Alois Data mining is a term used to describe going back through...

Alois


Data mining is a term used to describe going back through your results to find any sub group where results were better. As DNDN did in the early days and ELN more recently. What PBT found was a mistake, which caused the figures to look worse. It was not data mining, just checking accuracy of tghe figures for the complete trial population.

I dont have time, so read this Nobuddynoz post from investorvillage about the tests

Which brings me to Prana's press release (http://biz.yahoo.com/iw/080226/0366713.html), which includes this conclusion, "PBT2, in this study of early disease progression, had no effect on ADAS-cog, a test of cognition not designed to measure Executive Function changes." BUT, as everyone who cares must know by now, it also says, "Encouraging signs of cognitive improvement, as measured by the Neuropsychological Test Battery (NTB), were also observed. Statistically significant improvement was evident in two of the four Executive Function NTB tests: the Category Fluency Test (p=0.028) and the Trail Making Test part B (p=0.005), both after 12 weeks of treatment at the 250mg dose compared to placebo."

So the PhIIa showed no improvement with ADAS-cog, but showed statistically significant improvement with NTB. I personally had to go look up the difference and found this (http://archneur.ama-assn.org/cgi/content/abstract/64/9/1323). "Conclusions The NTB exhibits excellent psychometric properties and seems to be a reliable and sensitive measure of cognitive change in patients with mild to moderate Alzheimer disease. The psychometric properties of the NTB suggest that it may have particular utility in evaluating drug efficacy in clinical trials in which patients with mild Alzheimer disease are included." Is it safe for me to assume that Neurology hasn't been infiltrated by TEVA hacks yet? :)

Does anyone else think NTB is valid? This company (one I admire :) seems to think so since they included NTB as a key end-point.

"The key end-points include: ADAS-Cog (assesses cognition), Neuropsychological Test Battery (NTB) and DAD score (measures quality of life)." (http://www.elan.com/news/full.asp?ID=1004531"


So what do we have?
1) Cochrane concluded a LONG time ago that PBT1 didn't work
2) Prana says this trial, not considered by Cochrane, showed no benefit of PBT2 when using ADAS-cog
3) Prana says this trial, not considered by Cochrane, DID show benefit of PBT2 when using NTB
4) Neurology says NTB is valid
5) Elan included NTB as a key end-point