Mood disorders are categorically dimensional!

The debate regarding whether mood disorders are distinct entities, separable from normalcy and from each other, or whether instead they are arbitrary divisions along what is actually a continuum is age-old. In recent years, the pendulum has once again swung in favour of a dimensional model of mood disorders – referred to more commonly as a spectrum (Ghaemi and Dalley, 2014). However, opinion regarding whether this is the best model for understanding mood disorders continues to oscillate, because the issue remains unresolved. Not only is there debate about whether there is a distinct break between normal and mood disordered but also whether there are distinct categories of mood disorder, for example, recurrent major depressive disorder (MDD), bipolar I disorder (BD-I) and bipolar II disorder (BD-II). At the root of this debate is the problem that characterising the fundamentals of mood disorders has proven to be far more difficult than anticipated.

The axiomatic problem is how to investigate an entity the nature of which is unknown. Clinically, the extremes of mood – catatonic depression and psychotic mania, are clearly aberrant, but lesser gradations are less distinguishable – especially when short-lived (Sara and Malhi, 2015). Hence, in practice, for the purposes of clinical description, classificatory coding and investigational studies, diagnoses have to be made with a degree of replicability, and, to achieve this, the key features of a mood disorder and their severity need to be specified. This pragmatic approach has meant that classification has relied heavily on the presence of particular symptoms that together form syndromes, which then form the foundation for diagnostic categories and for ratings of severity. Cut-offs are then used to define whether an individual has a mood disorder.

The problem with this approach is that having to acquiesce to clinical and research needs has meant that a key problem has been circumvented rather than solved. First, rating symptoms using discrete cut points assumes a priori that the various dimensions of mood disorders can be meaningfully broken up into units and that there are distinct discontinuities. For example, for a Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition (DSM-5) diagnosis of major depressive episode (MDE), a patient must have depressed mood ‘most of the day, nearly every day’, but it is evident that there is not a clear cut point which defines which patient is depressed for enough of the time to trigger a diagnosis of MDE. This system also implies that there is continuity between normalcy and abnormal mood, with ‘normal’ low mood differing only by virtue of the amount of time it is experienced for. But, in practice, those with clinically depressed mood often describe the experience as qualitatively different to the symptoms of low mood associated with ‘normal’ experiences that are usually transient and do not cause functional impairment – such as being disappointed (Malhi et al., 2015). The connotation is that the very fabric of these mood states is different, and that they are not on a continuum or similar plane. In other words, clinical depression is an altogether different experience from variations in mood that occur normally, and that it perhaps involves completely separate or at least additional processes. This suggests that depression is possibly on a spectrum with other forms of mood disorders such as mixed states and mania, but mood disorders as a whole are not necessarily on a continuum with normal mood.

Second, when attempting to form categories by dissecting dimensional constructs into units for the purposes of measurement, an important question that arises is whether these components can be weighted equally. This applies both to the symptoms being rated in diagnostic systems and to those being rated in rating scales designed to measure severity. In DSM-5, there is an explicit system whereby not all symptoms are equal. For example ‘depressed mood’ or ‘diminished interest’ have to be present for a diagnosis of major depression while other symptoms only have to be present in sufficient numbers. However, there has been little research into what constellation of symptoms most reliably defines an entity which is different from ‘normal’.

In contrast, in rating scales used to measure mood disorder ‘severity’, each item is equal. For example, in the Hamilton Depression Rating Scale (HAM-D), fidgetinesss ‘equals’ feelings of ‘life is not worth living’ (both score 1) and attracts a similar rating. But is this correct, and is it clinically meaningful? Furthermore, when assessing individual dimensions of mood such as rating guilt using the HAM-D, it is scored discretely and incrementally from ‘self-reproach’ and ‘ideas of guilt’ to ‘delusions’ and ‘hallucinations’. But these divisions are not necessarily equal or indeed on a linear gradient of severity.

Psychiatry is not at fault because this approach is the scientific method used in medicine as a whole (Kapur et al., 2012). A disease entity is first defined on the basis of characteristic features (symptoms and signs) and then investigations are conducted to develop further tests that pin down the disease processes biologically. In psychiatry, we have become interminably stuck in the first phase of this process – partly because there are many views as to which symptoms are important, and how these should be clustered. This phenomenology – syndrome – diagnosis paradigm that implies the existence of underlying disease entities has, on the whole, failed to advance our understanding of mood disorders, and disillusionment with this approach (Craddock and Owen, 2010), for which the Diagnostic and Statistical Manual of Mental Disorders (DSM) and the International Classification of Diseases (ICD) have been emblematic, is part of the reason that researchers in the United States proposed Research Domain Criteria (RDoC). Briefly, this pursues individual phenomena across putative phenotypes and through inquiry at all levels’ attempts to explain symptoms in terms of underlying dysfunction in neurocircuits, neuronal cellular processes and genetic vulnerability. But this too is proving to be difficult, primarily because it does not solve the problem of having to capture a dimensional concept that is manifest clinically using a means that is measurable and reproducible with reliability.

The field therefore needs to decide first which domains are important alongside mood, and we would argue that activity and cognition are equally important if not more so than other symptoms given equal prominence. This is exemplified in evidence in mixed mood states and in melancholia, in which these symptoms appear to have a degree of predictive value. Second, if there are ‘natural’ breaks in these dimensions, these need to be identified so as to develop measurable components that can be usefully grouped and examined en masse. We argue that a new approach to research into mood disorders is required, perhaps similar to that advocated in RDoC but one that also reappraises the symptoms currently used in diagnostic systems and rating scales. The critical question which needs to be asked is – in a broad group of people who complain of mood-related symptoms, regardless of current classification – which symptoms or constellations of symptoms determine different neurobiology, course of illness and response to treatment. When this is known, then diagnoses may be re-drawn either as distinct entities or dimensions.