Wednesday, January 14, 2009

Side Effects of Behavioral Treatments

[Note: in this article, 'behavioral treatments' does not refer solely to the field that BF Skinner and Ivar Lovaas were involved in, but to any treatment in which the 'active ingredient' is some kind of experience rather than a biological agent. For example, psychoanalysis, floortime, music therapy, remedial education and sensory integration therapy are all examples of behavioral treatments.]
A lot of people understand that biological treatments, such as medications, can have unwanted effects. Some people understand that these unwanted effects can be permanent. Therefore, most people don't want to try some biological treatment unless they feel pretty sure it's the best option. (Granted, some people aren't so cautious, but many are.) Most people understand that making the wrong choice regarding a biological treatment can have serious impacts.
But a lot of people don't understand that behavioral treatments, too, can have side effects. They think 'well, it can't hurt, may as well try it'.
This view is encouraged by the people who offer behavioral treatments, because unlike medications, most behavioral treatments have no studies into the potential side effects. Indeed, many behavioral treatments have few or no studies of any effect of the treatment, much less side effects. However, even treatments such as ABA, which have a lot of research into their effectiveness, seldom study side effects. When I searched on PubMed for '"applied behavioral analysis" side effects', the only study I found was "Behavioral epidemiology of food additives" (PubMed ID 299572), which seems irrelevant. (PubMed has a tendency, when it can't find what you searched for, to select something that might have some of your search terms but isn't actually what you're looking for. It's just a quirk of their search engine.)
A big part of the problem is that the people studying a behavioral treatment are often quite blinded by their particular theory. With biological treatments, researchers tend to pick what treatments to try and how to measure effectiveness based on their theory, but side effects seem to be a mismash of theorized potential side effects based on several theories as well as things that are just common side effects of medications and whatever other side effects someone happens to notice. With behavioral treatments, however, many researchers seem to stick so much with their theory that they can't see the side effects, because those side effects don't fit their theory (in theory, the treatment is typically just fine).
In ABA, for example, the few times they happened to mention side effects, these are solely behaviors. Very often they're talking about something they actually wanted (ie generalization) rather than an adverse effect. The few times they discuss adverse effects, these are increases in unwanted behavior, often fairly minor things (eg a child punished for climbing on furniture starts sitting on the back of her chair instead of on the seat) which are felt to be an attempt to find other ways of getting the reward they used to get from the target behavior. The treatment is typically to change contingencies so these side effects are eliminated the way the target behavior was.
In other words, only the side effects that fit their theory are seen, and typically they're interpreted in a way that assumes the treatment is good. This is true in so many contexts. Bruno Bettelheim described children regressing in bowel/bladder control in his school, which he viewed as a sign of progress - a form of self-expression in a child who previously couldn't show those feelings at all. Another common method is to describe the side effects as being a sign of an improperly done treatment, which can get pretty ridiculous when you start claiming (as Ron Leaf did at the conference I attended) that 90% of people using your favorite treatment are doing it wrong, on no evidence other than that the treatment had adverse effects. While people involved in biological treatments often say that using them improperly is a common cause of adverse effects, only the quacks think their treatments can't have adverse effects when used properly. (The problem is that almost all the people involved in behavioral treatments basically think like quacks, even many of those using well-documented treatments.)
Given that it's so hard to find accurate information about adverse effects of behavioral treatments, many people assume these do not exist. But they do. Here are some ways to estimate the risks:
  • Look for stuff written by people who oppose that treatment. Studies are best, because then you'll have documentation. But if they don't have studies to back them up, you'll have to go to the studies by the proponents, and 'read between the lines' to see if there's evidence that supports what the opponents say. Focus on the data, not the interpretations of it, to find this evidence. For example, if you find stuff by opponents of ABA saying that extinction of a useful behavior can cause learned helplessness, and then find a study in which a child showed reduction of many different behaviors when one behavior was being extinguished, you've found some evidence in favour of the opponents' theory (not proof, but evidence suggesting that theory).
  • Imagine how you'd feel in that situation. If you think you'd have trouble with that, why? If you think you'd be fine, what could change to make you have trouble with it? Imagine it in the context of each theory you find. For example, to assess the risk of adverse effects of faciliatated communication on an autistic person, you'd have to try to imagine yourself both as a person who is truly communicating through FC and as a person who is having words 'put in their mouth' by FC. When assessing the risks for others involved, use the same kind of method - eg how would you feel if you thought your child was communicating with you but it was just the faciliatator, as opposed to how you'd feel if your child really was communicating. Remember, when assessing these, that the different theories aren't necessarily equally probable. Use what evidence you can find to assess the probabilities.
  • Lastly, keep watch on the impact it has when you try it. A good marker of the risk of a treatment is how happy or sad the person is when being treated. If they're happy, usually (not always, but usually) the worst that can happen is that the treatment doesn't help. If they're unhappy, that's when the risk of harm is higher.

Labels: , ,


Blogger Alison Cummins said...

Thank you Ettina.

As an autonomous NT adult, I've experienced severe side effects from talk therapy, but have never seen a serious mainstream discussion of the risks. Sometimes a talk therapist will point out that a risk of therapy is that you will get in touch with negative feelings that you had been avoiding, and that you might cry or otherwise feel bad. But I have never heard anyone say that a risk of talk therapy is losing confidence in your own judgement.

The routine advice to seek out a talk therapist with professional qualifications implies that any problems with talk therapy are necessarily due to "doing it wrong," not to the nature of talk therapy itself.

Again, thank you.

1:02 PM  
Blogger Lindsay said...

Wow, great post.

I think you are right that side effects of behavioral treatments are invisible to the people promoting them --- and to most health-care professionals and researchers.

(At least, the null result you got searching on PubMed seems to argue for this. Also, WRT the tendency for PubMed to yield completely unrelated results that might contain one or two of your search terms, I see that on almost every database I search. I think it's because the search engine is only looking at whether certain words or phrases occur at all, and not at the context in which they occur).

I think part of the invisibility of these side effects is a bias in medicine toward the physical and the quantifiable --- it's a lot easier to prove, and to explain mechanistically, something like, say, neuroleptics causing tardive dyskinesias than something like behavioral therapy causing excessive inhibition (or whatever). The one thing I can think of that I have seen mentioned as a side effect of behavioral treatment is PTSD, and that almost always invites the "doing-it-wrong" copout.

I also think there's a tendency, certainly when writing about neurodiverse people, but maybe even when writing about NTs, to avoid mentalizing. This probably also grows out of the bias in research, and in clinical practice, toward the concrete and mechanistic over the subjective and experiential. It also has roots in simple ableism, though: our feelings, our experiences, are less real to them because we are different. We're wrong, our emotional responses are themselves pathological. So when we have an emotional response to a treatment, it's much less likely to be connected to the treatment and more likely to be blamed on our own affective inappropriateness.

4:51 PM  
Blogger Dr.Rutledge said...

Hi Ettina,

I'm an academic physician (formerly at Harvard and Stanford) who found your blog while looking for the best health writers. I think your writing is great! I would like to feature you in the Autism & Autism Spectrum Community on Wellsphere, a top 10 health website that has well over 2 million visitors monthly.

If you would like to learn more, just drop me an email to

4:17 PM  
Blogger Ettina said...

That sounds really cool! I'm honored that you think my writing is that good.

7:29 AM  
Blogger Andrea Shettle, MSW said...

Ettina: I wonder if there might actually be a few studies on the "side effects" of various behavioral or talking interventions, but with different terminology than the terminology that is used to talk about biological medications.

For example, did you try, "risks" as a key word instead of "side effects"? Often in the phrase "risks and benefits."

Then there might be further discussion somewhere of what the consequences can be when therapy is "done wrong", though I'm not sure what key words to try for that. Maybe variants on "incompetent", or "malpractice"

6:32 AM  

Post a Comment

<< Home