Sunday, August 02, 2015

The most important conclusions from the Franken study on early-childhood intervention

Here is my summary of the Franken study (see here and a StutterTalk interview) after a first reading. Add your comments if you disagree!

The two most important clinically relevant statements are
  1. If your child has been stuttering for six months, does not have any obvious speech and language development issues and is treated by a specialist, it has a three-in-four chance to fully recover within 18 months of treatment.
  2. It is not important whether your child receives direct (Lidcombe) or indirect (DCM) treatment.

Here are safe statements:
  1. It is the outcome trial for close-to-onset kids with by far the largest sample size, 200, and best follow-up data, 18 months.
  2. It is about comparing the relative performance between two different intervention methods (direct and indirect)
  3. Both methods have similar outcome with no clear and significant difference.
  4. It is not about absolute performance comparing to natural recovery (i.e. recovery without professional therapeutic intervention).
  5. It contains no strong evidence whether they work, whether they only speed up natural recovery, whether they only improve psychosocial adaptation, and why they work.
  6. It is wrong to say that the direct method Lidcombe is the only scientifically tested method.
  7. It is by far superior in quantity and quality to the Lidcombe outcome trial but lacks the crucial control group out of ethical reasons.

Statements one cannot make safely but might be true:
  1. The trial might have picked up subtle differences between the two methods.
  2. The direct method might both be a bit better at dysfluency reduction in the short-term but also show relapse in the long-term.
  3. Only allowing kids who stuttered for six months or more in the trial should make the natural recovery rate lower in the treatment arms.
  4. Only allowing kids without other developmental issues in the trial should make the natural recovery rate higher in the treatment arms.

What still needs to be done:
  1. We absolutely need 3-year, 5-year, and 15-year (after teenage years) follow-up data to study relapse and recovery.
  2. A follow-up of the rejected kids or kids that did not do the treatment for various reasons. For example, they could be used to create a fake control group. 
  3. An analysis of sex-specific differences. For example, we could give a more specific prognosis to kids depending on sex and family history.

Some concerns I have
  1. I cannot understand why the two samples were not randomized to %SS so that they both start with the same %SS, because the %SS is 4.9% for direct treatment arm and 4.0% for indirect treatment arm.
  2. The article's wording and arguments are a bit sloppy, e.g. "... that both direct and indirect treatment... reduced stuttering..." should be "that both treatment arms showed reduced stuttering" as it is not clear that it is the treatment method and not natural recovery or the claim that the follow-up period is double the time as in Jones et al. even though they have a follow-up paper with low quantity and quality or the natural recovery is "estimated to be 63%" but I have also seen much higher numbers.
  3. The lack of a control group due to ethical concerns.


Ora said...

The authors write "Our results do not enable us to distinguish the potential effect of treatment from spontaneous recovery."

I'd be interested to see the same graph showing spontaneous recovery (without treatment).

Tom Weidig said...

But that is exactly the key issue. We cannot have a control group for ethical reasons.

But I would expect it to be very similar to the indirect method approach.

We should also not forget that spontaneous recovery is not really spontaneous, and that natural recovery of course includes all the corrective measure that the brain, the child, and its social environment undertake. This is a kind of treatment but not focused and directed and not done by an expert.

Tim said...

The claim that we can't have a control group for ethical reasons assumes the very thing to be tested by a control group! The movement to early intervention began long before there was ANY good data on the efficacy of early intervention therapies.

Tom Weidig said...


I agree with you and I have said that many times before.

How can it be unethical if you haven't proven it yet.

Here are some reasons why it is still not possible in the real world:

1) Parents just don't accept it. Especially many mothers are very concerned (bordering to irrational panic) and want to do the best possible for their child. For them, it is emotionally much more satisfying to have tried anything even if it did not work, because they can also tell themselves: I have done everything I could. If they don't do anything, they might blame themselves for ever.

This was also the reason why the original Lidcombe trial was unable to keep up the control group any longer because the parents rebelled.

2) Even if you have a control group, it is not a real control group because everyone knows whether a child is in treatment or not. With a placebo pill, it is different. No-one knows whether the child gets treatment of not.

3) Ethical boards are often filled with a certain kind of people. They are cautious and rather want to deny something that could have an upside rather than support something that could have a downside.

4) The Lidcombe group has shown, not very convincingly, that being in treatment has an impact on stutternig behaviour in the *short-term*.

Bruce said...

Hi Tom

I am a little confused reading these comments. I appreciate the difference between a placebo pill and a treatment but why can't the intent be replicated? If you can make up a pill with nothing but an ineffective powder in it, why can't you make up an ineffective treatment?

To be honest I would rather be on the end of an ineffective treatment for stuttering than a placebo pill for an incurable cancer, but the latter actually sounds the more likely scenario.

Tom Weidig said...

Hi Bruce,

1) children are in a delicate phase of their development. It is argued that it is unethical to withhold treatment because their brain is much more adaptive at that age than let's say 6 months later. For adults postponing treatment is not that much of an issue, also because they can give consent.

2) making up a fake treatment means that therapists would need to fake treatment, they would still need to be paid, will probably refuse to do so or disclose it. etc....