First, I understand that your main argument is that a patient has a right to say, "It's my body and my illness, I've a right to decide myself!" I won't argue with that, or that experiments can through up surprises and these shouldn't be ignored because of inconvenience. Personally, I think that one big advantage Bayesian statistics is that te method allows for additional information to be incorporated very simply as when the new information becomes available, but that's another matter.
One of my points was that by considering only a subsection of the Gleason scale after the results were known, they are screwing up the statistics. It's analogous to tossing a coin, losing, then saying, "best of three." Three tosses is fair, so is best of three. But if one player chooses the number of throws after the first toss, that is not fair. Yes, there might be a relationship with the Gleason scale, and it could have been reversed. If the poor response had been concentrated in the lower end of the scale, then the company would have argued for approval in that range. The chances are that there would be more response in some parts of the population than others - nothing is distributed uniformly.
People are individuals, but I disagree strongly with the idea that the experience of others can not inform the individual. If two out of three people get sick after eating five-day-old reheated turkey, I don't know whether I'd be in the class that would get sick, but I'll stay clear. And I'll continue to stay clear the next day, even though the stats might no longer be valid.
The exciting thing about new the new genetic advances is that they are opening the way for much more individualised treatment. Your right that that could, I suppose, create more difficulty with FDA : Ok, we know there are six genotypes, prove that the drug is safe and efficacious for individuals of each type. On the otherhand a drug that might have been discarded could be approved for , say four genotypes. Again though, this should be established before the experiment. Otherwise the stats might be as meaningful as noticing after the event that while two out of three died, all Asians survived the experiment. Perhaps the company would ask the FDA for approval limited to Asians?
Ashley |
