A further piece of my mind

Early concerns as a trainee psychiatrist

While recognition that I had found my lifetime career occurred rapidly, I had three immediate concerns.

First, the high risk of intellectual incest – not just narrowly expressed as being a ‘Freudian’ or an Adlerian’ but writ large. Long a free-range animal, I had no wish to be an acolyte and so employed several prophylactic strategies.

College training required providing 50 psychotherapy sessions for one patient, closely supervised by one supervisor. I took on three patients and sought three supervisors with differing therapeutic approaches. By taking each’s best strategies, I avoided being a paradigm captive and developed a more pluralistic approach to psychotherapeutic practice.

Similarly, I attended ward rounds run by Sydney’s two leading academics – Leslie Kiloh and David Maddison – comparing their differing diagnostic and management approaches.

Second, our lecturers commonly weighted their own sectarian model. One emphasized that schizophrenia was caused by parental ‘skew and schism’ rather than by genetic factors – a model counterintuitive to my judgment in interviewing patients with schizophrenia – and perhaps the earliest expression of my ‘gut instinct’ in play. I reviewed the lecturer’s key reference ¹ – an anecdotal paper reporting on only 14 subjects. Years later I asked its first author, Ted Lidz, how they had fared. He stated that most became successful professionals and none were significantly impaired. In essence, adolescent reactions to disputative parenting had led to an incorrect diagnosis of schizophrenia. A double bunger of invalid diagnosis and theorizing.

Years later, British researchers postulated a related model – that high parental ‘expressed emotion’ or ‘EE’ was a causal driver of relapse in those with schizophrenia. I initiated a study ² which identified a converse process. In essence, as schizophrenia in a child predictably caused parental distress – high parental EE was the consequence but not the cause of their child relapsing.

Third, I was struck by the high rate of psychiatrists not offering patients a diagnosis – as they judged a diagnosis of doubtful value or redundant - as they provided the same therapy to all patients. I judged then and today that any psychiatric patient should – at the end of initial interviewing – be given (i) their presumptive diagnosis or diagnoses, (ii) the psychiatrist’s level of confidence in those diagnoses and (iii) information on strategies needed to clarify any diagnostic dilemmas.

An academic career

I took up a Lecturer position within the UNSW School of Psychiatry in 1974 and progressed up the academic ladder at an acceptable clip. In 1983, I succeeded Leslie Kiloh as the Head of the School and acquired a similarly demanding attendant position as Clinical Director of the hospital psychiatry service. Here began my preoccupation with service models. The composite positions allowed me to implement a condition-based model – whereby academics were consolidated within clinical nodes.

Each clinical node had its own model, as illustrated by the Mood Disorders Unit or MDU, established in 1985 with Henry Brodaty as the administrative director and myself as the research director. My research model was a ‘trunk and branch’ one: all MDU psychiatrists contributed to ‘truncal’ studies (e.g. the classification of mood disorders) while individual psychiatrists pursued their own ‘branch’ studies.

MDU model success was recognized by an NHMRC Program Grant – the first awarded to a psychiatry team. Further, a 2001 audit of Australia’s highest ranked researchers in the field of psychiatry and psychology listed seven MDU psychiatrists in the top twenty.

Two MDU non-psychiatrist team members mandate highlighting. Firstly, Dusan Hadzi-Pavlovic – an exceptionally talented statistician but showing wisdom when judging that he needed to tell me to pull my head in. Secondly, Kerrie Eyers, our MDU Coordinator and my long-term editor – applying a light hand to refine my turgid prose and sprinkling high-order whimsical touch ups to the text. As noted, ³ they have ‘thoughtfully swept the ice and friction as I have curled my way forward to achieve professional targets’ – both now for over 40 years.

The Black Dog Institute model

While proud of the MDU, its clinical service had assessed only 3000 patients after 10 years. A new model was needed – and the Black Dog Institute (BDI) was founded. My BDI model had four nodes: research, clinical services, educational programs (for professionals and consumers) and a consumer support centre. All linked iteratively – clinical observation shaping research studies and salient research findings fed back into clinical and educational services and promulgated via publications and our website. A marker of the last was that in 2009 our website ranked first in the world for quality and accurate information on bipolar disorders.

The consumer focus allowed us to cement a partnership model. A key destigmatization strategy was to invite distinguished people to go public about their mood disorder and we turned to politicians and sportsmen – with the latter scoring more goals.

Another consumer-weighted strategy involved writing competitions – a process generating seven books. The first invited essayists to trace the origins of the ‘black dog’ metaphor and we learned that it preceded Churchill’s self-description and that the ‘black dog’ or ‘black fog’ metaphor went way back into Celtic times. Subsequent essay competitions (generating many hundreds of entrants) focused on specific mood topics, and we combined keen observations and witty asides by gritty writers with our professional views. Melding ‘inside out’ wisdom and ‘outside in’ perspectives provided a very powerful educational model.

The suicide rate in Australia dropped sharply from 2000 to 2004, and I like to think that establishment at that time of the twin lighthouses of Beyond Blue and the BDI contributed to that trend break.

My preoccupation with modelling of mood disorders in diagnostic manuals

For some two thousand years, a binary model of clinical depression held sway – one type (melancholia) was quintessentially biological, having a minimal response to placebo but responsive to antidepressant drugs. The other type (reactive depression) was more situational and rarely benefitted from antidepressant medication.

In 1980, the DSM-III manual introduced the diagnostic category of ‘major depression’ which merged depressive types and marginalized ‘melancholia’. Its low symptom cut-off point pathologized minor depressive states and led to implausibly high rates of ‘clinical depression’. I have long challenged such a diagnosis and its ascriptions. While commonly described as a ‘disease entity’, it is essentially a pseudo-entity encapsulating differing conditions, each with differing causes and preferential treatment strategies.

As a pseudo-entity, it is not surprising that the huge treatment evidence base quantifies every treatment modality – antidepressants, psychotherapies, counselling and even St. John’s Wort – as equally effective. Thus, treatment risks being influenced more by the practitioner’s discipline than by the depressive type: thus, see a GP and receive an antidepressant; see a psychologist and receive CBT; see a counsellor and receive counselling.

Further, meta-analyses of trial data for the most widely prescribed antidepressant drugs have shown minimal or no superiority over placebo for major depression – reflecting trial subjects having a low rate of melancholic depression – just as might occur if a genuinely effective drug for asthma was tested in those with ‘major breathlessness’ if few had asthma. Unfortunately, such ‘results’ lead many to argue that antidepressant drugs are effectively placebos.

I have long challenged the concept of major depression. While generally ignored, in response, to one such paper, ⁴ Jay Fawcett – Chair of the DSM-5 Mood Disorders’ Work Group – generously acknowledged ⁵ that ‘Right now, with major depression as our target, we are shooting in the dark’.

Similarly, I have recently challenged ⁶ the 2024 published ICD-11 (CCDR) mood disorder criteria. ⁷ While stated objectives were to determine valid and reliable diagnoses so as to differentiate each defined condition from another and from normative states, I critiqued each claim. The CCDR’s principal depressive condition (akin to DSM’s Major Depression diagnosis) is a ‘depressive episode’ but the symptom set alone provides a low diagnostic bar, none of the symptoms are needed to be present at an intense degree and, in stating that, individuals are ‘usually’ distressed and merely need to have ‘some difficulty’ in functioning risks pathologizing human misery/depression as a clinical state. Criteria for a dysthymic disorder require a persistent depressed mood lasting 2 years or longer, but no minimal symptom number is imposed, while symptom-free periods of up to 2 months are allowed – challenging the base duration definition, disallowing any distinction from the CCDR ‘recurrent depressive disorder’ condition and challenging the manual’s objective of clearly differentiating conditions.

While DSM allows a ‘melancholia’ specifier for major depression, many of its listed criteria are the same as those defining major depression. In effect, no clear boundary. The same concern holds for the CCDR ICD-11 classification – but here no minimum number of symptoms is imposed for melancholia – merely ‘several’, once again compromising stated objectives of generating valid and reliable diagnoses. I remain bewildered as to how such lapses in logic have emerged and seemingly escaped challenge.

I have long sought to resurrect interest in melancholia. One strategy was to weight the historically accepted feature of psychomotor disturbance – and the MDU team developed a sign-based measure, the CORE. In numerous publications and one monograph, ⁸ we detailed the CORE’s capacity to differentiate melancholic from non-melancholic depression and show differential treatment responses to differing interventions. More recently, we developed a symptom-based measure of melancholia: the Sydney Melancholia Prototype Index or SMPI, quantified its high diagnostic accuracy ⁹ and have demonstrated its capacity to predict differential responses by those with melancholic and non-melancholic depression to differing treatments.

After modelling and measuring – Managing

Measuring melancholia more precisely allowed salient management studies. In one, ¹⁰ only 40 assigned melancholic subjects were required to establish a statistically superior response to antidepressant medication as against cognitive behaviour therapy – supporting the historical ascription that melancholia responds minimally to a psychotherapy and preferentially to antidepressant medication. Despite claims by pharmaceutical companies that all antidepressant class drugs have comparable effectiveness, we established differential class rates for melancholia, with the most commonly prescribed antidepressants, the SSRIs, being the least effective, the SNRIs more effective and the old-fashioned TCAs even more effective for melancholia. ¹¹ I also pursued new augmentation strategies for when antidepressant drugs alone are ineffective to alleviate melancholia, generating the first reports of olanzapine as an augmenter ¹² and also quantifying the benefit of stimulant drugs as augmenting agents. ¹³

A focus on the bipolar disorders

I sought to put bipolar II disorder more clearly on the map, as many sufferers fail to receive the diagnosis or receive an inaccurate one such a borderline personality disorder. Thus, in 2008 I edited the first book published on this bipolar sub-type ¹⁴ and with subsequent editions in 2012 and 2019.

I have also challenged DSM and ICD diagnostic criteria for the two bipolar disorders. Both list the same symptoms and symptom cut-off numbers for hypomanic and manic states and have questionable differentiating criteria. For example, DSM-5 assigns Bipolar I status if hospitalization occurs – an idiosyncratic criterion for any medical or psychiatric condition. Duration criteria effectively exclude some 40% of those with a true bipolar disorder and, while manic and hypomanic states mandate impairment, numerous studies have quantified improved functioning in a distinct percentage of those with hypo/manic states. Thus, I recruited an international Task Force of researchers to address such limitations and to generate new criteria sets. ¹⁵

A focus on burnout

For several decades Burnout has been defined as a triadic symptom state, with its key measure – the Maslach Burnout Inventory or MBI ¹⁶ – having scales assessing exhaustion, depersonalization and compromised personal accomplishment – and with such constructs similarly captured in the ICD-11 definition. As compromised performance may simply be a consequence of the first two symptoms, such syndrome definition is narrow. Further, as there are no formal cut-off criteria for MBI scales, cut-off scores imposed by researchers risk generating problematic data, well evidenced by a systematic review of nearly 200 studies evaluating burnout in physicians and quantifying rates varying from 0% to 85% and allowing disparate conclusions that burnout is non-existent or ubiquitous in physicians. ¹⁷ Thus, another ‘condition’ inviting questions about syndrome definition and measurement. In a series of studies, we empirically developed a model of burnout involving six constructs and the Sydney Burnout Measure. ¹⁸ While burnout is widely viewed as a singular stress reaction and benefitting from stress reduction strategies, our analyses alternatively modelled burnout as a diathesis stress condition, with the diathesis involving personality factors (especially perfectionism and dutifulness) and thus arguing for management addressing any predisposing personality style and not simply precipitating stressors.

Questioning the credibility of management guidelines

I remain perplexed as to why diagnostic criteria (and management guidelines) appear more eminence-based than evidence based.

For example, we reviewed ¹⁹ (Parker et al, 2017) 11 bipolar management guidelines published by authoritative bodies and with all stating their guidelines as evidence based. If so, they should be in accord, but they were distinguished more by striking differences. As Abba Eban observed, an exemplar of consensus occurring when ‘everyone agrees to say collectively what no one believes individually’.

Consequences of being a prickly bastard who questions

Financial journalist Alan Kohler once observed that ‘Collegiality is too highly valued; every board should contain a prickly bastard who keeps asking questions and won’t shut up’. Psychiatry did not make me provocative – it simply provided a channel for a predisposed contrarian position to later question diagnostic and management models that struck me as worthy of challenge. The commonest counter has been the ‘totschweigtaktic’ or ‘death by silence’ response, where the critic and their opinions are deprived of the oxygen of attention as a consequence of their not sticking with the herd or of being judged as effecting a tone of superiority.

I argue that consensus should not be enforced when issues are inconclusive and that lack of unanimity of views can actually be instructive. Thus, in editing my bipolar II books I invited multiple experts to provide independent views on specific issues. Being exposed to the experts’ (commonly contrasting) views allowed the reader to weigh the evidence. One reviewer praised the advantages of eschewing the standard format of blandly compiling accumulated data and striving for deadening impartiality. Instead, he favoured making room for the full cacophony of opinion and conflict constituting the field.

Pattern analysis and its benefits to diagnosis and tailoring of treatment

A clinical psychiatrist’s priority should be to generate ‘the’ valid diagnosis or diagnoses as it allows specific treatment choices rather than risking a ‘one size fits all’ therapy. Further, for a patient to know their diagnosis and its causes is empowering.

In medicine, diagnoses are often confirmed by a laboratory or pathology test but psychiatry has few. Selecting a diagnosis from a DSM or ICD manual hazards a ‘painting by numbers’ approach. The ‘tool’ central to psychiatric practice is clinical reasoning or phronesis – a pattern analytic approach that I have favoured for decades.

It is a Bayesian approach, seeking to identify the ‘loudest’ symptom or sign and then the next most-weighting signal and so on until the most likely diagnostic pattern match is achieved. It involves ‘implicit’ thinking or use of ‘gut instinct’ – being quick, intuitive and categorizing by reference to prototypes. Not all of my highly weighted symptoms are listed in text books. For example, in a depressed patient, I judge ‘loss of light in the eyes’ as a strong diagnostic marker of melancholia, and its return as a marker of treatment effectiveness.

Implicit thinking, however, neglects ambiguity, suppresses doubt and ignores absent evidence. Thus, I then need to move to a second stage of ‘explicit’ thinking: seeking data confirming – or rejecting the diagnosis and considering alternatives.

Such a two-stage model has also underpinned many of my research inquiries. In essence, if theories, diagnostic criteria sets or management guidelines do not accord with my ‘gut instinct’, I am up for the chase.

A gut feeling about a gut finding

In late 2023, I was contacted by Jane Dudley. She had had a bipolar disorder since adolescence and in her twenties was manic for 5% of the year and depressed for 80%. Her psychiatrist had trialled all appropriate medications without success, while Jane experienced major side-effects and required 12 hospital admissions. And then, several months after a procedure, her mood swings settled, she started to experience joy and she was progressively taken off all medications, losing many kilograms. She has since published a book, started a business and in 2024 commenced university with distinct success. Her bipolar I disorder has now been in remission for 8 years.

I gave that vignette to some 20 psychiatrists. None could guess the successful intervention. The procedure was a faecal microbiota transplant (or FMT) to be precise or a transplant of shit to be crudely precise. Not suggested by any medical practitioner but by her husband, Alex, a zoologist. It didn’t reflect serendipity (or the faecal finger of fate) but his logical pursuit of salient literature.

Her psychiatrist submitted a descriptive article but it was initially rejected – with the consequences of such a treatment being judged as implausible – and his paper was only later published as a ‘comment’. It was the first such case report in the world literature, ²⁰ and thus Jane is ‘patient zero’.

For decades, we have managed the bipolar disorders with medications impacting on brain neurotransmitters. Jane’s story – implicating the gut microbiome and how bacteria-produced neurotransmitters in the gut can travel up the gut-brain axis to impact on mood functioning – took my breath away.

As the sociobiologist E.O. Wilson observed: ‘A paradigm shift is the best a scientist can hope for’. I spent 9 months seeking to understand gut microbiome nuances and detailed this new paradigm in a recent book. ²¹ I am now of the view that pursuing the contribution of gut dysbiosis and addressing treatment-resistant mood disorders by microbiome modulation offers a new paradigm for psychiatry. So, at the end of my career as a paradigm chaser I’ve been exposed to the most fascinating model of my career – and while it has turned to crap – that is how many careers end.