Family Interaction and Health: Recalling Demise of the Psychosomatic Family Model

Families Family Interaction and Health Washington DC 09-page-0This slide presentation occurred at a NIH conference in 2009, but in important ways they are even more relevant now. Claims of being able to influence biomedical variables are important for getting papers accepted into scholarly journals and getting media coverage of the resulting articles. Such claims are also important for securing funding.

Thus, health and family psychologists feel pressure to demonstrate that they can affect not only adjustment to physical illness, such as quality of life, but also physical health outcomes such as progression of cancer, recurring heart attacks, or mortality. Claims of directly influencing biological processes, rather than indirectly, as through improving adherence, are given more attention, but typically have less scientific support. It is characteristic of hot topics that standards are relaxed for making claims so that hot topics often generate more false discoveries and exaggerated findings that cannot be replicated, as described by the work of John Ioannidis.

There have been consistent failures to demonstrate that psychological interventions directly affect the biology of cancer, tooth fairyincluding survival, but it does not stop claims from being made. I have referred to claims so-called psychoneuroimmunology, as tooth fairy science. That is because explanations are sought for claims when the basis for these claims has not been established. It’s just as if a child is doing experiments by wrapping up baby teeth in various ribbons and packaging to see what the tooth fairy favors, without the tooth fairy’s existence ever having been established. I got this concept from fellow Science Based Medicine blogger  Harriet Hall.

The slideshow starts with references to the Psychosomatic Family Model which was popular in the 1970s. The model focused on direct links between observation of family interaction and physiological changes in adolescents with anorexia, asthma, and poorly controlled diabetes. My papers (1, 2) with Barbara J. Anderson pointed out the flaws in in both experimental studies and the basic logic of a model.

We argued that Minuchin and colleagues had not demonstrated links between family functioning and clinically significant outcomes. In hindsight, their sample sizes were absurdly too small to expect to support such claims. But we also focused on the negative implications for working with families, particularly those with adolescents with poorly controlled diabetes. Essentially the Psychosomatic Family Model provided the basis for blaming conflict between the parents about other issues for their adolescent’s difficulties managing diabetes.

Minuchin argued that the adolescent having difficulties managing health conditions served the function of distracting the parents from their own problems. So, the parents had to resolve their own marital problems in order for their adolescent to achieve the control of the diabetes. Minuchin went further in arguing that if a clinician attempted to deal directly with the adolescent’s health problems, the parents and adolescent would sabotage the efforts. The adolescent’s health problems served a homeostatic function, like a thermostat. If the adolescent’s health problems decreased, parental marital problems would increase, stimulating a renewal and even an aggravation of the adolescent’s problems. Real relief of the adolescent’s problems required a restructuring of the family. Minuchin’s approach to therapy therefore was termed structural family therapy.

Here is the case example that sparked Dr. Anderson’s and my skepticism about the model. A high school athlete with diabetes had showed up at the emergency room in a state of ketoacidosis. He previously had meticulously managed his diabetes. In fact, he had established routines of eating and exercising such that he didn’t even need to check blood sugar levels. It was unrealistic to believe that he no longer needed to check blood sugar levels, but his control had been good enough to avoid problems so far.

Before Dr. Anderson I could reach him in the emergency room, a psychologist who knew very little about diabetes but was familiar with Minuchin’s model, arrived first and summoned the teen’s parents. Discovering that they had recently divorced, she insisted that the team was trying to keep them together by suddenly having problems managing his diabetes. The only evidence that she had for this inference was that his difficulties managing his diabetes had an onset right after their divorce. Both the teen and the parents expressed outrage and refuse to talk to the psychologist anymore, which convinced the psychologist that she’d hit on an important issue and so understandably they had become defensive.

Interviewing the teen I learned that he took great pride in his previous management of his diabetes. What had recently disrupted his regimen was that he’d shifted from football to wrestling, with a very different exercise schedule. I asked him about his meter and he said that he no longer even had the batteries for it. I suggested that he get batteries. Against the medical advice of the other psychologist, he and his parents left the hospital. They were strongly inclined to do so before I talked with them.

Two weeks later the teen was back in the emergency room. He apologized profusely and said that he had gotten the battery for meter, but was determined to recalibrate himself so that he didn’t need it. He had thought he was successful until the ketoacidosis recurred. He promised to keep checking his meter and never have to come back to the emergency room again. We kept in touch by telephone and he indeed succeeded in avoiding the emergency room.

Dr. Anderson and I felt that a little knowledge of the Psychosomatic Family Model on the part of the other psychologist was a problem. Her confidence in the model that kept her from gathering the basic information about the circumstances of the teen’s nonadherence. If she had carefully inquired about this, she would’ve seen an opportunity for a simple intervention that stayed focus on the teen, not his parents having gotten divorce.

In the papers that Dr. Anderson and I wrote, we criticize the idea of symptom serving functions as being too glib and as interfering with good clinical observations of who was doing what when. We also took issue with the Minuchin’s key concept of parental overprotectiveness. It was being used by therapist to blame parents for their children’s intractable medical problems. We instead developed a concept of “miscarried helping” in which parents or other carers can get involved in increasingly on helpful ways when they see their efforts failing and when they see a patient failing to get control of a health issue. We suggested a “principle of charity” and attempting to understand these efforts, rather than jumping to the assumption that the carers gain something by the persistence of the patient’s problems.

Bernice Rosman and Bruce Baker responded to our first critique on behalf of the Minuchin, but declined to reply after we wrote our second paper.

The Minuchin claimed that their model only applied to adolescents with “brittle diabetes,” not all children and teens having trouble managing their diabetes. In the 1970s, there were some children and teens being hospitalized with poorly controlled diabetes, after previously having managed their condition well. By the 1980s, it was recognized that physiological changes associated with adolescence could disrupt previously effective routines for managing diabetes. But the solution was to understand the difficulties and bolster new routines and new self-monitoring. This responsibility should be left at first with the adolescent. If he or she were incapable of achieving better adherence, the parents could step in. With a new understanding of diabetes and adolescence and more adherence-focused interventions, the diagnosis of “brittle diabetes” fell out of favor as vague and unscientific or even invalid.

In this slide presentation, I recall the demise of the Psychosomatic Family Model, and try to extract some lessons that we should have learned. A key one is that we should not jump to the assumption of direct links of family interaction and physiological processes, particularly when they involve inferences about variables that we cannot observe or measure. Wherever possible, we should first rule out adherence as the issue and try to understand how family interaction patterns either facilitate or impede better adherence.

I focus on some of my own work involving heart failure patients but also cancer patients undergoing radiation therapy. In each instance, I tried to make hypotheses about observable and potentially modifiable interactional processes. In specific instances, these hypotheses can be tested clinically by asking the right questions about adherence. In terms of research, they can be tested with manipulation of adherence, such as in randomized controlled trials of interventions to improve adherence.

I close the presentation with some work that seems to demonstrate a link between marital interaction and wound healing. This work is theoretically important, but does not necessarily demonstrate that changing marital interaction can facilitate wound healing in a clinically significant fashion. In the absence of evidence that it can, it would be best to focus on simple biomedical efforts to increase wound healing.