Thursday, February 17, 2011

Why apparent short-term success in Lidcombe treatment

Perti asks me a question about Lidcombe:
I am an SLP student who has been wondering about the Lidcombe program for a while [...] I've just been wondering about the Lidcombe success rates. How come the Jones et al (2005, British Medical Journal) had such a huge difference between the control group and the actual subjects. The field of stuttering is a complex one, as our professor said. One could fill a library with books that are all about stuttering and still none of them is absolutely right about it. The problem is that somehow they managed to pull off such a huge difference between the groups. The same incidents seems to occur with other studies as well like Miller et al (2009, American
Journal Of Speech-Language Pathology) and Latterman et al (2008, Journal of Fluency Disorders). The sample was very small so some of the differences could be explained by pure statistics. I don't really believe in completely behavioral therapy because stuttering has (probably) a mostly neurobiological background. I am wondering what were your thoughts about the systematical, but yet small data, concerning the success of Lindcombe. The spontaneous recovery rates might easily have huge differences between such a small sample groups but it's consistent between few studies that I've quickly read through.
 Here are possible explanations for the big difference:

1) Every adult treatment of nearly any type leads to a significant reduction in dysfluency. Therefore, we must NOT look at outcome after therapy, but six months or one year after therapy. And that's EXACTLY what happened in their follow-up study, some relapsed.

2) The trial were not blinded at all, and the control group did nothing. That's VERY different to the standard random control trial, where both groups are given IDENTICAL pills. Here only one group was given treatment, and the other group waited.

3) The children in treatment have much more familiarity with the therapy environment. They have been conditioned to behave in a different way. They have been told that stuttering, or non-control of speech, is undesirable.

4) The treated kids are used to perform in the clinic environment. For them, it has become a kind of playground, and so later measurements are probably biased downwards.

4) You cannot even talk about success of Lidcombe, because the success factor might be a component common to all interventions, e.g. parent-child intervention.

5) Sample size is small.

6) I would not be surprised if behavioural treatment makes kids who recover naturally recover faster. I speculate that the neurobiological issue resolves within weeks or months but that the learned behaviours are still there without the neurobiological basis and will only gradually wane. But treatment might eliminate these learned behaviours faster.

And more questions:
And I was also wondering about the high spontaneous recovery rate you mentioned (~80%?). That seems a tad high and if I remember right it was originally conducted by the Yairi group at the University of Illinois. A systematic review about the articles widens and drops the scale by a large margin. I have to admit that since stuttering isn't something I've really kept up with I can't remember the exact numbers and right now I can't open up the article database to check it... the clinical data I've heard of was also less than 80%. But by no means, it had absolutely nothing to do with science. This just came up so that's why I even mention about it. I am very interested about everything in a very strict and scientific way. That's why I am pretty suspicious about the Lidcombe program.
The number depends on the definition. What exactly is stuttering? For example, the latest large scale survey has many more than 5% of kids stuttering, leading to even higher recovery rates. But some criticise the too stringent definition.

I personally prefer to work with 80% or 90%, because I want to be conservative. If I say a lower number and I am wrong, treatments will suddenly be effective without being effective. But if I aim for a higher number and I am wrong, the burden of proof has just been higher for treatment efficacy.

The answer is that there is no real answer. The best I can say is that the vast majority of kids recover. As a first approximation rule, if your child stutters, expect that he or she will recover.

Charles Van Riper, the father of what we consider stuttering therapy, once said (I've got it on tape as well) that he had never ever seen any stuttering person cured -  except that maybe one or two adults were capable enough to control the stuttering so that they could avoid the persons they interacted with from noticing that they actually had some trouble with their fluencies. Charles Van Riper himself knew his 'techniques' very well and could control his stuttering but after listening to his audio recordings I could easily notice that he had some short pauses and other minor dysfluencies in his spontaneous speech. That is why I don't really believe in anything like that. They taught us that the early intervention has (probably) a positive affect on child's attitudes and feelings towards his/her stuttering. By addressing that early on the speech therapy would be easier and therefore more successful. As stuttering has a huge, huge and huge impact on the stuttering person it's really hard to work on stuttering when someone has been stuttering without any kind of (professional) support or help through his/her childhood and even early adulthood.
I agree.


Anonymous said...

"As a first approximation rule, if your child stutters, expect that he or she will recover."

Did that approximation rule work out for you, Tom? You recovered and all was well afterwards?

For all the rest of us who stutter, when that advice was given to our parents, it turned out to be very wrong. And it comes at a very high price, as we all know when we struggle every day. By the way, based on your constant complaining about others making fun of stuttering, it seems that for you too it comes at a high price.

Even if treatment could help only 1% of stuttering kids who would not show spontaneous recovery (and in reality all science shows that the percentage is much much higher, regardless of how much you hate that science for purely personal reasons), it would be wrong and irresponsible to tell parents that they should just assume that their stuttering child will recover without help.

Tom Weidig said...

I am talking about an approximation.

The statement "Your child will recover" is true in 80-90% of the cases. So it's fine as a first order approximation.

but fails in 10-20% of cases.

Anonymous said...

The problem is that your statement is not even true and that, even if it were true, you are potentially causing harm to stuttering children whose parents decide to not seek help because of your stating that one can "expect" that recovery will occur.

As someone who stutters yourself, why would you want to do that to even one single child who stutters? You might argue that it is the parents' responsibility to make the decision, but if you are the one telling them that one can expect that recovery will occur, then the responsibility lies with you.

So can you provide us with references to published papers (and given your claim to be a good scientist, you want to rely on replicated data of course) for scientifically-sound studies showing unambiguously that somewhere between 80 and 90% of the cases recover?

I have looked into this myself and can only find 80% as the upper end of the range (the range being 10-80% with an average around 50%) and that upper limit number came from a study in which kids were considered recovered by the researchers even if either a parent or SLP still considered the child to stutter (by the way, why do you never attack such problems in recovery studies? it is worse than anything that you can pick on in the treatment studies that you dislike yet you happily rely on these problematic data when they can be [mis]used in your fight against Lidcombe). So that upper limit definitely does not meet the criterion of being unambiguous and in fact the same researchers said around 60% in other studies by their own group.

So in the end, recovery most likely occurs in around 50-60% of cases and thus what you should tell parents is that "If your child stutters, expect that the chance of not recovering without treatment is pretty much the same as recovering -- that's bad odds. So I recommend that you seek professional help asap. It may be that your child would recover anyway, but better to offer help to someone who would also have recovered without it, then to withhold help from someone who will not recover and then the critical time window during childhood may be missed."

You actually have a very nice opportunity to make a POSITIVE contribution and HELP children who stutter (just at the start of what may be a lifetime of dealing with this tough problem) and their families. You could make yourself a really important force in this regard. Wouldn't attempts to help kids who stutter be energy much better spent than remaining caught up in negative feelings resulting from being ignored/contradicted by certain treatment researchers? Feeling like you are "losing face" would take only a few days and then after that you will feel rewarded for having done the right thing. Think about it, it would be a much more valuable contribution to the community of people who stutter.

Tom Weidig said...


I never said that I would say to parents that their kid will recover and that they should not go to the clinician for a first assessment and monitoring.

I am just saying that the first order approximation rule is that they will recover.

Again, the rate depends on the definition. If you look at all kids who have dysfluencies at some point, you get a very high recovery rate. Obviously, if you only look at kids in treatment after age 3, you get a lower recovery rate.

It is true that the high recovery rate makes the threshold higher for outcome studies, and that the rate itself is not clear.

That's why, there should really be a control group in order to have a consistent measure for both groups, but this will not happen.

There is a large scale study in Australia, which should have been published by now and where Onslow and Packmann are co-authors. That study certainly suggests a very high recovery rate.

I take a high recover rate, typically 80%, because I want to be conservative.

Your suggestion of me making a positive contribution is totally unscientific. I am commenting because I am making a point and i don't care about its impact.

What is a valuable contribution? To adhere to a consensus?

Perti said...

The thing is that there aren't any other well documented and scientifically tested treatment methods besides Lidcombe - especially for children. It's very hard to compare Lidcombe to "traditional speech therapy" because there really isn't one "traditional" way of doing it. Lidcombe involves a lot of conditioning but it involves the most important thing as well: speaking. Like when a child is asked to repeat a stuttered word/sentence in a controlled way. That's exactly one of the things that are done in so called traditional speech therapy.

Lidcombe just seems to put together lot's of old and new information. Maybe too much is put on conditioning but yet it's the only well documented thing I know of. The problem with Speech-Language Pathology is that efficacy studies don't exist in many instances. Lidcombe and LSVT (for Parkinson's disease) are one of the few ones I know of. Most SLPs don't document their work and progress well enough to turn it in to a scientific release. Too bad and unfortunate for the field.

Tom Weidig said...


I agree that Lidcombe is well documented intervention.

However, I dispute that it's scientifically tested and shown to be effective in eliminating stuttering in the long-term.

Moreover, there is no control group. So most effects could happen in other treatment approaches, too.

Perti said...

Yes, that's what I mean. SLPs need to address the clinical work in a way more scientific way. Right know we don't have any other scientifically proven and effective way for early intervention. Lidcombe is better than doing nothing - that's for sure. I don't know why or how but it just means that we don't know why people stutter. It has something to do with our speech motor system but yea... what? and how? Are the PET/fMRI findings the etiology or caused by the stuttering? During SLP training we are told to pick the most scientifically valid treatment option. Lidcombe has results - and some relapse (but not complete). So how do you tell a parent not to treat with Lidcombe when there really isn't another valid option for children? The classic way of learning to speak at a slower way, doing the Van Riper techniques, giving positive experiences of fluency, having fun and dealing with the psychological side... it's not tested in any way.

The main problem with classic stuttering therapy techniques is that those techniques are made for adults. Children are not little adults. We can't really turn an adult's program into something for children. Speech therapy with children is a lot different than working with adults. Especially when comorbidity with things ADHD exists. Doing those techniques is very very hard and slow.

Unknown said...

Data is more fractal than normal distributed. You have essentially a control system with a feedback loop. A small change may result in a big impact. You can't repeat the process as you can newer return to the same "starting point". Data is not normal distributed and a normal statistical evaluation will fail.

This means that care have to be taken when evaluating studies as well as making them.
People tend to miss significant indications and over estimate what they believe is correct statistically.

If you have a fractal behaviour and large number of sample you should en at noise(50%) and you will with 100% miss the significant indications that you are looking for.


Tom Weidig said...


I don't fully understand, but I quite of vaguely suspect what you are saying.

Can you please make your argument clearly and give a concrete example?