John Bowlby and contemporary issues of clinical diagnosis

Introduction

The Wellcome Trust Library in Euston contains the Bowlby Archive, with John Bowlby’s unpublished papers, correspondence and notes. In the Archive there is an undated transcript of a talk to the Institute of Psychoanalysis, containing a recollection of Bowlby's first child patient:

After qualifying in psychoanalysis I started child training in probably 1938, and I suppose Mrs R[iviere] was keen that I should have analysis at that time. Well, I was allocated a small boy of about three who was hyper-active, anxious, aggressive, all over the shop, running in and out of the room, and this and that. His mother was very anxious, fidgety, troubled. Her job was to sit in the waiting room whilst I analysed this boy. Because of my experience with the Child Guidance Clinic I was very concerned that we were leaving out half the problem, but I was under strict instructions to have nothing to do with the mother, which went very much against the grain. Well, after I’d been seeing this boy for about 3-4 months [inaudible] to the effect that this boy’s mother had been admitted to a mental hospital. Now this was not particularly surprising in view of the condition that [inaudible] was in, but what shocked me was that Melanie Klein was not interested, except that this interrupted the analysis. That this woman’s personal problems were relevant to the boy’s problem seemed to me something that she was totally uninterested in, and I may say I subsequently had another case where – I won’t go into the details – something similar was going on. So I realised, and I say it was rather a belated realisation, regarding the importance of real-life experience was very much at variance with that of many people in the society. (ref. PP/BOW/A.5/37)

It is a pleasure to hear Bowlby’s own retrospective account of his development as a psychoanalyst. However, it is also remarkable quite how much about this case feels contemporary. The constellation of symptoms – “hyper-active, anxious, aggressive, all over the shop” – will be familiar to clinicians. Yet despite the decades that have passed since 1938, many practitioners still today have difficulties with, or lack confidence in, fitting these problems into available diagnostic systems. For instance, debates remain inconclusive in both research and clinical contexts regarding whether externalising symptoms in childhood are best understood as innate, within-the-child traits, or as consequences of the family system and the wider social environment, or as both (Belsky & Pluess, 2009; Rutter, 2009).

There are many ways in which Bowlby’s work has influenced the world we now inhabit. Three, especially, are important in thinking about what Bowlby took from his work with this first child patient:

1) One legacy of Bowlby’s work in this early period of his career was the growth of family therapy in Britain, of which Bowlby was an early and influential advocate (see, e.g., Bowlby 1949).

2) A second legacy, of course, was the development of attachment theory: an empirically-based model of individual socio-emotional development that drew from Melanie Klein but rejected her disavowal of the importance of the actual caregiving environment. Attachment theory is among the most influential perspectives on child development; it has formed the basis of a number of research programmes and has led to significant advances in clinical and welfare practices across the world (Kozlowska & Elliott, 2014). Attachment theory, like psychoanalysis, highlights the ways through which normative processes that support individual mental health can also lead to pathology. Bowlby regarded this aetiological strategy, rather than a categorical separation between health and pathology, as basic to both psychoanalysis and analytic biology. In an unpublished manuscript from 1983 addressing links between his project and that of Darwin (ref. PP/BOW/F.3/132), Bowlby criticised the focus of the psychological establishment on category-centric practice where this is divorced from a developmental perspective that is attentive to lines of continuity. Writing in 1983, this text is composed three years after the global event that was the publication of the DSM-III, the first of the American Psychiatric Association’s diagnostic manuals to achieve widespread use as an accepted and credible tool among psychiatrists:

The categorists are still searching for diagnostic criteria that distinguish the mentally ill from the normal, though today their search is more likely to be for genetically determined biochemical anomalies than for any behavioural criterion. [On the other hand, there are] those others who, like myself, believe continuity to be a more fruitful perspective.

3) A third legacy of Bowlby’s early work was his description of the behaviour of children who had suffered early separations from their caregivers or privation in their relationships, and in particular those who had experienced a period of institutionalisation. Bowlby and his small research group at the Tavistock observed children who variously did not seek or respond to their caregiver when distressed; froze out their caregivers when social or emotional reciprocity was attempted; had a brittle quality to their emotion-regulation leading them to erupt in rage at the slightest provocation; and/or appeared willing to wander off with strangers. In 1980, the DSM-III introduced the category of “reactive attachment disorder in infancy” as a recognised diagnosis, to capture the behaviours that Bowlby and colleagues had observed. There has been a growing body of research on attachment disorders since 1998, and especially since the increased prominence and detail of the diagnosis in DSM-IV (1994) and DSM-5 (2013). These developments in diagnostic practice have been influenced by the pivotal series of studies of children adopted from Eastern European orphanages (O’Connor et al., 1999; Zeanah, 2000; Zeanah et al., 2005; Zeanah et al., 2006).

The focus of this article will be on two questions, which will be addressed in order. Firstly: what might clinicians make of Bowlby’s first patient if he were seen by services today? Though the potential literature on diagnosis, trauma and caregiving environments is vast, we will review just some of this in considering how Bowlby’s case could be regarded now. The historical anachronism involved is not especially substantial: the constellation of symptoms remains a familiar one, and Bowlby’s name is invoked as support and consecration for how clinical practice and research go about interpreting them. However, in the second part of this article, we reverse the lines of interpretation between present and past, and consider what Bowlby might still have to teach us. The wealth of Bowlby’s published and unpublished writings can take us back to a period before many of our common assumptions about families, attachment and different disorders. A second question, then, is: How might thinking with Bowlby help us critically consider children whose presentations span a number of diagnoses, and children who do not readily fit any existing diagnostic category? In a sense, our discussion is of the tensions between the three key legacies of Bowlby’s work: the focus on family systems; attachment theory as a theory of individual development; and psychiatric classification of attachment. We will consider what Bowlby might regard as particularly important in a case like this, and the purpose and function of diagnosis.

Diagnostic Considerations

Today, there are two psychiatric diagnoses relating to childhood attachment: 1) reactive attachment disorder (RAD); and 2) disinhibited attachment disorder (ICD-10, World Health Organisation, 1992), with only the former now recognised in the DSM-V (American Psychiatric Association, 2013; for a recent review, see Zeanah & Gleason, 2015). A formal diagnosis of RAD would imply the absence of an attachment relationship and thus the establishment of any attachment relationship, even one of poor quality, would generally rule out a diagnosis of reactive attachment disorder. The attachment disorder diagnosis has seen quite widespread application – perhaps too widespread, with loose application of an unspecified “attachment disorder” classification causing confusion within clinical formulation and diagnostic practices. Although the RAD diagnostic category was clearly not introduced with the intention of capturing such cases, many clinicians might consider using the concept of “attachment disorder” if Bowlby’s first child patient came through their doors today.

It is true that attachment processes may impact a number of areas of development. This is perhaps one factor that has led to excessive use of an unspecified “attachment disorder” diagnosis even when the diagnostic criteria for RAD are not met (Woolgar & Baldock, 2015). This overuse has been particularly prevalent among children who have been maltreated (McCullough et al., 2014) or adopted (DeJong et al., 2016; Woolgar & Scott, 2014). In fact, RAD is exceptionally rare, even in disadvantaged samples of young children (Zeanah & Smyke, 2009), and disinhibited attachment is distinctly uncommon in children who have not been exposed to profound and pervasive institutional deprivation (Rutter, 2009). There are substantial problems with the extension of the RAD category beyond its intended target. In doing so, inadequate attention is paid to the question of whether the attachment system specifically is malfunctioning, as opposed to behaviour reflecting other factors such as exposure to trauma or neurodevelopmental problems. Additionally, Woolgar and Scott (2014, p. 363) have warned that “one of the major risks of resorting to an unspecified attachment disorder formulation is that it can leave systems feeling helpless or with excuses to avoid engaging with families.”

From what we know of the case, in the case of Bowlby’s first child patient, RAD should be ruled out. For reactive attachment disorder, we would expect an inhibited presentation with signs of sadness, withdrawal and low affect (WHO, 1992). From Bowlby’s description, the only aspect of this young boy’s presentation that could be mapped onto this diagnostic category would be the diffuse “anxiety” term, which on its own gives little helpful information for the purposes of diagnosis. One of the clearest reasons to rule out a diagnosis of RAD would be a lack of evidence of severe neglect, including institutional deprivation, multiple changes in caregivers or extreme deviations in the provision of basic care. If the birth mother was considered to be responsible for the latter, then the likely treatment would be to remove the child within the context of a safeguarding framework. We would expect the presentation to quickly recover, as RAD symptoms have been found to respond readily to environmental change (Bos et al., 2011).

Having excluded reactive attachment disorder, conduct disorder may seem the most likely diagnosis in this case. The severity of the clinical presentation might well meet specific criteria for oppositional defiant disorder (ODD), a subcategory of conduct disorder (CD), which is defined by the presence of markedly defiant, disobedient or provocative behaviour which does not include more serious aggressive or dissocial acts (WHO, 1992). It is not unusual for pre-school children to show temper tantrums on a weekly basis (Egger and Angold 2006) and behavioural outbursts for a child of three years would only be considered a symptom of ODD if they were clearly outside the normal range of behaviour for a child of the same age and within the same sociocultural context. Nonetheless, the manifestations of emotional symptoms in childhood may have longer-term legacies beyond early behaviour problems (Stringaris and Goodman, 2009). In keeping with the clinical profile of Bowlby’s case, the symptoms of ODD typically appear during the pre-school years, with the highest prevalence rates typically found in children aged 3 to 4 years (Egger and Angold, 2006; Lavigne et al., 2009). The ICD-10 system thus issues caution in applying this category to older children (i.e. those over ten years of age). Moreover, the prevalence of ODD is consistently found to be greater in boys, even after controlling for social factors and baseline psychopathology (Demmer et al., 2017). However, it is important to note that even severely mischievous or naughty behaviour is not in itself sufficient to warrant a diagnosis of ODD and there is no clear consensus about whether clinical distinctions should be quantitative or qualitative in nature.

The diagnostic profile of ODD is just one aspect of a complex pattern of problematic social interactions. Oppositional and defiant behaviours are most evident in interactions with adults whom the child knows well (Rowe et al., 2010); it can therefore be difficult for clinicians to unpick the relative contribution of the child’s behaviour to their problematic interactions with others. Often, children with ODD have experienced a history of hostile or inconsistent parenting (Pederson & Fite, 2014). In clinical practice, it is almost impossible to determine if the child’s behaviour caused the parent to act in a more hostile manner toward their offspring, if parental hostility led to the child’s behaviour, or if there was some combination of both. Yet, while the relative contributions of possible causal factors should be separated in routine practice, clinical guidelines state that this should not influence whether or not a diagnosis is made. The heterogeneity of the CD/ODD cluster makes it a broad construct that needs to be personalised. Several differential diagnoses may also be considered in this case. The disruptive and aggressive behaviour exhibited could have been elicited by the child’s forced separation from his mother in the clinical setting: his oppositional behaviour could be secondary to separation anxiety disorder (SAD). It is normal for young children, particularly toddlers, to show anxiety over threatened (or real) separation from a primary caregiver to whom they have an attachment. A clinical diagnosis of SAD would only be considered if the behaviour is developmentally inappropriate, the fear about the separation is the focus of the child’s anxiety and the symptoms are associated with significant impairments in social functioning (WHO, 1992).

The behaviours that Bowlby described in this case are agitation and emotional dysregulation following separation from the primary caregiver, and aspects of this boy’s presentation suggest problems in the attachment relationship. The National Institute for Health and Care Excellence (NICE) Clinical Guidelines for Children’s Attachment (2015) have attempted to shift practice away from psychiatric classifications of disordered attachment and towards an emphasis on early identification of “attachment problems”. In part, this is an appropriate and necessary response to the fact that true instances of RAD are rare, especially for children who have not undergone institutional care or extreme neglect. Yet the “attachment problems” construct used by NICE is also a fuzzy compromise and a placeholder, the result of both lack of clarity regarding what exactly is under discussion and fundamental disagreements about how to conceptualise and operationalise attachment.

A notable contemporary debate in this regard relates to whether attachment should be treated as an individual or a relationship-level construct. To assess attachment quality, clinical practice guidelines from the American Academy of Child and Adolescent Psychiatry (Zeanah et al., 2016) recommend structured separations and reunions with the primary caregivers and interactions with strangers. These practice parameters explicitly recommend looking at attachment quality rather than attachment patterns. This guidance appears intended to counteract tendencies among practitioners to press into service a category from the research literature: the disorganised attachment classification (Main & Solomon 1990). This is despite the fact that this classification has not had any validation for clinical purposes.

Disorganised attachment is coded on the basis of a child’s response to reunion with their caregiver in the Ainsworth Strange Situation Procedure, where these responses seem to indicate conflict at the level of the attachment system. The indices of disorganised attachment include fearful, fidgety or out-of-context aggression towards the caregiver that otherwise interrupts the child’s response on reunion. A well-established antecedent of disorganised attachment is parental frightening, frightened or dissociative behaviour towards their child (Madigan et al. 2006). Disorganised attachment has been linked to later externalising symptoms (Fearon et al., 2010), and on this basis Lyons-Ruth and Jacobvitz (2016) have argued that the disorganised attachment classification should be revalidated for diagnostic use as a measure of later psychopathology. However, Zeanah and Lieberman (2016) have explicitly argued against such an approach. Interestingly, a key reason that they give is that diagnosis by definition has to be of individual trait-like properties, not of the dynamics particular to a particular relationship. There is no association between disorganised attachment behaviour with one caregiver and with another in the Strange Situation (van IJzendoorn et al. 1999), which indicates that the phenomenon is relationship-specific at least in infancy, though for older children the properties of their individual relationships may become increasingly trait-like. Zeanah and Lieberman therefore argue that not only is it inappropriate to press disorganised attachment into service in a diagnostic role, but that doing so would contravene the basic function of diagnosis.

Continuing to Learn from Bowlby

Though much contemporary research into infant development is predicated on Bowlby’s original insights, those texts of Bowlby’s that lie off the beaten path have not always been considered as carefully as, for example, the Attachment and Loss trilogy. These other texts, including the unpublished material in the Bowlby Archive, nonetheless may have value for us today in thinking about clinical work and how diagnostic categories are used. In particular, renewed consideration of Bowlby’s key ideas can help us to retain focus on the importance of the family context and utilise diagnoses in a limited and effective way, even within a clinical and institutional environment that tends toward expecting answers from diagnoses about individual prognosis and treatment.

Bowlby understood the complexities of labelling children’s expressions of distress. Thirty years prior to the formal introduction of attachment theory, Bowlby had already begun to emphasise the interlocking between biological, psychological and systemic processes. Writing in Personality and Mental Illness, published in 1940, Bowlby described the critical role of the family environment in the course of child development. He emphasised not just the observable parent-infant relationship, but also the contribution of maternal genetics and biological systems:

It now appears indubitable that some people are born more liable than others to mental instability. The incidence of psychosis shows that certain families are far more liable to insanity than would be expected on chance or even allowing for the effect upon children of unstable parents. But of course this tendency may never result in actual mental illness. The environment may be so congenial that a perfectly stable personality results. The tendency, however, is there, and moderately bad surroundings (probably especially in early life) will bring it out. On the other hand it seems probable that there are other people in whom the tendency is minimal. (1940, p. 191)

However, even within the complexity of different interlocking causes, Bowlby recognised that children’s self-image and relationships were critically affected by their image of their primary caregiver. He maintained that parents’ relative ability or inability to create a secure attachment during the child’s early years increased the risk of emotional and behavioural difficulties in the future. If a child does not perceive their caregiver ultimately as available and responsive to their needs, this creates anxiety and anger. Over time, through repeated parent-child encounters, such relationships may increase the child’s likelihood of developing distress, depression or aggression.

In interpreting such symptoms, Bowlby emphasised the importance of “developmental pathways”. For Bowlby, neither knowledge of general patterns nor knowledge of the specific case should be abandoned. He considered diagnostic systems to provide an important framework on which to develop epidemiological research and expand evidence-based treatment: a view that has since been echoed by others (e.g., DeJong, 2010). Today, this might be discussed in terms of the distinction between diagnosis and formulation (Johnstone & Dallos, 2014). On an individual level, diagnostic classifications tell clinicians what one child has in common with other children with similar symptoms. They can help clinicians to develop an integrative account of the child’s difficulties, encourage more consistent care-planning and encourage professionals to consider the treatments that may help. Formulation, on the other hand, is an individual approach that considers the ways in which the child is unique from others. We suspect that attachment concepts find their primary role in the context of formulation, and problems arise when their use in the context of formulation is confused with use in the context of diagnosis. As Waters and McIntosh (2011, p. 478) have emphasised, “using scores on attachment measures to make decisions is not the same as using attachment-based observations and assessment to inform your formulation of what’s going on in the family”.

Bowlby was keen to emphasise the need to distinguish between the generalisations of diagnostic or research frameworks and individually-targeted clinical understanding, and to establish a productive and considered relationship between the two. For Bowlby, the key was how these antimonies were brought together:

In the past there has been a deplorable tendency for the experimentalist to despise the clinician’s lack of precision and the clinician to reciprocate with contempt for the experimentalist’s lack of insight into human nature. Each has stoutly maintained that his own method was the one true way to knowledge. These claims are absurd: each method is indispensable. It is the clinician who usually has the earliest insights, defines the problem, and formulates the first hypotheses. But the detailed minute study of the feelings and motivations of his patients, and the complicated intellectual and emotional repercussions to which they give rise, the clinical worker provides information regarding the relations of psychic and environmental forces which can be obtained in no other way. (1951, p. 61)

Confronting the multiple factors that are influencing the parent-child relationship, a single construct that captures the whole thing has great appeal. We suspect that there has been a tendency for professionals to make appeal to “attachment difficulties” when faced with complexity. However, this may well be at the risk of short-circuiting attention to details and specifics. Similarly, foster and adoptive parents of children referred to looked-after children’s services still continue to seek out problems and answers within the framework of attachment (Barth, 2005). In part, this appeal to attachment is in line with Bowlby’s legacy, since his theory of attachment was proposed as an account relevant to the whole personality. He was an early thinker in the field of developmental psychopathology, and understood that development was multiply determined by a range of transactional or interactional processes. Writing in the Health Education Journal in 1954, Bowlby and colleagues wrote that:

There has been a tendency to stress the symptom too much in the past so that it has sometimes come to be regarded almost as the illness itself rather than as a particular manifestation of a more general disturbance within the total personality of the child. The former viewpoint connotes a static conception of the mind and precludes the investigation of the conflict between the emotional forces within the child’s mind which emerges in the form of symptoms. (Bowlby et al., 1954, p. 62)

Yet Bowlby’s own reflections on the purpose of classification can help us here. In an undated text from the Bowlby Archive, Bowlby offers his reflections on his formation of diagnostic groupings in children younger than 5 years:

Classification and diagnosis in child psychiatry is at present in a state of anarchy. Very few of the children seen correspond to any of the classical psychoneuroses or psychoses – the fact unjustly being simply “character-problems”. For these character cases there is as yet no good classification for adults, let alone for children…Even in adults it is sometimes difficult to distinguish clearly between the habitual personality and the particular syndrome of symptoms from which the patient is suffering. This difficulty is increased in childhood. Consequently the classification used here is only provisional. (ref. PP/Bow/C3/9)

It is not clear whether or not Bowlby would have wished attachment terms to figure within formalised diagnostic systems. However, we do suspect that he would have wanted clean distinctions drawn, where possible, between such diagnostic usage and discussion of attachment within formulation.

In diagnosis, attachment problems ought to be conceptualised recognising qualitatively different forms of developmental antecedents and sequelae. For example, with regard to reactive attachment disorder specifically, current classification systems now specify that it is severe neglect, not maltreatment, which is the key aetiological mechanism behind the failure to develop an attachment relationship. On top of that, there is a lack of clarity regarding how reactive attachment disorder is diagnosed in practice. We would emphasise that if a reactive attachment disorder is being considered, then assessment needs to take into account the child’s attachment-related behaviours to their primary caregivers and to contrast this with strangers, typically in the context of stressors such as separations and reunions with the primary caregivers. Whatever else, diagnosis of attachment disorders should focus on specific attachment-related behaviours, their directedness towards a preferred familiar figure, their activation in times of stress, and their use for proximity seeking and contact maintenance. For a diagnosis to be made identifying a disorder of the attachment system, assessment needs to focus on the functioning of that system specifically. Attachment processes can readily impact other areas of functioning, but this does not mean that the resulting behaviours are therefore to be considered as attachment: for instance, as Waters et al. (1993, p. 220) argued some years back, “we can view attachment as a potentiator of disruptive behavior problems, without implying that the behaviour problems are themselves attachment behaviors.”

When accurate and relevant diagnoses are given, children and families are better able to access the evidence-based treatments and support that may make a significant difference to their emotional and social functioning. Diagnoses perform a function of elaborating the ways in which one child has features in common with other children, and that these commonalities are likely to be responsive to particular interventions, and they are helpful for this function, given the available knowledge in the field of child psychopathology. Yet making these diagnoses should not be confused with or detract from the importance of improving relationships between children and their caregivers. And it is these latter issues that will often fall under the ideas of “attachment problems” or “attachment difficulties”, outside of a diagnostic system. In the UK, NICE (2015) now provides guidance for professionals to assess this broader range of problems described as “attachment difficulties” for children who are adopted or are at risk of going into care.

In formulation, away from concerns with diagnosis, it should be remembered that the attachment construct should be viewed as a subtle, reciprocal and transactional process that sits best within a personalised formulation of a child’s difficulties. Used for this purpose within formulation, attachment theory is not shoehorned into diagnostic systems. Neither does it necessarily represent a challenge to the validity or need for diagnosis. It is serving a different purpose. The attachment construct should be applied to formulation during complex and nuanced thinking about a child’s behavioural profile, potential risk factors and considering the role of the child’s external environment. To take an example: a clinician may find interest in an assessment that indicates an insecure attachment between a child and a depressed caregiver. This could be very relevant to formulation, in thinking about how to work with this family. However, it would be ill-advised to regard the insecure behaviour like a symptom, in itself an indicator of pathology. Take for instance findings by Milan et al. (2009) that in families where mothers reported intermittent depressive symptoms, children with insecure attachment classifications at age 3 in the Strange Situation actually reported less depression at age 11 than did children with secure histories. The pattern was reversed for children of mothers with sustained depressive symptoms. The authors point out that the kinds of communication strategies shown by insecure children may have some advantages in the case of an uncertain caregiving environment where intermittent depression makes the caregiver less available but generally retrievable. However, where the adversity was sustained, the child’s greater trust in the relationship captured by the secure classification in the Strange Situation became the greater asset.

Conclusions

Bowlby recognised the importance of theory in clinical decision-making and he understood the challenges faced by professionals when using existing general frameworks, whether conceptual or diagnostic, to conceptualise individual cases:

Many analysts and other psychotherapists do excellent work using their intuition and without very clear ideas on theory, and often, I believe, in spite of the theories they nominally subscribe to. I have not that sort of mind, nor am I strong on intuition. Instead, I tend to apply such theories as I hold in an effort to understand my patient’s problems. This works well when the theories are applicable but can be a big handicap when they are not. Perhaps my saving graces have been that I am a good listener and not too dogmatic about theory…I often shudder to think how inept I have often been as a therapist and how I have ignored or misunderstood material a patient has presented. Clearly the best therapy is done by a therapist who is both naturally intuitive and also guided by appropriate theory. (Bowlby [1985] 1991, p.29-30)

Returning to Bowlby’s texts for guidance suggests to us that a sharp sense is needed of the point at which a disturbance of the attachment system – the capacity to utilise a familiar caregiver as a secure base and safe haven – is the core feature of a clinical problem, rather than a co-occurring feature. We suspect that Bowlby would have argued against using broad, diffuse concepts – even “attachment” – where this usage places a large proportion of the explanatory burden on a single subtle construct, and thus directs attention away from the story of the individual child. The personality should be taken as a whole, with recognition that attachment is only one diffusely influential aspect. The best type of clinical thinking will incorporate a clear diagnosis, with a clinical rationale, alongside detailed formulation.