Many Zeros
Several topics in the air here. All appropriate, to be considered by all.
There are continuing questions (by some) that the very positive symptom stabilizations or improvements during the recent clinical study can not be used because of the small trial participant numbers (n=25). It is stated that proper clinical trials require much larger populations (say n=300), in both a placebo or control arm, and a parallel experimental arm. Half of the participants, say, 150, in the control arm take a daily pill visually and palatably identical to the Anavex-containing pills the 150 in the experimental arm are taking. Neither the patients nor the doctors or nurses know who’s getting what; only a locked computer recording pill lot numbers. At the end of the trial, the blinded computer data are unlocked and matched against symptoms assessed during the trial. Appropriate statistical procedures are used to crunch and connect the data, which are then plotted (as in the CTAD presentation).
As a biologist, I know this well. The study must be double-blinded (neither patients or doctors know) so only authentic drug-caused results appear. Minimization of placebo effects.
If I were a reviewer of an Anavex clinical trial on humans, to be published in a peer-reviewed journal, all of the above (and more) would be required. Journals don’t like data appearing that were by chance or poor trial design showing unrelated or complicating drug-related phenomena. Bad numbers reveal bad science.
But, when I made my several AVXL purchases (last one just after parsing the CTAD data plots), I did not act upon my knowledge of strict clinical conduct rules, used to obviate erroneous by-chance or bad-influences results. I was (and am) confident in my equity purchases for the following arithmetical reasons.
Although the final trial cohort number was very small (n=25) — normally too small to offer a high-probability rendering of data accuracy — I was convinced otherwise.
Simply stated, all of the participants had one of two things transpire during the clinical period: they either remained symptomatically static or stable, or, actually had symptoms improve. Not a single instance of normal symptomatic decline. How could that be?
Only two explanations. Perhaps by remarkable happenstance, against all odds, 25 trial participants (the entire cohort in contention) just happened to be Alzheimer’s patients who so rarely show no decline over the many months of the trial.
Or, the cohort included typical Alzheimer’s patients, for which symptoms would normally get worse, but they didn’t because of Anavex 2-73 efficacy.
Which was it? An exceptionally atypical sampling of Alzheimer’s patients, all of which either held their symptom progression at bay or even suppressed them, regardless of any putative inefficacy of A2-73; or, do the simple sampling stochastics (probabilities) make the former explanation improbable?
It’s the latter. For those questioning the validity of the trial’s results, attributing them to pure happenstance, where an atypical number of Alzheimer’s patients got included who had rare abilities to innately stop or reverse symptomatic progress, let’s do simplest statistical probabilities for such an outcome.
First, we have to know the fraction of Alzheimer’s patients who (at the mid to mild stages of the disease, as per trial participants) can innately remain symptomatically stable for 52 weeks (or whatever the actual trial period in question was). Frankly, I’ve never read or learned of any spontaneous remission of Alzheimer’s symptoms. Once they appear, they continue to progress undiminished and in ever greater magnitudes.
But, I’ll presume that one percent of Alzheimer’s patients DO have this capability; that’s one out of a hundred, from time to time, who experience year-long remissions in symptom severity increase. For a year, they stay stable, as show on the CTAD plots.
What, then, are the chances of including everyone (n=25) in the trial cohort with such a trait? If the cohort had but one person, the chances of such results would be four percent, one out of 25. But we had 25. What are the chances of stochastic (pure chance selection) inclusion of 25 such symptom-controlling people? Multiply each individual’s chances with all the others.
The chance of this happening — if one out of a hundred Alzheimer’s patients have the rare ability to spontaneously yield flat or ascending favorable symptom data points — is 1E-50. To get 25 of these patients, it’s the number 1 with 50 xeros after it.
But, maybe in the unique population of Alzheimer’s patients the clinical cohort was drawn from, one out of ten had the unique symptom-controlling traits. Much better chance of Anavex inefficacy, then. How much? Just 1E-25. The number one followed by 25 zeros.
I forked over my few hundred dollars to my online brokerage and made my AVXL purchase with the greatest confidence that Anavex 2-73 actually does suppress or reverse symptomatic progression in Alzheimer’s. The many zeros tell it all.
