Can separation anxiety disorder escape its attachment to childhood?

The definition of separation anxiety disorder (SEPAD) has undergone significant changes in DSM‐5, the most important being the lifting of the age restriction (18 years of age in DSM‐IV) for assigning the diagnosis. There may be resistance, however, amongst some clinicians and researchers to extending the diagnosis to adulthood. We consider the arguments in favour and against this change in the hope of stimulating debate and research aimed at achieving a consensus on this issue.

Why do clinicians traditionally restrict the diagnosis of SEPAD to childhood (here used broadly to cover the period from infancy to early adolescence)? The main reason is that the construct of separation anxiety (SA) has long been central to developmental theories that exert a strong influence in guiding clinical practice. Within the broad developmental framework of psychoanalytic and attachment theories, SA is regarded as representing a repertoire of neurophysiological, intrapsychic and behavioural responses specifically designed to protect children from danger by ensuring the maintenance of close proximity to an adult caregiver, typically the mother. The SA mechanism is of particular importance to our species because of the prolonged period of dependency of the child on the caregiver1. In attachment theory, heightened expressions of SA are regarded as indicating disturbances in the child's working models or internal representations of attachment figures, shaped by past and ongoing bonding experiences with primary caretakers2. SEPAD as a diagnosis therefore lies at the extreme end of a spectrum of responses that extend from the normative to the pathological, its presence signifying that the child has been exposed to severe disruptions and/or disturbances in his/her primary bonds2. Classical symptoms of SEPAD (excessive clinging, tantrums, school refusal, abdominal pain and headaches, refusal to sleep alone, and nightmares of being attacked or abducted) reinforce further the phase‐specific nature of the response.

Yet attachment theory has long acknowledged that the drive to form and maintain close bonds is fundamental to humans throughout the life course3. The corollary must be that the SA response can occur in persons of all ages. Indeed, reciprocity in the SA response between the mother and the child is critical to the mechanism's protective function; by mirroring the alarm signals of the lost child, the mother's anxiety ensures that she engages in intensive searching behaviour to rescue the young person from potential harm. More generally, in collective species such as homo sapiens, the drive to maintain proximity to close others is fundamental to ensuring the survival of individual members1.

In summary, there is an evident tension within attachment theory between the tendency to regard SA as a specific characteristic of childhood and the recognition that attachment anxiety extends throughout the life course. From a clinical perspective, Bowlby's developmental model of agoraphobia provided a partial resolution for this problem. He proposed that, if high levels of SA persisted into later years, they manifested as typical symptoms of agoraphobia4. According to this model, symptoms such as carrying transitional objects, reliance on phobic companions, and the preference for staying at home (as a symbol of a secure base) reveal the underlying SA roots of adult agoraphobia4.

Initially, empirical research provided support for the SA‐agoraphobia model; in a series of studies, adult patients with agoraphobia reported much higher levels of early SEPAD (assessed by the proxy indicator of school phobia) in their early lives compared to those with other anxiety or depressive disorders5. The SA‐agoraphobia model became firmly embedded in developmental theory over time, incorporating panic disorder as an adult outcome when DSM‐III linked that category to agoraphobia. Since then, researchers have searched for evidence of a common biological substrate underlying SEPAD, panic disorder and agoraphobia, by examining the family aggregation, shared pattern of genetic inheritance and distinctive psychophysiological responses associated with the three constellations6, 7.

In parallel, however, other studies have produced evidence that calls into question the SA‐agoraphobia model. In particular, several studies have found that the link between early SA and panic disorder/agoraphobia is not specific, but represents a general characteristic of adults with a range of anxiety and depressive disorders8. Two decades ago, observations at a clinic for anxiety patients at the University of New South Wales led to the formulation of an alternative developmental model of SEPAD9. The team found that, when symptoms were specifically inquired into, many adult anxiety patients revealed the presence of SEPAD, commonly dating the onset of the problem to childhood9. This discovery suggested a continuity model in which SEPAD was a disorder that extended across the life course, although symptoms showed pathoplastic changes commensurate with maturation. For example, adults feared for the safety and whereabouts of a wider range of attachment figures, including parents, romantic partners and spouses. Moreover, symptoms manifested in more subtle ways: for example, adults employed complex rationalizations to avoid work or travel and tended to find pretexts to make repeated phone contact with attachment figures throughout the day.

Following these observations, several measures were developed to assess SEPAD in adulthood9, 10. The clinic‐based studies that followed indicated that 20‐40% of patients attending ambulatory facilities met criteria for SEPAD10, 11. The relationship between reported early SA symptoms and adult SEPAD proved to be highly specific; once that relationship was accounted for, there was no evidence to support a specific link between SA and panic disorder or agoraphobia.

A recent analysis of the World Mental Health Survey dataset indicated that the lifetime prevalence of SEPAD across countries approximated 5%; persistence of the disorder into adulthood was common; and adult onset occurred in 40% of all cases12. SEPAD showed a high level of comorbidity with a range of common mental disorders, not specifically with panic disorder and agoraphobia. Adults and children with SEPAD reported a consistent pattern of disturbances in their early family lives and high levels of exposure to a wide range of traumas12. Taken together, these findings offer support for the model of SEPAD proposing that symptoms in adulthood commonly represent the continuation or recurrence of those experienced in childhood.

Why, in the face of these recent findings, has the SA‐agoraphobia model persisted? Several factors are likely to be at play. The overriding reason is that adherence to established developmental theory discourages clinicians from recognizing SEPAD symptoms in adults. Also, by its very nature, SEPAD occurs within an interpersonal field, involving the family and close attachments. It is common in clinical practice to find that close attachments accommodate and adapt to the person's SEPAD‐related fears, particularly as the anxieties are directed at safeguarding others13. A pattern of collusion therefore may arise in which the person with SEPAD, the family, and ultimately the clinician, all underestimate the role of SEPAD symptoms as a source of dysfunction in the patient. Definitional overlap in symptoms, particularly between agoraphobia and SEPAD, may further confound the picture. SEPAD may also occur in response to the disruptions and losses associated with other severe mental disorders, such as bipolar disorder14. In these contexts, the mood‐related symptoms will often overshadow those of SEPAD which, as a consequence, will go undetected, even though they add to the person's overall disability. Severe SEPAD may also present in a variety of ways – for example, as suicidal behaviour or stalking in response to actual or threatened separations – which are not indicated in the DSM‐5 criteria for the disorder.

In the end, only one of the two developmental models outlined herein, the SA‐agoraphobia model and the continuity model, can be valid. Resolution of this issue is not merely one of theoretical importance. SEPAD in adulthood is associated with high levels of disability and signifies a poor response to treatment when conventional pharmacological or cognitive behavioural therapies are used to treat comorbid anxiety disorders11, 12. As a consequence, there may be a substantial cost in disability and suffering by overlooking the diagnosis of adult SEPAD. The critical question, therefore, is whether the DSM‐5 reformulation of SEPAD is a turning point that will release SEPAD from its over‐attachment to childhood.

Derrick Silove1, Vijaya Manicavasagar2, Stefano Pini3
1Psychiatry Research and Teaching Unit, School of Psychiatry, University of New South Wales, Sydney, Australia; 2Black Dog Institute, Prince of Wales Hospital, Randwick, Sydney, Australia; 3Department of Clinical and Experimental Medicine, University of Pisa, Pisa, Italy