Martha Welch Meets Bruno Bettelheim, Edgar Cayce, and the Rules of Research Design

I’ve devoted several posts and parts of some publications to the claims of Martha Welch, a New York psychiatrist, who used to claim that she could cure autism and now says she can prevent it. Welch has, inexplicably, the support of Columbia University, which has appointed her a sort of peripheral non-teaching faculty member with certain faculty privileges, and which not long ago gave her an award as a distinguished alumna.

In the 1980s, Welch visited Evergreen, Colorado, home of Holding Therapy as practiced on adopted children who were said to have attachment disorders. I have no idea exactly what happened there, but after her visit Welch decided that autism must be a form of attachment disorder, curable by facilitating emotional attachment to the child’s mother. She developed a new form of treatment that she named Holding Time. Rather than the Holding Therapy methods, which usually restrained the child in the therapist’s lap and used intrusive and painful shouting and poking to intimidate the child, Welch proposed that smaller children be held face-to-face with the seated mother and embraced so they were pressed belly to belly with her. This hold was to be maintained for an hour or more at a time, every day, as the child screamed and fought for release and eventually gave in and relaxed in the  mother’s arms. Older children were to lie on their backs while the mothers lay prone above them, partly supported on their arms. In both cases the mothers were to speak to the children, communicating all of their negative and positive feelings; the children too communicated a good deal of negative emotion.

Welch’s approach would probably have remained among the various obscure “complemetarty and alternative” approaches to autism, except that in the 1980s she made a connection with Elisabeth Tinbergen, a special education advocate and the wife of the Nobel laureate in medicine Nikolaas Tinbergen. Niko Tinbergen was a specialist in ethology, the study of innate, species-specific behavior patterns; although ethology had focused largely on insects, fish, and birds, the 1960s and ‘70s were a time when studies were attempting to find such species-specific behaviors occurring independently of culture and learning in human beings. The Tinbergens saw parallels between what Welch was doing and the occurrence of “imprinting” connections to other animals in ducks. They enthusiastically wrote a book declaring “hope” for autistic children in Holding Time methods, and included a long appendix written and illustrated by Welch. The Tinbergen book led to the publication of Welch’s 1989 Holding Time and an extensive book tour in Europe during which Welch presided over roomsfull of mothers holding tightly to screaming autistic children. (The Tinbergens also encouraged the promulgation of similar methods by Jirina Prekopova, a Czech psychologist working in Germany.) However, little was made of a point clearly stated in the Tinbergens’ book: that there was no empirical evidence to support the effectiveness of Holding Time.

I’ll leave this brief summary now because there is much more to say about Welch’s current activity. She is now running a program called Family Nurture Intervention and claims that physical contacts between premature babies and their mothers can reduce the frequency of autism, which occurs more frequently in premature than in full-term babies. You can see a demonstration and discussion of this method at https://www.youtube.com/watch?v=DIXAMaIOK64
 (my thanks to Yulia Massino for calling my attention to this video; for some reason I cannot create a link here). Welch is carrying out a research project on the effectiveness of Family Nurture Intervention as a prevention of autism and has described the research plan, a randomized controlled trial, at https://bmcpediatr.biomedcentral.com/articles/10.1186/1471-2431-12-14. Among other things, Welch proposes that autism is related to the presence of abnormal levels of peptides released from the gut, for example secretin and oxytocin (both popular suggestions for treatment of autism in CAM circles).

I want to comment on three aspects of Welch’s theory and research. First, her assumption that babies become autistic because they miss experiences of physical contact, exchange of odors, and so on with their mothers is a direct reflection of the very much outmoded “refrigerator mother” theory put forward by people like Bruno Bettelheim (an art historian who became focused on child mental health). Bettelheim claimed, without evidence, that cold, unemotional mothers caused autism by their failure to engage physically and emotionally with their children. He did not mention premature babies, but those babies can certainly be included in this hypothesis because their medical care makes it difficult for parents to touch or engage with them in pleasurable ways, and even the most loving parents of premature babies will not be able to care for them as they would normally want to do. In the video mentioned above, a scientist who works for Autism Speaks points out this problem of Welch’s approach—that she is accepting something like the “refrigerator mother” theory in spite of decades of evidence that genetic factors rather than experiences cause autism.

Second, Welch’s emphasis on literal “gut responses”, the production of peptides and their effect on the brain, is an example of an old theme in American CAM. Edgar Cayce, an early- 20^th-century “prophet and seer”, emphasized the work of the “enteric brain” as a source of emotion. The Meridian Institute of Virginia Beach, VA has continued to promulgate Cayce’s ideas, which are still part of many CAM approaches. For example, Andrew Wakefield, the disgraced physician who started the claims that immunization caused autism, looked to gut contents as evidence that materials of immunization remained in the body, and, presently working in America, he continues to focus on digestive tract issues as causes of autism and ways to treat it, in contradiction to all that is known about autism.

Now, the fact that Welch’s Family Nurture Intervention smacks loudly of outmoded and CAM beliefs is not enough to condemn it. It would be very reasonable to say, “who cares if it’s based on wrong ideas? If it works, that’s what’s important.” And Welch has reported some preliminary positive results—but this leads us to the third problem about Family Nurture Intervention studies.

Welch and her colleagues state proudly that they are carrying out a randomized controlled trial, and most certainly it is randomized, with families assigned randomly to one of two treatment groups. The problem is that it is not controlled, despite the presence of a putative comparison (control) group. Here is the trouble: according to the BMCpediatrics article mentioned earlier, the two groups are receiving treatments that are different in a number of ways other than the Family Nurture Intervention itself. In the FNI treatment group, mothers have 4 1-hour intervention sessions per week, or more, plus standard care. The comparison group has no specified treatment, standard care, and 1 meeting a week with research staff in the hospital. In other words, the FNI treatment groups is experiencing 4 or more times the socially-supported time with their babies as the comparison group receives. Such social support is well-known to improve mother’s moods and therefore to help them interact more patiently and warmly with their babies.

The point of randomization is to help isolate a variable, which is the crux of all experimental work. Randomizing families to conditions means that no special characteristic of the people (like knowing about and wanting a treatment) can interfere with finding the effects of the treatment itself, which should be isolated from any other possible effects on the outcome. But randomization alone cannot isolate the effects of the treatment unless the rest of the design and implementation of the study are correct. The time spent in treatments is clearly a factor that needs to be controlled, as it is common for more of a treatment to have a greater effect on outcomes than less does. When a treatment involves social support for people in a stressful situation, the time spent and the attitudes of staff are important factors. This study should have provided 4 or more sessions per week for the comparison group, devoting them perhaps to supportive interviews and motivational discussions with the mothers. Otherwise, we can only conclude that having more sessions has a better outcome--  we can’t conclude that the Family Nurture Intervention itself is supported.

A 2015 publication by Amie Hane et al (with Welch as a co-author), https://psychology.williams.edu/files/Hane-et-al-15-JDBP.pdf, concluded that mothers’ care for babies in the NICU was improved by experience of the Family Nurture Intervention. However, this article very properly stated some limitations of the study.  “The control group received standard care (SC) (i.e., usual care) as part of our randomized controlled trial that included holding and skin-to-skin care if the mothers chose to engage in these activities. But, there were not corresponding control conditions for each of the specific activities of FNI. For instance, SC mothers did not exchange sham odor cloths with their infants (cloths that were not exposed to mothers and/or infants). The effectiveness of FNI should thus be interpreted as a function of a comprehensive and integrative intervention, since during this preliminary trial of the intervention control manipulations of FNI activities were not entirely possible.” (p. 194). In other words, mothers in the comparison group may or may not have carried out some of the activities used for the FNI treatment group, but they did not share any version of some parts of the treatment, including the length of time per week (this last not mentioned as a limitation by Hane et al.). The standard care comparison group did not receive a family support session before discharge as the FNI group did.

Martha Welch has spent many years arguing in favor of alternatives to evidence-based conclusions about causes of autism and effective treatments for the disorder. Her connection with Tinbergen was apparently sufficient years ago to earn publication of a popular book and the support of Columbia University. The Columbia IRB approved her research design, weak as it is, although even non-harmful treatments can be considered as a harmful waste of time and resources if improper design and implementation of research can lead to incorrect conclusions. What is this all about? I would like very much to know.

N.B. Nothing in this post should be taken as rejection of the extensive legitimate research on neural factors in gastric and endocrine functioning; my point is that the existence of such physiological factors is not in itself evidence that supports any statement about autism.