Response: How one surgeon discusses BrAIST

[Pooka1]

This are issues researchers discuss. Cautionary comments have been posted by Dr. Hey among others.

Lay people cannot be expected to be interested or even follow along as we have seen in this thread.

[flerc]

The only other topics I've seen the experts talk about are: overtreatment (from most experts), long-term effectiveness (from Dr. Hey). But the central finding of the study - that braces are effective in keeping teens off of the operating table - has been accepted by every expert I've read..

The only place where I've seen that aspect of the report questioned is in these discussions.

If you would be someone wanting to confuse parents, trying to hide/distort important facts showed in braist studies, wanting to convince them than surgery is the only valid scoliosis treatment, what would you do? EXACTLY WHAT POOKA1 IS DOING OF COURSE! or helping her to do what she does in this forum.. the most visited scoliosis forum of the world. May someone imagine a better way to promote scoliosis surgery? I can't.

[hdugger]

On the expected long-term results from bracing.

From everything I've read, once one has stopped bracing, your chances of progression in adulthood are figured from your ending curve. And the risks are the same as any other person with that size curve.

So, assuming a <6 degree increase in brace (the expected average) - if you brace at 30 degrees and wear your brace as prescribed, the most likely outcome is that you'll enter adulthood at <36 degrees. You then have the same chance of progressing in adulthood as a kid who didn't brace and enters adulthood at 36 degrees.

If you want to know *before* you brace what your risk of progressing in adulthood is, the best numbers are to take your current (pre-brace) curve, add 6, and then figure out the risk that adults with that curve face of progression. Obviously, no one knows how it will turn out for a specific kid, but this gives you some sense of what you might expect.

[hdugger]

Back to the opening quote (in the other topic, where I couldn't post)

Here's Dr. Hey's headline about the BrAIST study:

"Are Scoliosis Braces Adolescent Idiopathic Scoliosis Effective for Stopping Curve and Preventing Surgery? New Weinstein Randomized Control Trial Published Today NEJM and Presented at SRS says "YES!""

And that simple "yes, they're effective" is exactly what I've seen from every other expert commenting on this report. That's the take-away, amongst orthopedic surgeons.

I'll just note that nowhere in any response I've seen to the report, from Dr. Hey or any other expert, are people suggesting that the effectiveness numbers *in the windows where effectiveness is measured* are going to change. The only place where I've seen that aspect of the report questioned is in these discussions. Given that - given that the numbers show effectiveness and that every expert I've read says that the study showed effectivenss - I'm going to accept that conclusion. Should some other study come down the road that shows something else, I'll re-examine that conclusion. But there is nothing in the study or in the expert response to the study that would lead me to any other conclusion.

The only other topics I've seen the experts talk about are: overtreatment (from most experts), long-term effectiveness (from Dr. Hey). But the central finding of the study - that braces are effective in keeping teens off of the operating table - has been accepted by every expert I've read.

[Pooka1]

One SD is the accepted standard. .

You completely missed the point.

[hdugger]

You are missing about one third of the variability by only showing one s.d.. .

One SD is the accepted standard. You're not trying to map all of the data - you're just giving a measure of variability around the mean.

On the rest, I believe we talked in depth about their study design before they started collecting data. I don't recall much resistance at that point to their measure of success. So, I'm resistant to complaints about this measure which only surface after the study is done.

Whatever their measure of success, it applies to the unbraced controls as well. So, it's the exact same measure being used in the discussions about over treatment.

I'll note that, again, that the measure of success in surgery is walking out of the hospital a few days later. I suspect that that, also, is not what a teen might think of as success, but it is how it is measured in the research.

[hdugger]

Not knowing the final curves nor the long term makes these numbers hard to interpret.

I'm not certain that I find it hard to interpret. It addresses one of the burning questions we've had about bracing - does it work or will the positive results we've seen disappear as soon as we randomize the patients. The answer to that question is now in - yes, the positive results we've seen were actually due to the brace.

Because we *can* interpret the data to answer that question, we have a tendency to rush forward to all those other questions that we've been (largely) ignoring while we worked out the effectiveness. You've raised some of them here. But I didn't really expect this study to address all of those questions. I see it more as the gate we've been waiting behind. Now that that gate is open, we can go on to really pin down which kids benefit, how we can keep the curve as small as possible, how other more tolerable braces stack up, and so on. So, for me, this just answers the preliminary question, and now we can go on, with other studies, to work out the rest of this.I don't expect that this study will answer these questions, even with more data points and a longer time frame. But I think those answers will come from future studies.

[flerc]

Hence why I keep asking why it's necessary to (continually) discuss the choices made in *this* study. It might leave someone with less experience in research wondering if this study is some special case. To which I'd respond, No. It's not. It's the same as any other study, except that I think it's actually somewhat better thought out, designed, and analyzed, and reported then most other medical studies.

I thought the follow up was more extense and the sample greater, but I'm not sure really . Even I saw other studies showing a greater succesful percentaje, but I believe the variable values as kind of brace, time in brace, reduction in brace.. were almost the same in all cases of those samples, something not happening in this study.
Anyway the conclusion about 50% effectivenes is right (under the study conditions) and is what parents should to know and of course is what brace haters here cannot accept.. hence it seems they wants this empty discussion never end..a good way to confuse parents.

[Pooka1]

There are things you can measure but, without ancillary information, are ultimately uninterpretable. Not knowing the final curves nor the long term makes these numbers hard to interpret.

Also, if I was a child, I would want to see all the variability and not just 68% of it.

The data a child would want to see and the data a researcher shows in a publication are not necessarily the same.

[Pooka1]

There is nothing special about this study apart from other studies. Many choices are available for graphing. They all address the hypotheses.

[hdugger]

Choices were made. .

My point was that the data display you were asking for - the end curve at the end of the study for every participant - would not more clearly address the central question of this study. It would be a poorer way of answering the question *these* researchers were looking at.

Beyond that, yes, I'm familiar with the choices made in putting together a medical research paper. The same choices are made in *every* research paper we've ever discussed on this forum. It's nothing unique to bracing, this study, these researchers, etc. Hence why I keep asking why it's necessary to (continually) discuss the choices made in *this* study. It might leave someone with less experience in research wondering if this study is some special case. To which I'd respond, No. It's not. It's the same as any other study, except that I think it's actually somewhat better thought out, designed, and analyzed, and reported then most other medical studies.

[Pooka1]

There are many ways to graph those data to address those exact hypotheses. This is what I am trying to get across. Various ways emphasize various aspects of the data.

I am waiting for subsequent publications. These will certainly have some graphs I am talking about.

Choices were made. Many choices. I don't know how else to state this. Ask any experienced researcher about this and they will reel off the choices if you tell then the study design.

[hdugger]

In my own work, I choose one way or others for a reason. These authors are doing that too. .

My sense is that the graph comes directly from the design of the experiment. Specifically, there was a lot of talk in the bracing research, before this study, about how we had to not just focus on how many degrees kids progressed (what had been shown in most of the previous studies) but in how this played out in terms of whether or not the kids progressed to surgery.

So, I assume, in response to *that* concern coming from the scoliosis medical community, they designed *this* study to look specifically at success (which they defined as avoiding surgery by study end).

So, while I get what you're asking, there are lots of studies (in fact, *most* of the bracing studies) which do what you ask. They say how many degrees these kids progress in brace. The most commonly reported number I've seen is 6 degrees. The other thing you ask for - the end point - is not as useful, in my opinion, because it depends so much on the starting curve. If I have a kid with a 25 degree curve, what I want to know is where *my* child will end up, not where lots of kids starting with lots of different curves will end up.

But, anyway, what we'd like to have a study focus on is no nevermind. This study is what this study is. The focus on surgical success was deliberate. It wasn't some mistake on the part of the authors, or some afterthought in how to present the data - it is specifically what they set out to study. And then they chart that. To me, it's all in keeping - what they want to achieve, what they do, and how they report it.

[Pooka1]

Medians versus averages. If those 29 points in each quartile are not normally distributed then median is correct. That might change some hearts and minds.

One s.d. = only about 68% of the variability. If you expand to include 95% (two standard deviations) then they would overlap. You are missing about one third of the variability by only showing one s.d..

And it's not avoidance of surgery, is <50* with up to 25% of the growth remaining. So a 49* with 25% of growth remaining would be a "success". Outside of BrAIST, who considers that a success?

What you are missing is that the data can legitimately be plotted a number of ways that may or may not lead to the same confidence. In my own work, I choose one way or others for a reason. These authors are doing that too. There is nothing nefarious in that.

Instead of summary stats, I would just like to see the raw data plotted. I think it would be interesting. It would show the range in "successes" given a particular average hour of brace wear, something a kid needs to know in deciding how much time is worth it.

[hdugger]

Here's a graph that I would like to see . . Given the errors bars on the original graph . . .

Correct, we don't have that data. Nor do we have a graph showing the success rate of the unbraced kids.

But we *do* have a table of success percentages with error bars.

And what we can see from that is that there is *no* overlap between the success rate of the kids wearing their braces at least 13 hours a day (the right half of the graph) and those kids wearing their brace less than 6 hours a day (the left quarter of the graph).

So, what we can say:

If you wear your brace 6 hours or less (and I think most of these kids are "or less") you have a roughly 25 to 60% chance of avoiding surgery, with the most likely outcome being a 40% chance of success.

If you wear your brace 13 hours a day or more, you a 70% to 99% chance of successfully avoiding surgery, with the most likely outcome being a 90% chance of success.

Again, I do not understand all the time spent imagining data which is not in the report. Obviously, that data is not available. Why not focus on the data which *is* there?