The anxiety of separation? Partitioning depression and anxiety in DSM-5

As noted by Goldberg (2013), there is an undeniable phenotypic association between anxiety and depression, borne out by genetic findings (Moore et al., 2000). The coterminous nature of anxiety and depressive symptoms and the fact that the respective disorders are often diagnosed as ‘comorbidities’ makes classification as prescribed in the Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition (DSM-5) an approximation that is arguably of limited value. This frustrating state of affairs is in part a consequence of the fact that our diagnoses remain contingent upon the elicitation of symptoms, which are then herded into syndromes and assigned labels. To achieve this, and to satisfy the requirements of a classification system that demands categorical separation, symptoms should fall naturally and neatly into mutually exclusive groupings. In reality, few, if any, symptoms in psychiatry are pathognomonic and, at best, only some are truly specific. Instead, most qualify simply as features on the basis of frequent association. This lack of specificity and correspondence between symptoms and diagnoses continues to hinder the meaningful classification of both the common and the low-prevalence disorders. Understanding the fundamental structure of anxiety and depression is therefore important, not only because it would enable accurate diagnosis and the administration of targeted treatment, but because it would also facilitate an understanding of the neurobiological mechanisms that drive them. Given the ubiquity of these symptoms, insights into anxiety and depression would also facilitate understanding of the other psychiatric disorders in which they feature strongly, such as schizophrenia, trauma-related conditions and substance misuse.

An influential structural model of anxiety and depression, based predominantly on phenotype, is the tripartite model of Clark and Watson (1991). It consists of unique anxiety factors, such as physiological hyperarousal and somatic tension, and depressive factors such as anhedonia and an absence of positive affect, along with shared features subsumed under a common factor called negative affect. The model has been shown to fit a variety of populations, including clinical and non-clinical cohorts, across a range of ages and contexts, and is widely supported (Prenoveau et al., 2010). However, a notable limitation is its failure to accommodate the clinical heterogeneity of anxiety and depression. To address this, the model has been extended. The integrated hierarchical model of anxiety and depression (Mineka et al., 1998) contains a broader, shared, higher-order factor termed general distress, which subsumes non-affective as well as affective symptoms. The key benefit of this hierarchical model is that it allows for a more nuanced interpretation of anxiety and a better understanding of the links between anxiety and depression. For example, within the integrated model each anxiety disorder has a unique component and this, along with more general factors, allows the development of structural models that better explain the covariance of anxiety and depression. But studies that examine the associations between anxiety and depression, either at a diagnostic or a symptomatic level, are subject to some inherent limitations. For example, examining symptoms alone yields little information pertaining to diagnoses per se, and, conversely, diagnostic-level research, which requires the application of thresholds to define categories, is unable to capture the dimensional nature of symptoms. Furthermore, it is important to note that grouping and clustering symptoms into disorders and diagnoses assumes the existence of particular associations between symptoms, and this ultimately limits the granularity with which the structural relationships between anxiety and depression can be seen at a symptomatic level.

Repetitive negative thinking, which includes worry and rumination, is a feature of many emotional disorders, and has prompted it to be viewed as a transdiagnostic phenomenon (Ehring and Watkins, 2008). Worry and rumination are two kinds of repetitive negative thinking that differ with respect to their focus, in that rumination involves dwelling on past events, whereas worry principally concerns the future. Both have been linked to anxiety and depression, respectively, and have been investigated as potential causes for comorbidity between the two sets of illnesses (McEvoy et al., 2013). Subtle distinctions between cognitions are likely to be important not only with respect to the treatment of anxiety and depression, but also perhaps their emergence in the first place.

Studies tracing the origins of anxiety and depression have shown that anxiety commonly emerges alongside subthreshold depressive symptoms (Cyranowski et al., 2012), and that depressed adolescents are more likely to experience relapse or recurrence of depression if symptoms of anxiety continue (Melvin et al., 2013). Recent functional magnetic resonance imaging (fMRI) research conducted by our group (Das et al., 2013) suggests that subthreshold emotional symptoms, in particular trait anxiety, are associated with hippocampal network dysfunction, a brain region that has been implicated repeatedly in the pathogenesis of depression. This suggests that, inter alia, chronic anxiety emanating from internal distress may be a key factor in the development of adolescent emotional disorders such as depression.

Adolescence is considered a time of increased physiological stress, and ‘normal’ heightened anxiety and stress-related symptoms occur, partly because of a differential between the developmental trajectories of emotional neural networks (Cohen et al., 2013). Specifically, prefrontal and limbic regions mature at different times, creating a transient but manifest period in adolescence during which prefrontal functioning lags behind subcortical capabilities. During this time it is possible that a prolonged failure to effectively regulate subcortical emotion processing (for instance, via prefrontal tempering) may produce a state of sustained emotional reactivity. This, if unchecked, may contribute to the onset of a depressive disorder.

The anxiety specifier in DSM-5 is therefore inadequate, especially given the importance of this putative link between anxiety and depression. As pointed out by Goldberg, the manifest existence of non-anxious depression and pure forms of anxiety lends itself persuasively to the conclusion that mixed forms of anxiety and depression are in fact an overlap of two entities. But is it possible that anxiety and depression are a single entity? Capturing comorbidity in any classification system is problematic simply because of the many combinations that can potentially occur (Tyrer, 2001), but if mixed anxiety and depression is a distinct process then perhaps it warrants separate recognition?

Tyrer proposed diagnosing ‘syndromal combined anxiety and depressive disorder’ and calling this ‘cothymia’ (Tyrer, 1989) to reflect the epidemiological prevalence of mixed anxiety and depression. Evidence for this combined approach was persuasively derived from neurobiology and genetic research, and, equally importantly, from treatment and outcome studies, which reinforced the fact that the outcome for mixed anxiety and depression is poor compared to purer presentations. Such reconceptualization is useful, although the term cothymia and a separate diagnosis were perhaps premature (Malhi, 2002). A key reason for distinguishing anxiety and depression, as noted by Goldberg, is that anxiety does not necessarily connote severity of depression. A complex, and iterative, relationship between anxiety and depression and the factors that lead to these presentations is to be expected, especially given their immense heterogeneity. The latter, again as alluded to by Goldberg, has implications for treatment.

In short, Goldberg is concerned that DSM-5 continues to neglect the significance of co-occurring anxiety, especially in the depressive disorders. How this omission has come about would be highly instructive in its own right. The omission nevertheless leads inexorably to four hard questions. In what ways does it matter? Has it made treatment inappropriate? Has it impeded an understanding of the neurobiological basis of these states? And, lastly, are there other instances in psychiatry where inattention to comorbidity may have led us astray? In clarifying the distinction between validity and utility in psychiatric diagnosis, Kendell and Jablensky (2003) pointed to the value of diagnosis in conveying information about outcome, treatment response and aetiology. The neglect of anxiety in the DSM-5 common mental disorders may be jeopardizing each of these.