Discussion of the statistics of the NEDU study on the redistribution of decompression stop time from

[rossh]

So the A1 profile can't be replicated? Also, Ross, when discussing science, let's please not quote experimental design guidelines from a Middle School web page.

The issue here is that the baseline pDSC rate, with all the backing of the NEDU experiences, proofed invalid..... the base line was not repeatable....

 

[Gozo Diver]

The original version, as reported in the UHMS decompression workshop 2008, the experiment method does NOT make mention of the complex mid point test.

It's mentioned in the Results summary (where the midpoint interim analysis is discussed) and p=0.047 is quoted there. I can't assume beyond that; I can only comment on what is reported.

 

[rossh]

It's mentioned in the Results summary (where the midpoint interim analysis is discussed) and p=0.047 is quoted there. I can't assume beyond that; I can only comment on what is reported.

...reinforcing my point... the half way stop was a reaction to the results, to avoid a forced rejection, and not a preset condition. The preset condition was added to the 2011 report only.

 

[Gozo Diver]

I cannot assume beyond what the authors report. If ethical considerations required that a statistically significant result from a one-tailed Fisher test meant that the trial had to stop, then the study would have to be terminated. As I said, I really can't comment beyond that.

 

[Beau640]

This argument comes up here once every few months and it’s always the same arguments back and forth.

We will never have the perfect study that we want to truly compare these two models. Just like in medicine, many times you have to make due with th studies you have to draw the best conclusions you can since you will never have the actual data or study you want (many times for ethical reasons which is also what is going on here).

Does anyone here truly believe the deep stop model is superior? Or just that it’s not inferior/there is no higher risk of DCS?

Right now we have two different models that are telling us to decompress in two different ways based on differences in physiological modeling and philosophy. Are the two models equal? Who knows. But what we have right now is data saying that one model MIGHT be better than the other. Even if you think that data is flawed and weak, why take the chance? If you start out thinking both models are equal and all the data we have says one might be better, why not just use the one that might be better? Even if they are both truly equal at the end of the day, why take the risk and chance?

During all of these repeated arguments on here I’ve never seen someone say deep stops are actually better and that everyone should be doing them, just that they don’t want people to think they are inferior because of the NEDU study. But again, even if they are both truly equal, if all our data (even if you think it’s weak) says one might be better, why not just use the better one?

These arguments have been clouded with people with financial interests in the deep stop models and personal attacks on character as well as wild accusations of unethical study practices and falsifying data. At this point, who are you going to believe?

 

[Dr Simon Mitchell]

Yes it does say that.... written after the test was done. We have no guarantee that was a decision point or value before the test.

Ross, are you accusing Drs Doolette and Gerth of fraudulently misrepresenting the endpoint of their study?

And even if it was a real planning point, its purpose is to act as a guard rail to avoid the impending rejection

How many times do you have to be told that there was no impending rejection: that like many other things in these discussions you have misinterpreted this?

and highlighting that the profiles were not an equal match in the first place.

What do you mean by this? If the authors truly believed the profiles were an equal match what would be the point of doing the trial?

What happened during the test? All persons involved were acutely aware of the running totals, and where it was headed. No sane person would let such an expensive undertaking run off the rails.

Like I said, I think they salvaged this test from imminent rejection, and the reason quoted above is the excuse.

Ross, there was no imminent rejection. Why would they "reject" a profile that was performing extremely well in comparison to another higher risk profile when the whole purpose of their study was to figure out whether the higher risk profile was superior or not.

And as I asked above, are you accusing Drs Doolette and Gerth of fraudulently misrepresenting the endpoint of their study?

Simon M

 

[Gozo Diver]

We will never have the perfect study that we want to truly compare these two models. Just like in medicine, many times you have to make due with th studies you have to draw the best conclusions you can since you will never have the actual data or study you want (many times for ethical reasons which is also what is going on here).

In fact, in case it might not be clear (although it should be), the aim of the article was to help those who might not understand the mathematics behind the study to inform themselves about what can be said on the basis of the data, it being as it is.

 

[rossh]

Right now we have two different models that are telling us to decompress in two different ways based on differences in physiological modeling and philosophy. Are the two models equal? Who knows.

Hi,

Right from the start, the experiment design was NOT equal.

In this thread we can see the A1 profile had a 2% head start. It was supposed to be a ~5% pDCS baseline, as per the extensive NEDU database, but it turned out to only be a 3% pDCS. But that error is never explained or accounted for. Instead that error is embraced as a winning feature.

Please understand what this test design was about. They took a basic plan with a known pDCS (the A1): a VVAL18 deterministic model. Then they created a delayed off gas version of it, by deliberately manipulating the stops towards the shallow end (redistribution). Then its fed through the BVM3 and probabilistic checks, to arrive at the same theoretical pDCS risk of ~5% (A2). As David states somewhere in RBW, the BVM model has the ability to take a plan and fit it to a desired pDCS.


So the comparison is, one ordinary 3% risk deterministic profile, vs a hand crafted ~5% risk probabilistic profile. No competition at all.


To that you can add the way the thermal stress was used to replace gas pressure stress, which has since been shown to be very unpredictable (TR06-07), and the different manner thermal stress affects different profiles types, adding another bias toward A1.


************

That's where this argument is.... Some people see the design and method as biased and unfair to start with, whiles other embrace the result for what it is.

.

 

[tridacna]

Talk about closing the stable door after the horse has bolted. Hard to imagine that this silly argument about this study is still going on. Ross, the world has moved on and Deep Stops is in the rear-view mirror.

 

[rossh]

And as I asked above, are you accusing Drs Doolette and Gerth of fraudulently misrepresenting the endpoint of their study?

Simon M

Simon, The nasty accusation above is your own...not mine. Please stop trying to project it onto me.


**********


What I'm doing is pointing out facts. Once again....


In the 2008 version of the report, the stated rules and conditions for early termination were the 3 and 7% rejection lines. In the two presentations that Wayne Gerth gave in 2008, he also only spoke of the 3-7% early termination rules.


In the 2011 version of the report, the above 3-7% condition is present, plus a new one:

"The trial was also to be concluded if a midpoint analysis after completion of
approximately 188 man-dives on each dive profile found a significantly greater
incidence of DCS (Fisher Exact test, one-sided α = 0.05) for the deep stops dive profile
than for the shallow stops dive profile"


Where did the extra 2011 rule come from? It's not in the 2008 reports.

This new 2011 rule and condition conveniently fits into the events as they happened.


****************

The test did stop half way.... because it was headed for rejection.... according to the rules at the time. Naturally it had to be stopped and salvaged, for any number of reasons. Its trend line and direction was obvious to all involved.

.