Research In Psychotherapy: Outcome Research Versus Process Research

In my Psychology Today blogpost about research in psychotherapy outcomes, and in my last book, I complained about the inflated claims of researchers in psychotherapy - particularly those made by purveyors of cognitive-behavioral therapy (CBT). They grossly overstate both the power and the significance of their results.

Unfortunately, a new report by the Institute of Medicine (IOM) falls for this baloney hook, line and sinker. The report on psychosocial interventions for mental illness and substance abuse has drawn a wide variety of responses from the field - including praise, recommendations for improvement, and some sharp criticism from psychiatrists and mental health professionals who are experts on psychotherapy. I am obviously sympathetic to the critics.

Of particular interest is that the report lauds the so-called "evidence-based psychotherapies" - code for those therapies which are "supported" by the incredibly weak psychotherapy outcome studies. One critic, Peter Roy-Byrne, M.D, summed up the criticisms of the report as follows:" In medicine, there is usually an array of different treatments for the same condition because of individual variability that is still poorly understood. Yet the field of medicine does not spend its time trying to understand what are the common elements between various effective treatments, though it will often explore comparative effectiveness as a way of improving care. It may well be that different kinds of individuals and problems demand different psychotherapeutic approaches rather than that there is one elemental Holy Grail that will be best for everyone.”

Of note is that, at least if you believe in free will as I do, patients always can choose to either respond favorably or unfavorably to any intervention a therapist makes. It is just not all that predictable, because everyone can choose to respond differently. In fact, the very same intervention given to seemingly very similar patients can lead to responses that are completely opposite from each other - in one case the patient improves on some dimension, while in the next the patient may get worse! In psychotherapy, patients are very different from one another in ways that vastly outnumber individual differences that affect treatment outcomes in any other field of medicine.

Holly Swartz, M.D., of the Department of Psychiatry at the University of Pittsburgh School of Medicine, brings up another criticism of the IOM report: “The recommendation to reduce highly complex interventions to their component parts, however, is misguided. A bias toward CBT and CBT-based interventions constitutes an essential flaw in the IOM report, placing affect-focused therapies such as IPT (interpersonal therapy) at risk for unfair negative evaluation and, ultimately, elimination from our therapeutic armamentarium. … [T]he IOM report should advance an inclusive research agenda that reflects and supports the diversity of psychosocial interventions that the IOM purports to represent.”

Yet another problem with the IOM report and similar viewpoints is that they completely ignore the fact that there is a vast literature within psychotherapy research that does not focus on outcomes but on process. Process research looks at the moment to moment interactions of patients and therapists within the context of their particular relationship.

As Les Greenberg, Ph.D, puts it, "Research on change processes is needed to help explain how psychotherapy produces change. To explain processes of change it will be important to measure three types of outcomes—immediate, intermediate, and final—and three levels of process—speech act, episode, and relationship. Emphasis will need to be placed on specifying different types of in-session change episodes and the intermediate outcomes they produce. The assumption that all processes have the same meaning (regardless of context) needs to be dropped, and a context-sensitive process research needs to be developed. Speech acts need to be viewed in the context of the types of episodes in which they occur, and episodes need to be viewed in the context of the type of relationship in which they occur.

Speech acts refers to the fact that speech does not merely convey information to, or exchange propositions with, a listener. Sentences do things that are frequently independent of the meaning of the actual words that are used. Speech causes others to perform acts. If I say, "I hear you're having a party Saturday," I am not describing only my recent experience of having heard about the party. I am probably fishing for an actual invitation. In all likelihood, I have made in advance a determination that this sort of statement is the best way to accomplish my goal of attend­ing the party. I have made a prediction about the future behavior of the listener. If this particular ploy leads to no response or a different response, I will consider alternative strategies.

Those who tout psychotherapy outcome studies, which study psychotherapy interventions as if they occurred in some sort of relationship vacuum devoid of context, seem to want to pretend that the highly significant process research literature does not even exist. In fact, the vast majority of articles published in the journal Psychiatric Research are process studies, not outcome studies.

Finally a Study on Borderline Personality Disorder That Looks at Family Dynamics

Karlen Lyons-Ruth, Ph.D.

When it comes to the study of personality disorders as described in medical journals, it seems like most of what's been published lately in the field looks at which superficial aspects of personality disorders and personality variables correlate with one another and to what extent, or how often some construct like "impulsive aggression" is found in various populations. Such correlations usually turn out to moderate at best and never ever close to being 100% predictive of anything. 

While group data like this may help therapists prioritize - to a minimal degree - which other issues in a patient's life on which to focus clinical attention whenever they happen to note one of these characteristics, it really tells us nothing about the individual patient. Not to mention that it neglects literally thousands of other important details that may apply only to the patient at hand.

IMO, such studies are just different ways of slicing up the same old bread, when it is the bakery that needs the attention. The bakery is the interpersonal environment in which personality disorder traits develop and flurish.

Of course, as readers of my blog know, it's very difficult to study the bakery "empirically," because a lot of what goes on involves subtexts or unspoken communication in family relationships. These subtexts not only change over time but, especially in the world of personality disorders, the subtexts of family interactions can contradict one another. To make matters even worse, some of these contradictory messages may show up only rarely, but become significant when seen in light of the history of the family over the lifetime of the participants as well as the family's history over several generations. All of this is almost impossible to quantify.

I was pleased to see in a recent issue of the Journal of Personality Disorders that, for the first time in a long time, some researchers tried to address these issues in at least a partially comprehensive way (Karlen Lyons-Ruth, PhD, Laura E. Brumariu, PhD, Jean-Francois Bureau, PhD, Katherine Hennighausen, PhD, and Bjarne Holmes, PhD: "Role Confusion And Disorientation In Young Adult-Parent Interaction Among Individuals With Borderline Symptomatology." Journal of Personality Disorders, 29(5), 641–662, 2015). 

Dr. Lyons-Ruth and her colleagues did their best to look at what they so nicely referred to as the "minute to minute fabric of interaction..."  Their data lent much support to my idea that, as a manifestation of what evolutionary biologists call kin selection, children sacrifice their own needs in order to stabilize unstable parents.

In fact, while I previously thought of so-called parent-child role reversal as applying only in cases in which children act more like parents to their childlike parents than the other way around, I realized one could conceptualize the spoiling behavior seen in borderline personality disorder (See the post The Family Dynamics of Patients with Borderline Personality Disorders) as just another form of children taking care of unstable parents - although a much more subtle and covert form.

In so called "empirical studies" to date, the quality of parent-child interaction has rarely been observed directly. In this study, 120 young adults were assessed for features of borderline personality disorder, for severity of childhood maltreatment through interviews and self-report measures, and for disturbances in parent-child interaction during a videotaped conflict discussion task. The last part addressess a huge criticism of the literature I have been making: no one actually looks directly at what is going on at home.  

Of course, the way the family behaves in a laboratory setting is only an approximation of what goes on when no one is watching, and much important behavior undoubtedly gets omitted under this type of observation (as oppposed to when Supernanny spends a couple of nights in the home) - but at least it is something that can be observed directly. To paraphrase a line from the movie The Big Short that showed how certain people were able to see through the fraud perpetrated by mortgage bankers prior to the burst of the housing bubble in 2008, "If you want to understand a process, actually looking at it is a very good place to start."

The conclusions of the study: Borderline traits, as well as suicidality and self-injury specifically, were associated with more role confusion and more disoriented behavior in interactions with the parent.

The definitions of these terms, quoting more from the article: In infancy, the term disorganized refers to the apparent lack of a consistent way of organizing attachment responses to the parent when under stress. The types of disorganized behaviors observed in infancy include apprehensive, helpless, or depressed behaviors, unexpected alternations of approach and avoidance toward the attachment figure, and other marked conflict behaviors such as prolonged freezing or stilling or slowed “underwater” movements.

Controlling children, on the other hand, “Actively attempt to control or direct the parent’s attention and behavior, and assume a role which is usually considered more appropriate for a parent with reference to a child. Controlling behaviors are thought to represent a compensatory attempt by the child to maintain the involvement and attention of the attachment figure by taking over direction of the parent-child relationship.

Two forms of this controlling behavior were observed. Controlling-caregiving behavior is characterized by organizing and guiding the parent or providing support and encouragement to the parent (e.g. child praises the parent or asks if a parent is all right). Controlling-punitive behavior is characterized by episodes of hostility toward the parent that are marked by a challenging, humiliating, cruel, or defying quality (e.g., child tells parent that the parent is terrible at doing a task). The study found that, among young adults with recurrent suicidality/self-injury, 40% displayed high levels of role confusion compared to 16% of those who were not suicidal. 

Neither form of disturbed interaction mediated the independent effect of childhood abuse on borderline symptoms. In other words, these patterns contributed to the development of the disorder in an additive fashion to the usual culprit of child abuse. As the authors say, "even when present, abuse events are often only the tip of the iceberg and index pervasive difficulties within the family in establishing caregiving routines, managing anger, and maintaining discipline"  (p.643) (Italics mine).

One interesting sidelight was noted on p. 656: with depression controlled (that is, subtracted as a factor in the mix), young adult-parent interaction associated with borderline traits included a particularly contradictory combination of role confused/caregiving behavior and punitive behavior. This contradictory mix is "similar to the mixed forms of childhood controlling behavior toward the parent that predict elevated BPD features in adulthood...", as had been noted by the first author in a previous work from 2013. This odd mixture of punitive and caregiving behavior "converges with the often-described shifts from idealizing to devaluing behavior experienced in therapeutic interactions with borderline patients."

This means that the so-called "splitting" seen in patients with BPD may be triggered by the demand characteristics of interactions with parents as they change over a period of time. This finding is highly consistent with what I have written about extensively.