DSM-5: Ordering disorder?

‘Watch out for the fellow who talks about putting things in order!

Putting things in order always means getting other people under your control.’

Denis Diderot (1713–1784)

The classification of mental illness

When assessing psychiatric symptoms, there are two important considerations at the outset. First, is there a disorder or illness, and, if so, what is it? Second, why has it presented now? Axiomatic to these questions is the definition of mental illness, but defining the latter is extraordinarily difficult, and even though there are many descriptions of mental illness, few are clinically meaningful. In practice, operationalising mental illness would be of immense benefit because it would allow the partitioning of illness and health, but any definition must capture the prevailing views and needs of society. Thus, in reality, mental illness is heterogeneous and ever changing, and perhaps by its very nature difficult to characterise (Wakefield, 1992). It is this fundamental problem that has created a black hole at the centre of psychiatric codification.

Our current assortment of psychiatric disorders is the culmination of haphazard attempts at ordering disorder. For example, some disorders are self-limiting and ephemeral, while others are chronic and intractable. Some are simply extrapolations of normality whereas others are distinct aberrations that have nothing in common with healthy experiences. The putative causes of psychiatric disorders also vary considerably, as does their onset, which spans childhood to old age. Consequently, psychiatric phenomenology is incredibly diverse, but because few signs and symptoms are characteristic and none are pathognomonic, constellations of clinical features are needed to ‘define’ psychiatric disorders. Hence, our current psychiatric classifications are imprecise and malleable, and our diagnoses have limited clinical salience (Frances and Widiger, 2012).

The DSM dynasty

At first, the Diagnostic and Statistical Manual of Mental Disorders (DSM) focused on the epidemiology of illnesses. However, by its third incarnation (DSM-III), it started making recommendations as to how to identify psychiatric disorders in clinical practice on the basis of psychopathology and clinical trajectory, instead of aetiology (American Psychiatric Association, 1980). The impetus for this change was that at the time (circa 1980) diagnostic consensus between psychiatrists was relatively poor, possibly because of differing approaches to clinical formulation derived from of a variety of schools of thought (e.g. psychodynamic, biological, behavioural). This dramatic shift in the role of DSM had a profound effect, in that it increased diagnostic agreement amongst psychiatrists and quelled aetiological discord. Inadvertently, however, it created a new set of problems.

Taking a pragmatic view, the authors of DSM-III used lists of symptoms to create diagnostic criteria, and managed to describe all manner of psychiatric presentations. It was envisaged that a comprehensive description of conditions would equip clinicians with clinical diagnoses and at the same time provide researchers the opportunity to investigate the underlying neurobiological substrates of mental illness. Ultimately, it was presumed that definitions of psychiatric disorders approximating to their ‘true nature’ would iteratively emerge, and that these would in turn spawn the development of novel treatments. However, the challenge of ‘translating’ psychiatric disorders into diseases was grossly underestimated and unfortunately the DSM dream never materialised. Indeed, currently, our fundamental understanding of most mental illnesses remains rather rudimentary. However, it is not all bad news. In practice, the listing of inclusion and exclusion criteria has certainly increased clinical reliability and made treatment more consistent. Researchers have also been able to better define populations for investigation and, arguably, DSM has facilitated advocacy for mental illnesses worldwide. However, the latest revision of DSM has prompted many questions regarding the validity of our psychiatric diagnoses and brought into sharp relief problems in our taxonomy.

In May of this year, DSM will be releasing its fifth edition (DSM-5). It was promulgated that this latest apotheosis would mark a significant shift towards dimensional aspects of psychopathology, in which categorical diagnoses would be accorded much less importance (First, 2010). The explanation proffered for this proposed paradigm shift was that many disorders lack clear delineation and, that in the absence of boundaries, psychopathology in the real world is best captured using continuous measures. However, enthusiasm for this conceptual realignment has gradually diminished and the actual changes are likely to be much less dramatic. This is partly a consequence of the multitude of competing interests in the DSM process. For example, one of the key difficulties faced by DSM committees is to balance the need to remain independent and yet consult widely so as to ensure that they are able to incorporate constructive commentary. To achieve the latter, drafts of the diagnostic system have been available on the Internet for feedback, but the process is mind-bogglingly complex, with umpteen committees attempting to synthesise both opinion and evidence and reach consensus on issues that are intrinsically contrary.

Those involved, and those who have contributed substantively to the process of DSM development, are undoubtedly well informed, and most are world-class experts in their respective fields. But even for the psychiatry elite, developing a classificatory system that seamlessly bridges the gap between scientific data and clinical wisdom is a formidable task. In many instances, visionary changes have been forsaken for conservative alternatives – often resulting in status quo. Perhaps one of the reasons for a conservative and somewhat muted approach is the increased scrutiny applied to psychiatry, in which academics and clinicians alike have been accused of promoting disorders (Insel, 2010). Furthermore, it has been argued that DSM has provided legitimacy for disorders such as Paediatric Bipolar Disorder and that ‘fad’ diagnoses have been facilitated by big Pharma (Frances and Widiger, 2012; Parry and Levin, 2011).

Diagnostic inflation?

One of the major concerns regarding DSM-5 is its potential to increase inappropriate diagnosis. This gives rise to unnecessary prescription of medications that are likely to cause harm. Such an increase in diagnoses, termed diagnostic inflation, can occur with the introduction of new disorders and by broadening current diagnostic criteria (Batstra and Frances, 2012).

It is interesting to note that during the development of DSM-5 a variety of ‘new disorders’ were considered for coding, and that even though most have been relegated to specifiers, the fact that they were proposed as diagnoses may give rise to future problems. For example, in the context of bipolar disorder, using the ‘specifier’ mixed features does not significantly clarify how to assign the diagnosis, given that a mixed state can occur either in the context of a depressive or manic episode. Furthermore, simply ‘specifying’ that a certain number of symptoms are required, without stipulating which ones in particular, means that, ironically, mixed states will remain a heterogeneous mix.

Equally as problematic as the introduction of new diagnoses is lowering the threshold of existing disorders. This further narrows the margin between illness and normalcy and increases the risk of medicalising health by way of over diagnosis. Once again, using bipolar disorder as a prototypical example, there is no consensus on how best to define the soft bipolar spectrum (Kuiper et al., 2012), and, specifically, where its boundaries should lie. Therefore, providing this spectrum with diagnostic status would run the risk of subsuming those with normal vicissitudes of mood within a diagnosable disorder (Malhi et al., 2010).

Similarly, the at-risk category has also caused considerable concern. The concept of attempting to identify those likely to develop a diagnosis is useful because it would allow prevention and early intervention. However, it is predicated on the assumption that those ‘at risk’ of a disorder will invariably develop the illness. Unfortunately, there are currently little, or no, data to allow accurate prediction of a psychiatric illness on the basis of early signs and symptoms, or risk factors related to heritability and lifestyle. With respect to interventions, there is even less evidence to suggest which treatments are effective for those potentially at risk of a mental illness, and simply extrapolating treatments from established disorders to at-risk populations is a myopic strategy, and one that is likely to be unsuccessful.

Diagnosis per se does have some uses. For patients, a diagnosis provides acknowledgement of their illness and recognition of its associated disability. For clinicians, diagnosis allows categorisation of psychopathology such that seemingly disparate symptoms and signs can be integrated into a single diagnostic construct. However, DSM categorical diagnoses have become reified as entities, transmuting from syndromes to diseases. But the pathophysiology of psychiatric diagnoses such as schizophrenia and depression is unknown and their underlying pathology is undoubtedly heterogeneous. We know, for example, that a single pathology (e.g. syphilis) can have a myriad of clinical manifestations (e.g. rash, fever and alopecia) and a single clinical presentation (e.g. blindness) can stem from diverse pathology (e.g. diabetes, trachoma). This raises concerns as to how often the same mental illness is erroneously diagnosed as multiple co-occurring disorders and termed comorbidity. This quandary further challenges the perception of DSM disorders as distinct clinical entities.

Another key concern is that diagnosis inevitably gives rise to treatment, and usually this entails pharmacotherapy. In some countries the cost of particular medications can only be recovered from health insurers if they have been administered for nominated conditions. In other words, in order to prescribe certain medications, a specific diagnosis has to be in place. Therefore, on occasion, rather than diagnosis determining treatment, ‘therapeutic need’ may in fact determine diagnosis. Clearly, this is problematic for all parties involved.

Remarkably, pharmacotherapy has also been used to ‘define’ diagnosis on the misguided assumption that mental illnesses such as depression and schizophrenia have a specific pharmacological countermeasure. Such ‘drug-driven-diagnosis’ is clearly a flawed approach and a specific antidote for any mental illness is yet to be found. But, in practice, many DSM diagnoses are informally substantiated by virtue of a clinical response to a particular medication.

Thus, psychiatric diagnosis is a poisoned chalice and the responsibility of drinking from this cup remains the remit of psychiatrists. In reality, clinicians have to carefully negotiate a fine line between under and over diagnosis. An inaccurate or missed diagnosis is damaging because it automatically restricts the repertoire of therapeutic interventions available to the individual or denies them treatment altogether. However, over diagnosis is equally unhelpful and potentially damaging, as it pathologises ‘normal’ behaviour and mandates treatment for those without an illness.

In conclusion, and on balance, DSM is a necessary evil and the lack of a suitable alternative has led to its aggrandisement. However, as a manual it does provide a recognised taxonomy that reflects an agreed manner in which to make diagnoses, albeit with many significant limitations. In particular, while DSM continues to adopt a polythetic approach to diagnosis, its disorders need to be regarded as purely nosological quantities that should not be mistaken for disease entities. Ideally, DSM will eventually have research, legal and consumer versions and if these are properly constructed and applied judiciously they may facilitate our understanding of mental illness. Specifically, we need to determine what is mental illness? And strive to underpin psychiatric phenomenology with brain biology. However, until such a time, DSM-5 and its descendants will be the order of the day.