DSM-5: Il buono,il cattivo,il brutto 1

I had not anticipated the excitement of thumbing through the new diagnostic and statistical manual of mental disorders (DSM-5) (APA, 2013). It had a new book smell and the allusion of knowledge, but it appeared slimmer than its predecessor, seeming to have lost weight. The key question, however, was whether it had lost substance.

Externally, DSM-5 appears much the same as DSM-IV (APA, 1994), except that it is purple in colour and the numbering of this fifth edition has changed from Roman to Hindu-Arabic. Internally, the manual has been divided into three sections, along with an appendix, and the grouping of disorders has been modified from the previous edition.

The first section includes a detailed introduction and guidance as to how the manual should be used, along with a cautionary note concerning its use for forensic purposes. This section outlines the deliberations of the ‘hundreds of people working’ … ‘over a 12-year process’ and touches on the need for reliability of diagnoses and the contextual use of DSM-5 alongside clinical experience. It carefully outlines the revision process, beginning with proposals, which were developed by members of working groups, for disorders in Section II. This process was guided by several principles: maintaining a focus on clinical practice to be informed by research findings, and the freedom to implement as much change as necessary, whilst maintaining as much continuity with DSM-IV as possible. Working with these standards in mind, working groups identified important issues within their designated diagnostic fields and also considered issues from their area that might impact other fields and diagnoses. A key task for the development of Section II was to appraise contradictory findings regarding diagnoses and also to consider the need for new diagnoses or subtypes and specifiers (APA, 2013).

Section I also contains an explanation of the role of Section III, which has been reserved for diagnoses that have clinical potential but that are yet to attain validity. The hope is that their inclusion in DSM-5 will encourage and facilitate the further research that is necessary before these categories can become fully-fledged disorders. Interestingly, these ‘diagnoses’ are not intended for clinical use. Section I additionally describes the use of field trials and expert, public and professional review and the goal of harmonizing DSM-5 with ICD11. Finally, it discusses how the manual has attempted to incorporate a dimensional approach to diagnosis, along with developmental and lifespan considerations.

The magnitude and detail of DSM-5 mean that it is beyond the scope of this Editorial to provide a comprehensive appraisal of the manual. But, given its pivotal role in research and clinical practice worldwide, it is useful to consider briefly some of its key aspects, while it is still new and undergoing critical evaluation.

Il Buono: The Good

The sheer size and depth of the manual are impressive and reflect the fact that this is the culmination of enormous effort. The production of DSM-5, or its ‘creation’ as it has been described, was a gargantuan undertaking that involved literally hundreds of people from more than a dozen countries. This army of contributors was drawn from a range of disciplines in addition to psychiatry, such as neurology, psychology, primary care, epidemiology, statistics, paediatrics and those with particular areas of expertise in research. The development process involved more than a dozen international conferences, many of which were conducted in conjunction with the World Health Organisation (WHO) Division of Mental Health and Substance Abuse and with support from the American Psychiatric Association (APA) and National Institute of Health (NIH). This process therefore spanned more than a decade. DSM-5 taskforces and work groups focusing on particular disorders also attempted to have a breadth of representation involving international members alongside members from the United States, so that cultural aspects could be considered and, where appropriate, incorporated. This collaborative, multidisciplinary approach warrants recognition and, to ensure and underscore this, a DSM-5 Culture and Gender Study Group was assigned to create guidelines for the working groups (Regier et al., 2013). Concurrently, one eye has also been kept on the development of ICD11, which has a tentative publication date of 2015, so as to harmonize diagnoses, where possible, so that statistics have more widespread application and research can be meaningfully translated from one classification system to the other. In a similar vein, DSM-5 also pays greater attention to gender and age, noting that these factors undoubtedly alter the manifestation of symptoms and the trajectory and course of psychiatric disorders. Therefore a deliberate attempt has been made to emphasize the importance of considering lifespan, developmental factors and gender when making a diagnosis. These are all positives.

Even the overall organisation of DSM-5 has been re-examined and given careful consideration. Groups of disorders have been ordered to reflect the strength of potential linkages. For example, Bipolar and related disorders, now separate from Depressive disorders, is positioned between the latter and Schizophrenia spectrum and other psychotic disorders because of its shared neurobiology and clinical overlap with both groups of disorders (Owen et al., 2007; Goldberg et al., 2009a; Goldberg et al., 2009b). Its positioning is thought to ‘link the two groups of disorders’, even though many consider bipolar disorders to be predominantly a component of mood disorders alongside depression. Interestingly, mood disorders, as a group, is no more (Regier et al., 2010; APA, 2013).

For similar reasons, depressive disorders and anxiety disorders have been juxtaposed and the latter then serves as a conduit to obsessive-compulsive and related disorders. Clinically, this is logical, but the failure to additionally define ‘mixed anxiety and depression’ as a disorder, or at least to include it as a condition for further study, is baffling. Two further groups, Trauma- and Stressor-related disorders and Dissociative disorders, complete this group of five (along with depressive, anxiety and obsessive-compulsive disorders) collectively termed internalising disorders. These disorders that have prominent anxiety, depressive and somatic symptoms and are seen as having shared genetic and environmental risk factors that drive negative affectivity are conceptualized as distinct from externalising disorders, which include Disruptive, Impulse-control, and Conduct disorders and Substance-related and Addictive disorders. Sandwiched between these two ‘dimensions’ are disorders (Somatic symptom and related disorders, Feeding and Eating disorders, Elimination disorders, Sleep wake disorders, Sexual dysfunctions and Gender dysphoria) collectively referred to as Somatic (APA, 2013).

While interesting and derived from a rationale of sorts, it is important to bear in mind that the evidence that underpins these associations and linkages is tentative at best. The strongest indications of shared biology come from recent genetic studies that have found shared liability across mood disorders and psychosis (Owen et al., 2007) and similarly within neurodevelopmental disorders such as Autism Spectrum Disorder (ASD) and Attention Deficit/Hyperactivity Disorder. However, these minor criticisms aside, the overall structure flows reasonably well.

Another move in the right direction is the explicit recognition that disorders and categories as defined by DSM are not distinct entities that connote discrete diseases/illnesses. Furthermore, in keeping with the realization that any underlying pathophysiology of psychiatric disorders is likely to exist on a continuum from normalcy to disease, so too there is awareness that symptom expression is also dimensional. Consequently, DSM-5 has attempted to integrate dimensionality into diagnosis by introducing measures that can be used in conjunction with diagnostic categories (Regier et al. 2013). For example, once a categorical diagnosis has been made it can be ‘dimensionalized’ using subtypes, specifiers and severity ratings, and to this end specific rating scales and measures that can be used to rate severity and to monitor long-term outcome and treatment response have been made available online. Additional cross-cutting assessments, which allow dimensional assessment across disorders, and provide a means for quantitative comparisons, have also been included, and perhaps an indirect consequence of such dimensional diagnoses will be to capture subsyndromal presentations (Clarke et al., 2013; Narrow et al., 2013).

Overall, these developments are welcome, but their true value will only become apparent once they have been tested in real-world settings. It will therefore be interesting to observe how often these additional metrics are actually used in practice, and whether they provide clinically meaningful information.

Il Cattivo: The Bad

The new DSM-5 is like an unwanted present: you’re gladdened by the thought, but remain unsatisfied because it’s not what you had hoped for. Even a cursory examination of the manual reveals that it is not what was promised. Firstly, many of the changes are cosmetic and the majority of diagnoses have remained much the same. Secondly, DSM has dispensed totally with the notion of incorporating neurobiological markers and restructuring DSM according to pathophysiology. Finally, the changes that have been made are not necessarily improvements. Clearly, many things altered along the way, from the time of inception to the eventual realization of the classification system.

The DSM-5 working groups became increasingly aware of a fundamental problem as they began the task of proposing revisions and considering how to describe the world of psychiatry: our understanding of disorders, in terms of aetiology and neurobiology, remains woefully incomplete. The anticipated advances in neuroscience that would inform clinical psychiatry have simply not eventuated and, in many cases, these necessary insights are not even on the horizon. Therefore, one way of adding sophistication is to create specifiers, such as ‘with anxious distress’ and ‘with mixed features’. Specifiers are not diagnosable or codeable disorders in themselves, but allow nuance to be added to the diagnoses by denoting a specific manifestation. DSM-5 suggests that this will allow subgrouping of more homogeneous diagnoses and that, in conjunction with subtyping, the use of specifiers will identify variants that have greater clinical significance. Furthermore, subgrouping in this manner will facilitate research into more specific diagnoses. However, this is unlikely because, ironically, the specifiers themselves lack specificity (Malhi, 2013). Note that specifiers are different from subtypes: the latter are ‘mutually exclusive and jointly exhaustive phenomenological subgroupings within a diagnosis’; specifiers are not, meaning that more than one specifier can be assigned to a diagnosis. This immediately creates a multitude of possible diagnoses, as does the fact that nomination of symptoms for the mixed symptom specifier, for example (which can be applied to both major depression and mania/hypomania), can comprise a variety and variable number of symptoms, and therefore mixed states ‘defined’ in this manner will necessarily tend towards being heterogeneous (Malhi, 2013). In clinical practice specifiers are likely to be applied relatively loosely and may even become shorthand for various ‘kinds’ of presentations; this will further blunt the accuracy of diagnoses. Therefore, the expectation that loosely defined, optional qualifiers will somehow fuel research and improve clinical understanding is unrealistic and misguided.

Remarkably, DSM-5 has managed to create a number of new disorders (Frances, 2013). One of the reasons proffered in the case of Disruptive mood dysregulation disorder (DMDD), for example, is to stem the over-diagnosis of bipolar disorder. The diagnosis of bipolar disorder in adolescents and children is a contentious issue that has been the subject of vehement debate over the past decade. All the while an increasing number of children and adolescents have been labelled as bipolar. It is therefore appropriate for DSM-5 to tackle this problem; indeed it is particularly fitting given that DSM-IV contributed to its development in the first place. However, it is unclear how renaming a subset of children in whom the onset of symptoms is between the ages of 6 and 10 years will help. DSM-5 argues that a key ‘distinction’ has been drawn on the basis of whether irritability is persistent or episodic, such that children with persistent irritability, along with extreme behavioural and/or verbal outbursts, will no longer be regarded as having bipolar disorder but instead will be diagnosed with DMDD. To emphasize this distinction further, DMDD has been placed within depressive disorders. Clinically, DMDD is thought to peter out with age as the individual transitions to adulthood, but it is unclear whether the condition occurs solely, or even predominantly, in this age-bracketed manner. This fundamental diagnostic uncertainty is cause for concern.

Equally as problematic as the ‘creation of disorders’, is the elimination of criteria that safeguarded against inaccurate and over diagnosis. Removal of the bereavement exclusion, for example, within depressive disorders, is a major concern that has been recently discussed in the Journal (see Porter et al., 2013).

Therefore, despite good intentions, DSM-5 has implemented changes that are likely to have serious consequences for all concerned, and for patients in particular.

Il Brutto: The ugly

In its defence one could argue that DSM-5 has attempted to provide an objective diagnostic classification and that this was its remit. However, because of its significant and widespread influence, past experience has shown that even subtle changes can lead to dire problems that only become apparent retrospectively, many years later. However, DSM-5 is not wholly responsible for its biggest failing, namely, its inability to tether clinical diagnoses to the neurobiology of the brain. This would only have been possible if a profound understanding of the pathophysiology of psychiatric disorders had been achieved, and the responsibility for this lies more broadly with the research community as a whole. In fairness, the human brain is by far the most complicated thing known to man and a full understanding of its inner workings was too ambitious a goal. The ugly truth is that, for the foreseeable future, we will have to rely on phenomenology alone to describe and define psychiatric disorders, as we have done for a century or more. A lack of knowledge of the intrinsic workings of the brain has hamstrung the creators of DSM-5, and the same limitations will impact the development of ICD11. Perhaps, therefore, it was not necessary to revise DSM-IV? But now that it has been done the key question is: are the changes made likely to be of benefit? The answer to this is unknown with views being firmly divided as to what is likely to unfold. Indeed, many continue to argue that, without any clear indication as to what to change, and how to change it, this revision of DSM-IV is difficult to justify other than in financial terms. Two of the main goals of DSM-5 were to improve diagnosis and to capture new information that has come to light since DSM-IV. However, neither genetics nor neuroimaging have managed to provide an underpinning to clinical disorders. If anything they have served to highlight the sheer complexity and sophistication of the brain and the multitude of factors that possibly lead to the development of psychiatric disorders. New areas such as oxidative neurobiology, though promising, are exactly that, new and promising, and hence are too nascent to offer any robust leads. The widespread criticism of DSM-5 is therefore perhaps understandable.

l’imprevedibile: The unpredictable?

Frustrated with the current approach to nosology, and frustrated by DSM-5, NIMH has launched a project called the Research Domain Criteria (RDoC) (Insel, 2013). The aim of this is to enhance the validity of diagnoses. In medicine, diagnoses (e.g. cancer) have a known pathophysiology (e.g. unchecked cell growth), which can be measured directly (e.g. biopsy, imaging) or indirectly (e.g. blood-borne markers). The purpose of RDoC is to make diagnosis more objective and more ‘lab-based’ by integrating neurobiology and symptomatology. Specifically, this approach intends to incorporate information from domains such as cognitive neuroscience, neuroimaging and genetics and to map this, along with clinical information (including symptoms), to identify ‘new categories’ that are an amalgam of all of these domains. This approach necessarily steers away from DSM, partly by design, but also because thus far attempts to correlate DSM disorders with underlying biomarkers have been largely unsuccessful, reflecting the fact that DSM disorders are not psychiatric diseases. Therefore the framework that RDoC provides may stimulate new avenues of research and identify treatments that address specific neurobiological dysfunctions. However, this is a new direction and one that will take time to gain momentum and to make discoveries; once again, the outcome is unpredictable. In the interim, and possibly for many years to come, DSM-5 and its progeny will govern clinical practice and many other aspects of mental health care.

The principal strength of DSM is its reliability, which stems from a consistent approach to making diagnoses. This has ensured a common language for psychiatrists and mental health professionals and has benefitted everyone enormously. However, the greatest weakness of DSM is that it lacks validity, and the best choice of treatment cannot be simply determined on the basis of symptoms alone. Fortunately, the changes in DSM-5 are few and relatively modest, reflecting DSM’s conservative and pragmatic approach. In the long term, the DSM approach is unlikely to advance our understanding of psychiatric disorders and, to achieve a better system of classification that is good for patients, we are going to have to tackle all that is bad in our current practice and nosology even if this means that things get ugly.