DSM-5 allows separation anxiety disorder to grow up

“A 23-year-old woman… was referred to an anxiety clinic because her symptoms intensified every time she had an argument with her partner or when the latter went out alone… At the patient’s insistence, her partner had moved in to live with her in her parents’ home… She continued to be anxious about her parents leaving the house and would imagine all sorts of disasters befalling them when they were out. She was plagued by similar fears in relation to her partner. Six months prior to presentation, the patient’s separation anxiety symptoms worsened substantially and she was obliged to give up work… She insisted on accompanying her partner everywhere… her insistence on ‘shadowing’ her partner led to arguments and threats by the latter to desert the relationship. It was at that point that the patient began experiencing panic attacks…” (Manicavasagar and Silove, 1997).

As illustrated by this case excerpt, the team at the Psychiatry Research and Teaching Unit, the University of New South Wales (UNSW), began to describe separation anxiety disorder (SAD) occurring in adulthood in the mid-1990s. Consistent with the characteristics of the childhood category, the core features of adult SAD involved fears of separation and/or harm befalling close attachment figures (Manicavasagar et al., 2000). As expected, symptoms were modified by the stage of maturation; for example, whereas the focus of childhood SAD tends to be exclusively on parents, adult fears encompass a wider network of attachment figures, including spouses, children, parents and romantic partners (not to mention dogs). Rather than overt behaviours like crying and clinging to attachment figures, adults exhibit their fears in more subtle ways, for example by delaying or avoiding leaving home to go to work, and finding reasons to maintain close proximity to or contact with family members, perhaps by phoning repeatedly throughout the day to check on their safety. The recent changes to the category of SAD in the fifth edition of the Diagnostic and Statistical Manual of Mental Disorders (DSM-5 Task Force, 2012) reflect the above observations.

The prevalence of adult patients with SAD in anxiety clinic populations appears to be substantial, as is the level of disability associated with the disorder (Silove et al., 2010b). Yet none of the patients with SAD had been so diagnosed by the mental health professionals with whom they had previously consulted, even though many of the patients themselves had been convinced that the diagnoses they had received, such as panic disorder or agoraphobia, did not accurately reflect their core anxieties (Manicavasagar and Silove, 1997; Manicavasagar et al., 1997).

Why the delayed maturation of SAD?

An important question is: Why has it taken psychiatry so long to recognize that SAD can occur in adulthood? The prevailing tradition of developmental theory undoubtedly has played a role, with the established canon tending to regard SAD as a characteristic of infancy and early childhood. For example, Winnicott (2002) regarded the task of separation-individuation as central to the psychological development of infants and young children, with inadequacies in the care given by the mother laying the foundations for separation anxiety and via that route to future psychopathology. Attachment theory extended this theme, drawing on evolutionary principles to argue that attachment to parental figures in early life is vital to survival in a species where there is a prolonged period of dependency on care-givers (MacLean, 1985). SAD therefore represents a core safety signal that alerts the infant and the care-taker to imminent threats to the security of primary bonds. Concurrent research in neurophysiology led leaders in the field such as Paul MacLean to conclude that the fear of separation, and the behavioural reactions that it triggers, are of such importance to survival that the establishment of the neural substrates that underpin these survival mechanisms were fundamental to the evolution of the human brain (MacLean, 1985).

One of the most influential and enduring hypotheses arising from attachment theory was Bowlby’s hypothesis that an anxious attachment style was the underlying psychopathology linking SAD in childhood, commonly manifesting as school refusal, and agoraphobia in adulthood (Bowlby, 1973). The separation anxiety (SA) – agoraphobia (Ag) developmental model has evolved over time, particularly with the close link established in DSM-III between panic disorder (PD) and Ag. The SA–PD–Ag hypothesis has been tested empirically over time, with most studies relying on the retrospective reports of adults about their early anxieties. Klein’s (1964) seminal studies supported the model by indicating that a substantial number of adult patients with PD–Ag reported a history of childhood SA. Neurophysiological and genetic studies appeared to support the link by demonstrating a common underlying diathesis to both SA and PD–Ag (Roberson-Nay et al., 2012), although this study only compared the associations between childhood overanxious disorder or SAD with adult onset panic attacks. Nevertheless, over time, studies in the field began to produce contradictory results, particularly when patients with a wider range of adult anxiety subtypes were included (Biederman et al., 2005; Lipsitz et al., 1994). These latter studies have cast doubt upon the idea that early SA is specifically associated with PD–Ag, suggesting instead that it represents a general risk factor to a range of adult anxiety (and other) disorders.

Research on the adult form of SAD

Systematic research on adult SAD undertaken at UNSW has helped to resolve this issue. In order to test the SA–PD–Ag model, we developed a retrospective measure of SA, the SASI (Silove et al., 1993), and later the self-reported Adult Separation Anxiety Questionnaire (ASA-27) (Manicavasagar et al., 2003), which was designed to assign a diagnosis of SAD in adulthood. In studies based on clinic samples of anxious and general psychiatric patients studied in Sydney and Italy, the prevalence of adult SAD has varied from 20 to 40%, depending on whether a strict or wider definition of the disorder has been used (Pini et al., 2009; Silove et al., 2010b). A large-scale epidemiological study (the National Comorbidity Survey Replication, NCS-R) conducted in the USA recorded a lifetime prevalence of SAD in adulthood of 6.6%, indicating that the category is one of the most common forms of anxiety (Shear et al., 2006). Importantly, a large number of cases in that study reported the onset of SAD in early adulthood. Clinic studies suggest that SAD in adulthood is disabling (Silove et al., 2010b) and, when occurring comorbidly with other disorders, is associated with a poor response to standard cognitive behavioural therapies (Aaronson et al., 2008).

Our clinic studies have examined the association between SAD in adulthood and reports of heightened SA symptoms in early life, based on the SASI. The findings have been consistent in showing a strong association between early SA symptoms and a diagnosis of SAD in adulthood. Moreover, once the association with adult SAD is accounted for, there is no direct link between early SA and PD–Ag or any other adult anxiety subtype (Manicavasagar et al., 2000). These observations offer an explanation for the ambiguous findings produced by the body of previous studies testing the SA–PD–Ag hypothesis, none of which took into account the presence of adult SAD. Once that newly discovered category was included in the analysis, the pattern became clear: a specific and unique relationship emerged between childhood and adult SAD. Previous studies showing an association between SAD in early life and PD–Ag may therefore be explained by the pattern of comorbidity amongst the adult anxiety disorders: the apparent link between SA and PD–Ag may be explained by comorbidity between the latter category and adult SAD, as has been found in some studies (Pini et al., 2009; Silove et al., 2010b).

SAD also appears to aggregate across the generations within families. For example, a small study undertaken in a childhood anxiety clinic found a strong and exclusive relationship between childhood and parental SAD, but no association between childhood SAD and parental PD–Ag (Manicavasagar et al., 2001). In summary, existing data suggest that SAD can occur across the lifespan and that there is a specific developmental relationship between the childhood and adult form of the disorder.

Attachment theory and SAD

Over recent decades, attachment theory has expanded its scope beyond infancy and childhood to attachment styles in adulthood and related relevance to psychopathology in later life. The discovery of SAD in adulthood may be regarded as a parallel development that consolidates the notion that anxious and other forms of distorted attachments are of fundamental importance to human adaptation across the lifespan – and not simply an issue of relevance to understanding developmental psychopathology originating in infancy. At the same time, it is important to caution against an assumption that SAD and an anxious attachment style are synonymous constructs. As a diagnostic category, SAD is a nomothetic construct based on the coexistence of operational symptoms; in contrast, anxious attachment is an idiographic construct whose meaning derives from its explanatory function within attachment theory. Although studies have shown that persons with adult SAD do score higher on indices of anxious attachment, the association is modest (Manicavasagar et al., 2009). In addition, some of the findings in the recent literature point to associations between SAD and other forms of psychopathology that may be unanticipated. For example, in a study of trauma-exposed Bosnian refugees, adult SAD was strongly associated with PTSD but not with symptoms of traumatic grief (Silove et al., 2010a). Fears for the safety and security of the self are closely linked to fears for others amongst the symptoms of SAD, offering a possible explanation for its overlap with PTSD. Indeed, in some instances, SAD may be the opposite side of the coin of PTSD, particularly in settings where the initiating trauma posed a threat to close others. As such, as a diagnostic category, SAD may prove to have complex origins that are not necessarily aligned with the constructs of attachment theory.

DSM-5 allows SAD to grow up

Modifications to the proposed criteria in DSM-5 reflect the aforementioned findings of research undertaken on SAD over the past 15 years. If adopted, the proposed changes will support the process of SAD growing up. First, DSM-5 will discontinue the section for disorders that have their first onset in childhood, where SAD was located in DSM-IV-TR. Instead, SAD will join the other subtypes of anxiety disorders in the general section for the anxiety disorders. Moreover, a key change in the criteria is the removal of the stipulation that onset of the disorder must be before the age of 18 years, a modification that is consistent with clinical and population-level research evidence (Manicavasagar et al., 1997, Shear et al., 2006). Finally, symptoms have been reworded to remove any implicit bias towards childhood, for example by widening the target of SA fears to a range of attachment figures instead of only listing parents, and by expanding the repertoire of avoidance behaviours to adult settings such as work, rather than focusing primarily on school refusal. In concert, these changes have the important effect of aligning SAD with the existing anxiety subtypes, particularly by removing any reference to age of onset in the criteria for diagnosing the disorder. These changes have a far-reaching impact on the interpretation of past epidemiological and clinical studies involving the anxiety disorders. In effect, the changed status of SAD means that the epidemiology, patterns of comorbidity and risk factor profiles of the anxiety disorders need to be updated because it is likely that past studies that have not taken into account persons with adult SAD have overlooked or misdiagnosed a substantial number of individuals included in the analyses.

It is to be expected that changes in the classification of mental disorders will be greeted with controversy and at times acrimonious disagreement, especially if categories are removed or expanded. The birth pangs of DSM-5 have not been an exception to this rule. It is of interest, therefore, that the modifications proposed for SAD have not attracted any public or professional debate, possibly because the evidence supporting these changes is relatively strong. Yet the maturation of SAD has profound implications for research and practice, opening up important questions about the origins, patterns of comorbidity and consequences in relation to disability for adults who can now legitimately be assigned that diagnosis. As yet, there are no treatment studies testing interventions for SAD in adulthood, a remarkable hiatus in itself. Hopefully, the changes ushered in by DSM-5 will act as a catalyst for research aimed at establishing interventions for SAD in adulthood. Perhaps the most profound immediate effect of allowing SAD to grow up, however, is that it acknowledges the suffering of countless patients, many of whom have contacted the present authors over time to express their despair at not having their core anxieties properly recognised or understood by the mental health professionals with whom they have consulted.