in the second edition of a monograph focusing on bipolar II disorder (Parker, 2012), an increased number of invited experts in managing bipolar II disorder had come to favour or judge lamotrigine as a foundational treatment for this condition than was evident in the 2008 edition. Such views – that lamotrigine is the drug of first choice for the bipolar II sub-type – led us to undertake the first randomised controlled bipolar II maintenance study comparing the two drugs in a 20-week study (Parker et al., 2021). We ceased the trial because of severe cognitive side effects reported by several receiving lithium (and where lithium serum levels did not exceed 0.8 mmol/L). The trajectory of mood stabilisation in the trial completers (16 receiving lamotrigine and 12 receiving lithium) was comparable. While the study was clearly underpowered, if the trajectories were maintained in an adequately powered sample, we calculated that a sample size of over 700 subjects would have been required to determine a significant difference in our primary outcome measure (measuring daily oscillations in depressive and hypomanic symptoms), some 650 on the Hamilton Depression Rating Scale (Hamilton, 1960) and over 3200 on the Young Mania Rating Scale (Young et al., 1978). In another single-blind study comparing lithium and lamotrigine (Suppes et al., 2008) – albeit for the treatment of bipolar II depression rather than maintenance – the authors quantified comparable response and remission rates in their 19 lithium and 21 lamotrigine patients who completed the study. However, they calculated from a power analysis that a sample size of over 10,000 subjects would be required to show a significant difference. Such power calculations suggest that it is quite unlikely that any research group will undertake an adequately powered study to determine whether the two drugs have differential efficacy as a bipolar II maintenance treatment.

While the two medications appeared broadly comparable in their efficacy in our maintenance study, they showed quite differing side-effect prevalence data and profiles. Specifically, for those who completed the study, 92% of those receiving lithium reported significant side effects as against 19% of those receiving lamotrigine (and with mean side-effect numbers being 5.9 vs 1.3). Cognitive side effects dominated in the lithium recipients (word finding difficulty in 58%, impaired memory in 50% and cognitive slowing in 42%), but tremors (58%) and thirst (50%) were also common. In the study by Suppes et al. (2008), patients reported an average of 9.2 and 4.2 side effects for those receiving lithium and lamotrigine, respectively, with those in the lithium group reporting distinctively higher rates of cognitive slowing, impaired memory, word-finding difficulties, dizziness/light-headedness, increased thirst, urinary frequency, nausea/vomiting and tremor. The author finds, in his clinical practice, that those with a bipolar II condition appear far more likely than those with a bipolar I condition to experience cognitive side effects with lithium, an issue worthy of formal study.

In the absence of definitive maintenance studies, the clinical observations and pilot data reviewed here suggest that lithium may be the preferred first option for managing bipolar I disorder, while lamotrigine is the preferred first option for managing bipolar II disorder (for the latter condition, the efficacy may or may not be superior, but the side-effect profile is distinctly superior which argues for a distinct cost/benefit advantage). Thus, a ‘horses for courses’ management model seems more appropriate than a ‘one-size-fits-all’ Procrustean model such as that advanced by the College Committee.

If, however, the Committee’s recommendations are accepted, then any individual with a bipolar II disorder would receive lithium as the prioritised mood stabiliser management option, despite it not seemingly being superior to lamotrigine. This risks a distinctively troubling side-effect profile for many individuals, even when the long-term risks of weight gain as well as renal and thyroid dysfunction are ignored.

Similar concerns apply to two other bipolar variants. Some individuals develop a first-ever hypo/manic (bipolar III) episode – often severe enough to require hospitalisation – in response to certain medications (with steroids perhaps being the drug type that most often triggers such an iatrogenic response). Clinicians also occasionally see those with a seeming ‘psychogenic’ manic psychosis, where an individual with no previous history of hypo/manic or even depressive episodes becomes clinically ‘manic’ following a personally distinctive stressful or traumatic life event. In managing these two scenarios, it is common (and common sense) clinical practice to treat the acute phase with an antipsychotic drug and/or mood stabiliser, and then slowly taper and cease such medications after some weeks or months to determine whether there is any logical need for any maintenance medication. If the Committee’s model holds, and ‘mania’ encompasses all hypo/manic states which necessitates the same management model, then many patients with temporary hypo/manic states risk being unnecessarily exposed to medications with concerning side-effect profiles for extended periods.

As a final point, let me here inquire about process. The Committee had 12 named members and 5 (unnamed) peer reviewers. Was the banishing of bipolar II disorder an agreed unanimous decision, or did others put a dissenting view? If the latter, what was the initial ratio of assenters to dissenters, and what were the arguments used to convince the dissenters? Why was dismissal of bipolar II argued as a ‘premise’ rather than empirically? On what basis were the DSM and ICD systems of the last quarter century (incorporating a bipolar II condition) able to be dismissed as ‘arbitrary’? Was there any consideration of the published literature considering whether a unitary or binary model was supported empirically? ‘The dismissal’ now has a second Australian connotation.