Ashley, thanks for pointing that out.
I'm with you on resisting those who trot out the line that statistics are a load of cobblers and can prove anything, which is usually an excuse for them to do what they want without the foundations of rationality being confronted.
In this instance, the problem looks to me like the bureaucratic one of more and more rules being less and less manageable until nobody can do anything. Or, under clause 242, subsection 3.312 everybody can do nothing.
WWII took 5 years [well, only 3 by the time the Yanks deigned to realize that Hitler was more toxic than CHOP]. If the rules were followed now, it would be 100 years to even get started. Yes, yes, I know that wartime is different from less urgent demands, but the principle is definitely a major issue in modern bureaucracies.
It might be the case that people lower on the Gleason scale can in fact benefit while those higher up don't.
One thing I learned in my expeditions into cancer treatment is that people are individuals and the only statistical information which matters is a single unit. The rest is trying to make comparisons between apples, oranges and potatoes. As we know, comparing apples and oranges is a common mistake.
Yes, there are broad similarities and they can all be killed by lack of water, and they all have membranes around their cells [other than the few which don't of course].
But each person's function is different. Very different.
Cutting the groups into smaller and smaller units is vital to find effective treatment. For example, in Non Hodgkins Lymphoma treatment, there were certain allele this and that which meant a person has only a 1:30 probability of successful treatment, whereas those without that particular characteristic enjoyed a 1:1 cure rate [actually, that might be 2:1 or 1:2, I can't recall the exact number].
Pooled, the 1:30 and 1:1 gave a probability of cure of around 1:3. [Or thereabouts]. But the poor buggers who had those wonky alleles didn't have a show [other than 3% of them]. The others were looking good.
Similarly, with the Gleason level. For all I know, that Gleason level is a correlation with an important issue for that group of individuals and yes, it should have been identified before the experiment started.
But it wasn't. To some extent, experiments find new stuff and it shows up in such statistical variation. Yes, it should have been properly designed to start with, but it wasn't. Meanwhile, WWII is on and people are dying.
If they want to try the product based on an inadequate study, they should be able to. It's not as though they are going to live to appreciate the statistical precision of later studies.
Admittedly, I haven't studied this particular product or the study or what the Gleason level means, but from what I've read so far, I'd say that if there's a 1:20 chance that the stuff is doing something, I'd take it [assuming the cost of treatment is reasonable and the benefits sufficient to pay that cost, or, more accurately, 20x that cost since there is only a 1:20 chance of it being good stuff - hmm, I might be getting my statistics tangled here, but you probably know what I mean].
My main argument is let people decide for themselves [with the advice of anyone they choose].
Mqurice
PS: Yes thanks, Xmas was enjoyable [though a niece's fiance's death 2 Dec from an industrial accident in Canada was very upsetting for her]. |
