The October 2009 meeting of the American Academy of Child and Adolescent Psychiatry (AACAP) included a research forum entitled “Investigating the Effects of Psychotherapy on Children and Adolescents: Learning from the Past—Moving Forward.” The research forum, organized annually by the AACAP Workgroup on Research is a day-long set of presentations and discussions meant to address an issue of current and central importance to the field of child psychiatry. The chairmen of this year's forum, David Shaffer, David Brent, and Neal Ryan, selected this topic because they felt that an important juncture in child psychotherapy research had been reached. In the past several years, a series of adequately powered (i.e., with a large enough sample size to reach meaningful statistical conclusions) multi-site randomized controlled trials (RCTs) have been published comparing medication and psychotherapy for common disorders of childhood and adolescence. These trials include:
MTA: The Multimodal Treatment Study of Children with Attention Deficit Hyperactivity Disorder (ADHD), comparing medication, behavioral therapy, and combined treatment
TADS: Treatment for Adolescents with Depression Study, comparing medication, cognitive behavioral therapy (CBT), and combined treatment
CAMS: Child and Adolescent Anxiety Disorders Study, comparing medication, CBT, and combined treatment
POTS: Pediatric Obsessive Compulsive Disorder (OCD) Treatment Study, comparing medication, CBT, and combination treatment
ADAPT: Adolescent Depression Anti-depressant and Psychotherapy Trial, comparing medication alone with medication plus CBT
TORDIA: Treatment of SSRI-Resistant Depression in Adolescents, comparing medication, CBT, and combined treatment
See links on page 26 for further details.
Despite the high methodological quality of these studies (according to the standards of the psychotherapy RCT community) the findings were far from uniform or conclusive. In two studies (CAMS, TORDIA), combination treatment or the addition of CBT to medication was superior to either medication or CBT alone, and both single treatments were superior to placebo. In one (POTS), combination treatment proved better than CBT alone which was better than medication which was better than placebo. However, in three studies (MTA, TADS, ADAPT), medication, combination treatment, or the addition of CBT to medication achieved similar effects and were both superior to CBT or a placebo. Meanwhile, most of these studies noted differences in the success of particular psychotherapies at one site or another, despite the use of a standard manual. Finally, researchers and administrators interested in disseminating “evidence-based practices,” in child psychotherapy, such as CBT, have expressed their growing frustration with the larger psychotherapy community for failing to uniformly embrace these therapies and with parents for not insisting on them.
These circumstances led Shaffer, Brent, and Ryan to pose the following questions: “How can we respect parent choice for or against a particular mode of treatment?” and “Psychotherapy is more time consuming and expensive than psychopharmacology…Is it reasonable to expect a more enduring outcome from psychotherapy and how do we interpret studies that fail to show this?” The 2009 AACAP Research Forum arose from these questions.
The questions posed by the research forum, as well as some of the answers suggested by forum participants, have important implications for research in psychodynamic psychotherapy (PDT) and psychoanalysis for both children and adults. Because of the advocacy of the AACAP Committee on Research (chaired by APsaA member Efrain Bleiberg), I delivered a presentation at the forum entitled “Empirical Research on Psychodynamic Psychotherapy: Can We Count What Counts?” This talk was co-authored by Nate Thomä, a Fordham University graduate student in clinical psychology.
My comments were divided into two sections. First, I did not fight the ground rules implicitly presented to me by the other members of the forum, as well as by the medical community as a whole, namely that RCTs sit atop a “hierarchy of evidence” and provide a unique set of data about whether a treatment, such as psychotherapy, is efficacious (i.e., works in the controlled setting of a careful experiment) or effective (i.e., works in the real world of community practitioners and clinics). It is not that I think RCTs are without flaws or that other methods of data gathering (open trials, process studies, case series, or even single case reports) do not have merits
Andrew J. Gerber, M.D., Ph.D., is chair of the Committee on Scientific Activities, a candidate at the Columbia University Center for Psychoanalytic Training and Research, and a clinical fellow in the New York State Psychiatric Institute/Columbia University Division of Child and Adolescent Psychiatry.
which RCTs lack. Psychodynamic researchers, particularly Drew Westen and Sidney Blatt, have argued powerfully and clearly against an unquestioned hegemony for RCTs, and I share their views. However, I do believe that RCTs allow us to systematically address the frequent challenge that all of the successes of PDT can be attributed to therapist bias or patient self-selection, even if these same studies fail to address a myriad of other important questions (e.g., what are the benefits of allowing patients and therapists to select one another and to decide together what the goals of the treatment should be?). And perhaps more importantly, I think it is a mistake to combine the already difficult battle to have PDT research recognized with a still harder, and maybe insurmountable battle, against the special status of RCTs. Rather, it makes more sense to do the studies that are being asked of us, even if it means temporarily sacrificing some of our long-term research goals.
Psychodynamic researchers, particularly Drew Westen and Sidney Blatt, have argued powerfully and clearly against an unquestioned hegemony for RCTs, and I share their views.
Fortunately, the data is on our side and this is what I presented at the AACAP forum. An ad hoc Subcommittee of the Workgroup on Research of the American Psychiatric Association, chaired by James Kocsis and including several dynamic researchers including Barbara Milrod, Steven Roose, Jacques Barber, and me, developed a 25-item measure of the quality of an RCT of psychotherapy. We rated all 70 RCTs of PDT (mostly in adults) that we found in the literature (through 2007) and found that the quality of these trials is steadily rising. (Unfortunately, there was not a single RCT of psychoanalysis for us to include.) More impressively, when Nate Thomä rated 105 RCTs of CBT for adult depression, using the same scale, he found the rising level of quality to be indistinguishable from studies of PDT. In other words, RCTs of PDT, though still fewer in number than those for CBT, are of just as good quality and are getting better at the same rate.
Interestingly, the vast majority of RCTs of PDT and CBT show the same pattern of results. When a well executed, carefully specified treatment (either CBT or PDT) is compared against a weaker treatment (e.g., a watered-down version of an existing treatment or “treatment as usual”) or no treatment at all, the well executed treatment wins. However, when two well-executed, carefully specified treatments are compared against one another, the study ends in a tie. Unfortunately, as Barbara Milrod has recently pointed out in an editorial in the American Journal of Psychiatry, such a tie tells us very little. When studies fail to show a difference between two treatment groups we cannot tell whether this is because the treatments are equivalent or because the sample size was just too small to detect a difference. In the jargon of psychotherapy research, these studies are not “powered for equivalence” and given how difficult it is to do studies of this size, it is unlikely that this will change in the near future. These findings are the latest confirmation of what Saul Rosenzweig famously referred to as the “Dodo Bird verdict” in psychotherapy comparison trials, quoting Alice in Wonderland: “Everybody has won and all must have prizes.” Though an important part of what I presented at AACAP, this was not intended to justify the abandonment of RCTs for PDT. Some in the CBT community, notably John March who presented at the forum, have stated, “The horserace is over and CBT has won.” We need more RCTs to show that this is not true. Our goal is for the horserace to be over, indeed, as these studies fall short of what we truly want to know, but first we need sufficient RCT evidence to demonstrate that PDT is useful for a wide range of adult and child psychopathology.
The second part of my AACAP presentation attempted to move the discussion forward in another direction. I proposed that the most interesting goal in psychotherapy research is for us to learn more about the components of so-called brand name psychotherapies such as CBT, interpersonal psychotherapy (IPT), and even PDT itself, and to determine in what patients, in what circumstances different types of interventions produce what kinds of change. Psychodynamic researchers have made a great deal of progress in developing reliable and valid measures for this purpose. Beginning with the problem of diagnostic assessment and the widely recognized failings of the categorical approach of the Diagnostic and Statistical Manual of Mental Disorders (DSM), I explained the Shedler-Westen Assessment Protocol. This 200-item Q-sort measure (items are placed in piles from most applicable to least applicable, following a set procedure for how many items are allowed to be in each pile) calls upon a clinician's skills to describe an individual in jargon-free yet highly descriptive systematic language. Next, I presented two continuous, multi-axial measures of individual functioning developed by psychoanalysts for the purposes of going beyond simple symptom description: Per Høglend's Psychodynamic Functioning Scales (PFS) and Fonagy and Target's Hampstead Child Adaptation Measure (HCAM).
Another important and unique contribution of dynamic researchers has been the development of scales that evaluate specific therapeutic interventions. These include, though are not limited to, Høglend's Specific Transference Techniques Scale (STSS), Barber's Multitheoretical List of Therapeutic Interventions (MULTI), and Jones's Psychotherapy Process Q-Sort (PQS). All are meticulously designed assessments of particular interventions in psychotherapy that we need to study in order to learn in what circumstances each does or does not have utility.
Finally, I presented the work of Susan Andersen at New York University recently replicated in our own laboratory at Columbia, demonstrating a way to show that healthy adults use specific prototypes of their own significant others to inform how they perceive and recall new information about people. Whether or not we call this, as Andersen does, “transference,” it seems likely that this basic psychological process has something to do with what we are addressing in psychotherapy. That is, if we are able to help individuals know more about the ways that they see their current situations through the lenses of the past, it seems likely that this will address some of their symptoms. To what extent this kind of intervention is the primary province of dynamic treatment or takes place in other therapies as well, is something we need to know more about.