A RADical Idea: A Call to Eliminate “Attachment Disorder” and “Attachment Therapy” From the Clinical Lexicon

Attachment disorder” and “attachment therapy” are common terms used in applied clinical practice. However, these terms are not typically employed in research settings or published scientific papers. In this article, the author reviews the theoretical tenets and empirical research of attachment theory and discusses how these two terms fail to coincide with the scientific knowledge. The historical development of these phrases is considered, as well as the potential impact they have on clinical practice. The ultimate conclusion is that the “attachment disorder” and “attachment therapy” constructs are hindrances to evidence-based clinical practice and should be eliminated from the clinical lexicon. Most mental health clinicians are familiar with the terms “attachment disorder” and “attachment therapy.” Some associate these terms exclusively with the notorious holding therapy and rage reduction approaches (e.g., Cline, 1979; Zaslow & Menta, 1975), techniques that resulted in serious physical injury and even death to some children (Mercer, Sarner, & Rosa, 2003). However, other treatment approaches employing ethically questionable techniques are similarly promoted from an attachment perspective (e.g., Dyadic Developmental Psychotherapy, Theraplay; see Allen, 2011b, and Mercer, 2015, for reviews), and a sizable minority of clinicians appear interested in these approaches (Allen, Gharagozloo, & Johnson, 2012). Still others understand these terms in a more pedestrian way, assuming they represent a common presentation and treatment approach, respectively, for maltreated children. In short, there is a lack of professional consensus on what exactly these terms mean and the manner in which they should be used. Given the immense confusion and multiple child deaths associated with purported “attachment therapies,” various professional organizations issued policy statements and practice parameters for the assessment and treatment of children displaying attachment-related concerns (e.g., American Academy of Child and Adolescent Psychiatry, 2005; American Professional Society on the Abuse of Children [Chaffin et al., 2006]). These reports largely yielded similar conclusions and recommendations: Attachment-related problems are poorly understood by many practicing clinicians, unsupported treatment techniques should be rejected, and scientifically supported interventions that strengthen the parent–child relationship are preferred. Unfortunately, anecdotal reports, court cases, news articles, and empirical research suggest that the recommendations of these organizations are not practiced by many professionals. Perhaps a new approach is necessary. Admittedly, the title of this article may confuse and surprise many readers. The concepts of attachment disorder and attachment therapy are well ingrained into graduate education, clinical practice, child welfare, adoption, and continuing education programs. However, the reader should note that I did not include the academic arena in the list of areas where these phrases are widely accepted. There are many reasons for this, which are highlighted throughout this article. Suffice it to say that neither of these two concepts is empirically sound as commonly practiced. Again, this may be confusing and surprising to many. However, if CONTACT Brian Allen [email protected] Center for the Protection of Children, Penn State Hershey Children’s Hospital, 500 University Drive, Hershey, PA 17033. The terms “attachment disorder” and “attachment therapy” are used with some frequency in multiple professions, including mental health, law, child advocacy, and social work. Although confusion regarding these terms extends to allied professions, this article discusses these terms specifically as they relate to the mental health field. It is recognized, however, that many of the points made in this article may be relevant to other professions. EVIDENCE-BASED PRACTICE IN CHILD AND ADOLESCENT MENTAL HEALTH 2016, VOL. 1, NO. 1, 60–71 http://dx.doi.org/10.1080/23794925.2016.1172945 © 2016 Taylor & Francis D ow nl oa de d by [ Pe nn S ta te U ni ve rs ity ], [ B ri an A lle n] a t 1 2: 50 1 3 Ju ly 2 01 6 the reader will afford me the opportunity to detail my reasoning, perhaps the title of this article will be seen as the logical conclusion. Attachment: A clinical or developmental theory? John Bowlby, who originally specified attachment theory, was a trained psychoanalyst, having been mentored by Melanie Klein. However, Bowlby was an empiricist who was disillusioned with the seemingly widening gap between psychoanalytic theories of the time and emerging scientific findings from the fields of ethology, biology, cognitive development, cybernetics, and others. He also was influenced by his own observations suggesting the primacy and necessity of a supportive and responsive caregiver for healthy development, findings most famously described in his work for the World Health Organization (Bowlby, 1951) and his classic volume Forty-Four Juvenile Thieves (Bowlby, 1946). Bowlby explicated his thinking on adaptive and aberrant development with the publication of his seminal “attachment trilogy” (Bowlby 1969/1982, 1973, 1980). Interestingly, although Bowlby was a clinician developing a theory to improve clinical practice, his ideas were primarily investigated by developmental scientists. Readers will likely recall the formative work of Harry Harlow (1958) with rhesus monkeys, demonstrating that the availability of a caregiver who provides contact comfort and felt security is more crucial to emotional well-being than a caregiver who provides oral stimulation (note the direct test of attachment versus psychoanalytic ideas). Mary Ainsworth (1967), after studying with Bowlby in London, traveled to Uganda to conduct detailed observations of mother–child interactions. She continued this work after securing faculty positions in the United States and trained a generation of attachment researchers. Although the considerable quantity and quality of work accomplished by these researchers resulted in attachment theory becoming the most widely accepted theory of socioemotional development, comparatively little research examined its clinical applications. Indeed, near the end of his life, Bowlby (1988) expressed disappointment that his ideas were so rarely studied for clinical purposes. The distinction between attachment behavior and attachment theory First, it is important to note that Bowlby and developmental scientists draw a clear distinction between attachment behavior and the broader context of attachment theory. Understanding this difference is a crucial point for understanding the remainder of this discussion. Attachment behavior is any attempt, whether verbal or nonverbal, to maintain proximity to and seek comfort from an attachment figure for the purposes of reducing distress (Bowlby 1969/1982). Using the Strange Situation paradigm with infants and toddlers, Ainsworth (1978) and her colleagues identified three patterns of attachment behavior that they labeled as secure, avoidant, and resistant/ambivalent. Secure attachment is most often fostered by consistently sensitive and responsive caregiving, although the strength of this relationship is moderate and other causal factors are likely (De Wolff & van IJzendoorn, 1997; Lucassen et al., 2011). Alternatively, rejecting and dismissive caregiving is associated with the development of an avoidant attachment, whereas a resistant/ambivalent attachment is linked to inconsistent caregiving that fluctuates between responsive and rejecting (Egeland & Sroufe, 1981). Each of these patterns of attachment is considered an organized and coherent set of characteristic behaviors designed to accomplish the same goal: to maintain proximity to a caregiver who can be called upon to help cope with a stressful situation. Although this may sound contradictory in the case of the insecure forms (i.e., avoidant, resistant/ambivalent), consider that a child displaying avoidant attachment behaviors, for example, actually is able to maintain proximity to a rejecting caregiver by not displaying approach behaviors that may prompt the caregiver to respond by moving away. Main and Solomon (1990) identified a number of children who displayed fearful, odd, contradictory, or otherwise bizarre behaviors during the Strange Situation procedure. In essence, these children appeared to lack a coherently organized set of behaviors for maintaining proximity to a caregiver and were thus identified as displaying a disorganized attachment. Subsequent research demonstrated that child maltreatment, particularly physical abuse, was a primary causal factor (although not the only causal factor) in the development of disorganized attachment behavior (Cyr, Euser, Bakermans-Kranenburg, & van IJzendoorn, 2010). Conceptually, disorganized behavior is believed to display the approach-avoidance conflict inherent in seeking security from a caregiver who is also a significant source of fear (Main & Hesse, 1990). Attachment classifications are somewhat flexible with 6to 12-month stability estimates of organized infant attachment classifications rarely noted above 65% (e.g., Belsky, Campbell, Cohn, & Moore, 1996; EVIDENCE-BASED PRACTICE IN CHILD AND ADOLESCENT MENTAL HEALTH 61 D ow nl oa de d by [ Pe nn S ta te U ni ve rs ity ], [ B ri an A lle n] a t 1 2: 50 1 3 Ju ly 2 01 6 Vaughn, Egeland, Sroufe, & Waters, 1979), and estimates of the stability of disorganized attachment are similarly modest (van IJzendoorn, Schuengel, & Bakermans-Kranenburg, 1999). In addition, attachment behavior is relationship specific. For instance, disorganized attachment is rarely observed with two different caregivers (van IJzendoorn et al., 1999). Although insecure and disorganized attachments are not psychiatric disorders, there are established links between attachment classification and later mental health outcomes, with the strongest association occurring between disorganized attachment and later externalizing problems (Fearon, Bakermans-Kranenburg, van IJzendoorn, Lapsley, & Roisman, 2010). As such, infant and toddler attachment classifications are typically considered indicators of the quality of a parent–child relationship, a potent risk/protective factor for later psychopathology. Attachment theory, on the other hand, is a broader conceptualization of development that incorporates and emphasizes the impact of parent–child relationships on psychological functioning. A central feature of attachment theory is the concept of the internal working model (IWM) or cognitive representation (i.e., schema). Through interactions with caregivers, children develop and revise IWMs of themselves, others, and typical social interactions (Bowlby, 1969/1982; Main, Kaplan, & Cassidy, 1985). For instance, a child may develop a model of themselves as competent and worthy of love or incompetent and flawed, a model of others as trustworthy or dangerous. Children then automatically implement these models when engaging in everyday circumstances, including social situations. A principle of these “working” models is that they are always capable of being revised in response to new information; however, a model becomes more inflexible the longer it is maintained and reinforced by experience. Research from both the attachment and cognitive development fields largely supports the concepts of cognitive representations and scripts, and the impact of these automatic processes on mental health (e.g., Cannon & Weems, 2010; Solomon, George, & De Jong, 1995; see Bretherton, 2005, for a review). Indeed, the discipline of cognitive therapy largely concerns itself with the alteration of these cognitive representations, a comparison made by Bowlby (1980) himself. Contrary to what many believe, attachment theory does not suggest that the first years of life and one’s early attachment classification determine later outcomes (Bowlby, 1988). Rather, attachment theory emphasizes an ongoing contextual understanding of the child and the multiple developmental pathways that may occur. For instance, Belsky (2005) reviewed the results of two major prospective studies examining the longitudinal impact of early attachment (i.e., Pennsylvania Infant and Family Development Project and the NICHD Study of Early Child Care). Among the many findings of these studies was clear evidence that (a) the developmental benefits and risks associated with early attachment are contingent upon the quality of caregiving received later in development, and (b) an increasing number of risk factors (e.g., poverty, maternal depression, single-parent household) predict poorer outcomes for children in areas such as behavior problems and social competence, even among those with secure attachments. Other longitudinal studies yielded similar results (e.g., Sroufe, Egeland, Carlson, & Collins, 2005), and these findings help explain the moderate bivariate associations documented between early attachment and later outcomes (Fearon et al., 2010). In short, attachment theory proposes that later experiences can alter one’s developmental trajectory in either a negative or positive fashion regardless of the attachment behavior displayed in the first years of life. Attachment ≠ parent–child relationship Confusing for many is that “attachment” and “parent– child relationship” are not synonymous terms. There are multiple interconnected and overlapping characteristics of the parent–child relationship, for instance, the discipline techniques utilized, the parent’s modeling of emotion regulation and impulse control, parental mental illness, the cognitive stimulation afforded the child, the child’s temperament and biological constitution, and communication skills. The child’s attachment to the caregiver, that is, the child’s understanding of the caregiver as a “secure base” who provides safety and security, is only one specific component of the parent–child relationship. One should be careful not to conflate these two concepts and ensure that respecting attachment theory and research does not overshadow or dismiss other aspects of the parent–child relationship. This point may best be demonstrated by way of an example. The seminal Minnesota longitudinal project provides numerous examples of this point. For instance, as one would hypothesize, children rated as insecurely attached in infancy were more likely to display behavioral problems in preschool (Erickson, Sroufe, & Egeland, 1985). However, in those instances where this association was not present (i.e., secure with behavior problems or insecure without behavior problems), children were more likely to display problems if their mothers were poor at setting limits on their child and/or displayed less confidence in their ability to 62 B. ALLEN D ow nl oa de d by [ Pe nn S ta te U ni ve rs ity ], [ B ri an A lle n] a t 1 2: 50 1 3 Ju ly 2 01 6 manage the child’s behavior. Conceptually, these results suggest that caregivers who are emotionally supportive and comforting to their distressed children may still confer risk for poor developmental outcomes as a result of other parenting variables. Indeed, multiple reports from the Minnesota project suggest that models including other parent–child relationship variables beyond attachment classification were often preferred for predicting outcomes when compared to attachment classification alone (Sroufe et al., 2005). One should remember that attachment classification is a specific risk/protective factor that indexes certain, but not all, aspects of the much broader parent–child relationship. A brief history of Reactive Attachment Disorder (RAD) As mentioned previously, Bowlby developed many of his ideas by observing the effects of parental deprivation on infants and children. Evidence had existed for decades that infants living in hospitals or other settings without consistent parental interaction would display depressivelike behaviors, many times resulting in the death of the child, what Rene Spitz (Spitz & Wolf, 1946) termed “anaclitic depression.” By the late 1970s, Bowlby’s theory on the importance of the parent–child relationship for development, Harlow’s research with rhesus monkeys, and Ainsworth’s observational studies were well known. Within this scientific and clinical climate, the Diagnostic and Statistical Manual of Mental Disorders (3rd ed.; DSM-III) was published in 1980 (American Psychiatric Association, 1980). DSM-III included the first codification of a disorder purportedly based in attachment theory. This presentation, termed Reactive Attachment Disorder of Infancy, described what is commonly referred to as non-organic failure to thrive (i.e., anaclitic depression or hospitalism). Diagnostic criteria included a lack of appropriate social responsiveness (e.g., smiling in response to faces, reaching for mother, engaging in playful games), lethargy or irritability, and weight loss or failure to gain appropriate weight. The criteria required onset of the condition before 8 months of age, and many of the individual symptoms were deemed valid indicators of the condition if the child was at least 2 months of age. Attachment researchers were quick to criticize the diagnosis, pointing out that children do not typically show attachment behavior to a discriminated caregiver until at least 6 months of age (Rutter & Shaffer, 1980), making it terminologically and conceptually incorrect to suggest that children younger than 6 months of age were showing signs of disordered attachment. Thus, RAD as defined in the DSM-III did not actually describe or attempt to describe disordered attachment behavior. Rather, the intent was to describe the symptoms of the disorder as being “reactive” to problems in a supposed attachment relationship. The astute reader will note that, given the tenets of attachment theory, practically any form of emotional or behavioral problem could be conceptualized as “reactive” to problems in the primary attachment relationship. In brief, the DSM-III unsuccessfully attempted to link a welldocumented clinical condition of infants (i.e., nonorganic failure to thrive) to a popular and prevailing developmental theory. By the time the DSM-III-R (3rd ed., rev.; American Psychiatric Association, 1987) was published in 1987, it was commonly recognized that toddlers and preschoolers raised in institutions (e.g., orphanages) were at risk for displaying problematic social behavior, whether it be significantly withdrawn and selfisolating or “overly friendly” behavior that lacked appropriate reticence toward strangers (Tizard & Rees, 1975). In recognition of these findings, and to address the criticisms of RAD as found in the DSM-III, the diagnosis was radically changed in the DSM-III-R to reflect “markedly disturbed social relatedness in most contexts.” This presentation could manifest as either “persistent failure to initiate or respond to most social interactions (inhibited)” or “indiscriminate sociability (disinhibited),” and there was a required presumption that the condition was caused by “grossly pathogenic care” (p. 91). In addition, the condition was described as RAD of Infancy or Early Childhood, as the aberrant social behavior could now be diagnosed as RAD if present before the age of 5 years. Although few doubted that the problems described by RAD in the DSM-III-R were linked to severe early neglect, RAD described problematic social behaviors and not disordered attachment behavior (i.e., seeking a caregiver when distressed; Green, 2003; Zeanah & Boris, 2000). In a series of case presentations, Richters and Volkmar (1994) similarly reached the conclusion that RAD, as defined in the DSM-III-R, is not compatible with developmentally derived conceptualizations of attachment behavior. Nonetheless, the diagnostic criteria for RAD were unchanged when the DSM-IV (4th ed.; American Psychiatric Association, 1994) was published in 1994. Beyond greater specification and inserting the term “attachment” into the criteria, RAD remained a disorder of social relatedness. Shortly after publication of the DSM-IV, intensive research examining children adopted from Eastern European orphanages began to appear. Studies demonstrated that within these samples of severely neglected children, symptoms of RAD were observable in a minority of the EVIDENCE-BASED PRACTICE IN CHILD AND ADOLESCENT MENTAL HEALTH 63 D ow nl oa de d by [ Pe nn S ta te U ni ve rs ity ], [ B ri an A lle n] a t 1 2: 50 1 3 Ju ly 2 01 6 cases (e.g., Smyke, Dumitrescu, & Zeanah, 2002). A more recent longitudinal study provided data showing that only 4.6% of these children showed inhibited RAD while living in the orphanages, and 31.8% displayed disinhibited RAD (Gleason et al., 2011). Two additional findings from these studies deserve mention. First, children placed in appropriate foster care homes who previously displayed the inhibited/ withdrawn subtype of RAD no longer displayed these symptoms when followed up months later (e.g., Smyke et al., 2012). Zeanah and Gleason (2015) recently summarized that “in studies of children adopted out of institutions, there are no reports of children with (inhibited) RAD ... suggesting that signs of (inhibited) RAD diminish or disappear once the child is placed in a more normative caregiving environment” (p. 217). They further opined that these results make it unclear “whether additional interventions beyond family placement may be necessary” (p. 217). Second, numerous studies showed that children displaying the disinhibited/indiscriminate sociability subtype of RAD continued to display these behaviors years later, after presumably developing a discriminated attachment relationship with their adoptive caregiver(s) (Chisholm, 1998; Rutter et al., 2007). These latter findings suggest quite clearly that disinhibited, indiscriminate social behavior, although potentially being etiologically related to early severe deprivation, is not related to the child’s concurrent attachment behavior and, therefore, is not a sign of disordered attachment. As a result of the criticisms and extant research on RAD, significant changes occurred with the publication of the DSM-5 (5th ed.; American Psychiatric Association, 2013) to align the diagnosis with a developmental conceptualization of attachment behavior. First, in light of the evidence that indiscriminate sociability is not related to concurrent attachment behavior, the disinhibited subtype is no longer considered RAD. Rather, it is now described as Disinhibited Social Engagement Disorder (DSED). Terms used to describe this presentation in DSM-IV, such as “diffuse attachments,” are eliminated in DSM-5 and replaced with clearer definitions. Second, RAD in DSM-5 refers specifically to a child who rarely or minimally seeks or responds to comfort from a caregiver when distressed. In the DSM-5, “RAD is essentially the absence of a preferred attachment to anyone” (Lyons-Ruth, Zeanah, Benoit, Madigan, & Mills-Koonce, 2014, p. 698). Given that maltreated children with disorganized attachment demonstrate attachment to a preferred caregiver (albeit potentially in a problematic way), these children are precluded from being diagnosed with RAD. The DSM-5 goes on to state that RAD occurs in less than 10% of severely neglected children (i.e., those raised in institutions) and the disorder is rarely seen in general clinical practice. Even with the revisions of the RAD diagnosis in DSM-5, the DSM conceptualizes psychopathology as residing within the individual. In many ways, the idea that an individual displays any form of attachment disorder is problematic. From its first theoretical iteration attachment was conceived as a process, a system that involves the behaviors and interactional responses of two individuals (Bowlby, 1969/1982). The child’s behavior is viewed as responsive to the caregiver and vice versa. As one would therefore expect, RAD resolves relatively quickly following placement with a supportive caregiver (Smyke et al., 2012). However, describing a child as displaying an attachment disorder, including RAD, unfortunately focuses clinical attention on the child and not the system. This concern is recognized and addressed in other approaches to defining attachment-related concerns, such as a popular typology of attachment problems provided by Zeanah and Boris (2000). The revised version of the Diagnostic Classification of Mental Health and Developmental Disorders of Infancy and Early Childhood (DC 0-3R; Zero to Three, 2005) noted the evolving research on attachment and renamed RAD to a more empirically justifiable “Deprivation/Maltreatment Disorder.” In addition, the DC 0-3R included a separate axis for specifying the quality of the parent–child relationship. In summary, given (a) the current definition of RAD, (b) the relatively low prevalence of the condition in even severely neglected children, and (c) the finding that RAD has not been documented in any children after a period of time living in a normative caregiving environment, the safest conclusion is that RAD is an unlikely clinical presentation that will rarely, if ever, be encountered by most mental health professionals. Even in the unlikely circumstance that a child does present with RAD, the diagnostic nature of the DSM raises the concern that the condition will be viewed in a “disorder-within-the-child” manner. DSED, although also exceptionally rare, is less responsive to appropriate caregiving and is more prevalent than RAD. As such, DSED is a more likely presenting condition, but the reader should remember that DSED does not appear to be related to attachment behavior and is no longer defined as RAD. Reification of the “attachment disorder” construct At this point I imagine a number of readers may be confused by my presentation of RAD. Undoubtedly, the description provided does not coincide with the 64 B. ALLEN D ow nl oa de d by [ Pe nn S ta te U ni ve rs ity ], [ B ri an A lle n] a t 1 2: 50 1 3 Ju ly 2 01 6 concept of RAD that many have practiced. Consider, for instance, a recent study by Woolgar and Baldock (2015), who reviewed 100 consecutive referrals from professionals in the community to a specialized adoption and foster care treatment program in the United Kingdom. Of these 100 referred children, nearly one third (n = 31) were identified in the referral letter as displaying an “attachment disorder” or an attachment problem. When evaluated by the experts at the specialty program using DSM-IV/ICD-10 diagnostic criteria, only three of these 31 children were diagnosed with RAD; however, 18 of the 31 children were diagnosed with conduct disorder. Interestingly, only one of the 31 referrals mentioned a potential conduct disorder diagnosis. One may conclude that adopted and foster children with conduct problems are frequently misdiagnosed with attachment problems. The findings from Woolgar and Baldock are not surprising; it seems that professionals commonly view RAD, or an expanded diagnosis simply called “attachment disorder,” as the display of significant externalizing behavior problems (e.g., aggression, conduct problems) following maltreatment in infancy or early childhood (Becker-Weidman, 2006; Wimmer, Vonk, & Bordnick, 2009). It should be apparent that “attachment disorder” in this sense does not describe disordered attachment behavior (see Allen, 2011a, for a review). Rather, it is an attempt to explain the child’s problems by invoking the ideas of attachment theory. For instance, some proponents of the “attachment disorder” construct suggest that early maltreating experiences by an attachment figure leave a child unable to love or form relationships, resulting in children who lack a conscience and act in violent and destructive ways (e.g., Thomas, 2005). Many times these advocates will justify their ideas by providing quotes from attachment researchers, often out of context, particularly Bowlby’s early theorizing. First, attempting to use attachment theory in this way says nothing about the child’s actual attachment behavior. Rather, it is a theoretical perspective on presumed etiological factors in the development of the emotional and behavioral problems. This would be akin to labeling a set of symptoms as an “operant conditioning disorder” or an “object relational disorder.” Second, these justifications neglect a large portion of attachment theory, particularly the portions that focus on postinfancy/toddlerhood development. Empirically, research shows that the link between early attachment difficulties and later externalizing problems is modest at best (Fearon et al., 2010; O’Connor, Bredenkemp, Rutter, & the English and Romanian Adoptees Study Team, 1999), with many studies failing to find such a relationship (Smyke et al., 2002; Zeanah, Smyke, & Dumitrescu, 2002). Noted developmental researcher Ross Thompson (1999) provided an appropriate summary of this body of research: “Two decades of inquiry into the sequelae of early attachment yields this confident conclusion: Sometimes attachment in infancy predicts later psychosocial functioning, and sometimes it does not” (p. 274). How did it come to this? Why have so many professionals accepted the “attachment disorder” diagnosis in the absence of empirical support? There are likely multiple converging reasons. First, the inclusion in the DSM of the assumption that early pathogenic care is the direct cause of RAD likely prompted many to presuppose that RAD was the correct term for a maltreated child with psychiatric problems. Second, numerous authors, often promoting “attachment therapy” or “attachment parenting,” published materials with lists of symptoms or behaviors that they claimed were indicative of “attachment disorder” (e.g., lack of a conscience, cruelty to animals, not being affectionate on the caregiver’s terms; Hughes, 1997; Randolph, 2000). Third, many of those practicing these beliefs specifically marketed their services to adoption and foster care parents and caseworkers. The “attachment disorder” explanation made intuitive sense to those attempting to help children with severe problems and offered hope that change was possible, thus increasing demand for recognition and treatment of “attachment disorder.” Finally, although developmental researchers have historically examined issues related to adoption and foster care, clinical research generally has not focused on this population outside of the infant mental health field. As a result, little empirical work related to clinical practice was available to counter the faulty assumptions and misconceptions that were taking hold. Developmental researchers collectively were in agreement that neither the symptoms described as “attachment disorder” nor the earlier DSM-defined criteria for RAD described disordered attachment behavior. Clinical researchers primarily were busy developing interventions for far more common presenting concerns, and few were interested in researching concepts that were poorly defined and lacking a sufficient empirical basis. This confluence of factors resulted in the applied clinical field coalescing around the concept of “attachment disorder” in spite of a complete lack of empirical validation for the purported syndrome. The common clinical wisdom accepted the attachment disorder construct rather uncritically and propagated the belief in this construct down to trainees through the years. It remains the case that children with significant maltreatment histories and subsequent emotional and behavioral problems are being diagnosed with EVIDENCE-BASED PRACTICE IN CHILD AND ADOLESCENT MENTAL HEALTH 65 D ow nl oa de d by [ Pe nn S ta te U ni ve rs ity ], [ B ri an A lle n] a t 1 2: 50 1 3 Ju ly 2 01 6 RAD or “attachment disorder,” and along with this diagnosis comes the presumed belief that “attachment therapy” is required (see Woolgar & Scott, 2014, for an excellent discussion of this issue, including case examples). (Mis)defining “attachment therapy” To begin this discussion, consider the following exchange I recently had with a child welfare case manager: Case manager: “Do you provide attachment therapy?” Me: “What do you mean by attachment therapy?” Case manager: “Umm ... Well, do you treat children with attachment disorder?” Me: “What do you mean by attachment disorder?” Case manager: “You know, children with attachment problems. Like, children who were abused and now they’re aggressive and manipulative and don’t care about their caregivers.” Me: “Yes, we treat that.” Case manager: “Oh, good. So you do attachment therapy?” This brief example is meant to highlight the challenge with defining what exactly is meant by the term “attachment therapy.” Oftentimes, the term is used specifically to refer to the treatment of children who are described as exhibiting “attachment disorder” with little consideration of what the treatment approach actually entails. As just discussed, the term “attachment disorder” is not empirically sound or scientifically accepted, and RAD as defined in the DSM-5 is so exceptionally rare that most clinicians will never encounter it in their professional career. What, then, can be defined as “attachment therapy?” Historically, clinicians who promote themselves as attachment therapists have not restricted themselves to treating problematic attachment behavior. Rather, these clinicians generally believe that they are applying attachment theory (incorrectly, as it turns out) in describing the child’s current problems primarily as the result of maltreatment within their early attachment relationships (e.g., Association for Training on Trauma and Attachment in Children, n.d.). In this manner practically all of the child’s problems are symptoms of “attachment disorder,” as they are traced to the early maltreatment experiences. Simply put, “attachment therapy” can be defined as treatment designed to mitigate the impact of early caregiving/maltreatment experiences, with the belief that this will then remedy the current presenting problems. Within this broad framework, numerous techniques have been described as “attachment therapy,” including forced and coerced holding approaches; power assertive parenting techniques; treating a child as if she or he were an infant by prescribing bottle-feeding, rocking, or other experiences of infancy (i.e., age regression); as well as traditional individual approaches such as nondirective play therapy and sandtray therapy (see Allen 2011b, and Mercer, 2015, for reviews). In essence, any treatment technique that the clinician believes may lessen the damage caused to the child early in life by an attachment figure may be implemented. The observant reader will notice several significant disconnects here between attachment theory and treatment approaches described as “attachment therapy.” First, attachment theory views the child’s behavior as a function of previous experiences and current circumstances (e.g., attachment behavior can change with circumstances, IWMs can be modified based on new evidence). Given that one cannot change the past and a child cannot cognitively process experiences he or she does not remember, empirically derived attachment theory stresses attempting to change behavior bymodifying a child’s current environment. Second, development and social interactions continue beyond infancy, and attachment theory values those later experiences as similarly important to understanding the child’s behavior (Bowlby, 1988; Sroufe, 2005). As such, treatment focused solely on mitigating the impact of early life experiences is oversimplified and neglects much of attachment theory and research as well as development that has occurred since infancy/toddlerhood. Third, attachment theory suggests that the child’s attachment behavior and current internal working models are largely the result of the caregiving received. Thus, the primary treatment target should be improving the caregiver’s responses to the child’s behavior, as these interactions will serve to improve both attachment behavior and internal working models (Allen, 2011b). The field of “attachment therapy” grew considerably along with the acceptance of “attachment disorder” as a diagnosis. Some clinicians, although distancing themselves from holding and coercive therapies, directly state that other purportedly attachment-based treatments are the only effective means of treating attachment problems (e.g., Becker-Weidman, n.d.; Buenning, n.d.). Given that more academically oriented clinicians and researchers were not accepting of this diagnosis, these therapists were the only ones providing a potential solution to the perceived problem. “Attachment therapy” as a field 66 B. ALLEN D ow nl oa de d by [ Pe nn S ta te U ni ve rs ity ], [ B ri an A lle n] a t 1 2: 50 1 3 Ju ly 2 01 6 remains poorly defined, with significant variability among practitioners identifying themselves as attachment therapists. There is no compelling empirical evidence supporting any treatment approach identified as “attachment therapy.” When providing treatment from an empirically derived attachment perspective, one must remember four key aspects: (a) treatment should focus on improving the functioning of the child–caregiver dyad; (b) the child’s behaviors and representations are in response to the caregiving received and, therefore, the quality of the caregiving is the primary target of change; (c) treatment should be present-focused with a goal of improving the child’s developmental trajectory; and (d) the child’s cognitive abilities should be respected and considered (see Allen, 2011b, for a further discussion). In practical terms, numerous attachment-derived recommendations and guidelines suggest that clinicians utilize evidence-based interventions focused on enhancing the caregiver’s ability to understand the child’s behaviors and emotions and to respond sensitively to the child’s needs (American Academy of Child and Adolescent Psychiatry, 2005; Chaffin et al., 2006; Zeanah & Gleason, 2015). Some excellent examples are available from the growing body of research on reputable attachment-based interventions with infants, toddlers, and preschoolers, such as Attachment & Biobehavioral Catch-Up (Bernard et al., 2012), Child–Parent Psychotherapy (Lieberman, Ghosh Ippen, & Van Horn, 2006), and Circle of Security (Hoffman, Marvin, Cooper, & Powell, 2006). Studies of these interventions often document a positive impact on attachment behavior/security and other attachment-related constructs. Clinical interventions with older children can make use of the same general directives while remaining sensitive to developmental differences. Some currently available evidence-based parent-training interventions are easily understood from an attachment perspective. For instance, Allen, Timmer, and Urquiza (2014) provided a discussion of how Parent–Child Interaction Therapy coincides with attachment-derived treatment directives and provided pilot data with adopted children. Similarly, O’Connor, Matias, Futh, Tantam, and Scott (2013) experimented with the Incredible Years parent-training program and found that the intervention was successful for improving attachment-related parenting outcomes.