Thursday, November 8, 2012
Diagnosing Reactive Attachment Disorder with Scales and Questionnaires
I’ve recently been having a correspondence with a birth mother whose two school-age sons have been placed in treatment with an attachment therapist by their stepmother. One child is now being placed in residential treatment. The attachment therapist, who is also evaluating the children and recommending a custody arrangement (he’s a social worker; a psychologist would not be permitted to wear two hats like this), says he has diagnosed one boy with Reactive Attachment Disorder by means of an unidentified scale.
What scale might this be, I wonder? Can there be a scientifically validated test of this type? Psychological scales and tests are generally developed in one of two ways. The first, the one used with intelligence tests for adults, involves developing a large set of questions, throwing out all the ones that everyone or no one answers correctly, and then determining what proportion of the population gets specific scores on the remaining questions. An IQ score is a shorthand statement of how well the person did in comparison with a lot of other people. (Tests for children are a bit different and need to be done separately for each age group.)
In the second method, there is some known accurate way to detect or diagnose a personality characteristic (for example), but it takes a long time to do it or requires special training. The test developers seek information that is easier to get and that correlates highly with the known accurate diagnosis. When the test is developed, it allows less trained people to diagnose in a short time a condition that otherwise takes a lot of work on the part of an expert. This is a good outcome if there really is a reliably accurate way of diagnosing, and if the new test is strongly correlated with that way of diagnosing. If there is no reliable way to diagnose a condition, or if the proposed test does not correlate very well with the accurate method, it may not be useful to employ the new method.
Both of these methods of test development require meticulous work and data analysis. The second method also demands a good design in which people collecting information from the test do not know that diagnosis someone has previously given the tested individual, for example.
Although there has been a good deal of research on the proportions of young children who show different attachment statuses, there is little on that topic for school-age children, and even less on proportions of children diagnosed with Reactive Attachment Disorder. One of the reasons for this lack of research is that no one has put forward a valid technique of diagnosing Reactive Attachment Disorder. The lack of such a method obviously also makes it impossible to follow the second method and develop a scale or questionnaire that gives an accurate diagnosis.
So what scale could the attachment therapist mentioned above have used in his diagnostic efforts? There have in fact been some attempts to create such a test, but so far all present serious problems.
The Randolph Attachment Disorder Questionnaire (RADQ), published by Elizabeth Randolph in 2000, is one of the most questionable efforts along this line. For those who have heard of this or even used it, let me point out an awfully important point:
The RADQ is not intended by its developer to diagnose Reactive Attachment Disorder. It addresses a different set of problems, posited by some attachment therapists and called simply Attachment Disorder.
Calling the test the RADQ and the disorder RAD does suggest that the test is about Reactive Attachment Disorder as commonly defined, but no--- the R is for Randolph.
Whatever the RADQ is meant to diagnose, it does not really matter very much, because Randolph herself observed and diagnosed the children and supervised the parents who filled out the questionnaire, as well as doing the data analysis on information whose background she knew. To comment on this is not to attack Randoph’s integrity; no researcher should trust himself or herself to avoid biases about known information. That’s why everyone is expected to be “blind” to any information that could skew diagnoses or measurements or mathematical analysis. Randolph did not take care about this, and as a result her complicated calculations amount to “garbage in, garbage out”. In addition, she did not clarify what she was looking for when making her own diagnosis, and since in a 2001 publications she stated that she could diagnose RAD from the fact that a child could not crawl backward on command-- well, need I go on?
IN spite of the shortcomings of the RADQ, Sheperis and his colleagues (“The development of an assessment protocol for Reactive Attachment Disorder”, Journal of Mental Health Counseling, 2003, Vol. 25, pp. 291-310) used the test along with various others and proposed that this battery would assess RAD in school-age children. However, in 2005, Cappelletty, Brown, and Shumate examined correlations between the RADQ and other sources of information (“Correlates of the Randolph Attachment Disorder Questionnaire (RADQ) in a sample of children in foster placement”, Child and Adolescent Social Work, 2005, Vol. 22,pp.71-84) and found that the scores for children in foster care (who would presumably be most likely to be categorized as having RAD) were not significantly different from those of other children. These facts suggest that it would have been a mistake for the attachment therapist to have been using the RADQ to produce the diagnosis.
What about other scales? Thrall, Hall, Golden, and Sheaffer (“Screening measures for children and adolescents with Reactive Attachment Disorder”, Behavioral Development Bulletin, 2009, Vol. 15, pp. 4-10), who apparently can’t spell Cronbach, used two questionnaires which they said discriminated between individuals who had been diagnosed previously with Reactive Attachment Disorder and those who had not. They noted that for those children with a RAD diagnosis, “the diagnosis of RAD, disinhibited subtype, had been made by a mental health practitioner specializing in attachment disorders” and that the children were now in treatment. This raises a question about the original diagnosis, because although most child mental health practitioners consider attachment issues to be of importance, very few of them would describe themselves as specialists in attachment disorders unless they were much involved with the whole attachment therapy belief system. Thus, it’s questionable whether the questionnaires used by Thrall and her colleagues would actually have discriminated between children diagnosed with RAD by conventional practitioners and those receiving other diagnoses.
One more approach to be considered here: Helen Minnis, a Scottish psychiatrist, has spent years in trying to develop scales to diagnose RAD. (The Thrall paper used one of her questionnaires.) In 2010, Follan and Minnis (“Forty-four juvenile thieves revisited: From Bowlby to Reactive Attachment Disorder”, Child: Care, Health, and Development, Vol. 36, pp. 639-645) proposed a battery of tests to identify RAD for research purposes (in other words, with less accuracy required than we would demand for individual evaluation). Very interestingly, Follan and Minnis argued that concerning behaviors were probably due to early maltreatment and genetic vulnerabilities rather than to separation, and that current secure attachments could exist side by side with the condition called Reactive Attachment Disorder. They also noted that “insecure attachment is likely to be very common in child psychiatric populations and although its measurement is useful to develop a holistic understanding of the child, is not an indication of disorder in itself.”
To conclude, then, there are some points to be made with respect to the attachment therapist who is using a scale for diagnosis of RAD. The first is that there is no scale of this type that provides the sort of reliable and valid assessment that should be required for use in legal decisions about custody of children. A second is that there may be no reason whatsoever to seek a specific diagnosis of this kind-- other than the fact that insurance claims demand one. As Follan and Minnis note, whatever the concerning behaviors displayed by a child, they may have little to do with attachment per se, and therefore, I would point out, do not require treatment that focuses on attachment. In addition, evidence-based treatments focus on specific characteristics of children and parents rather than assuming that they indicate attachment problems; treatments that make attachment issues the central concern are not at this time evidence-based, although some practitioners make this claim.