Kind of like the "errors" in hcq tests where they overdosed and declared it dangerous, and started too late and declared it ineffective. It's easy to perpetrate frauds when your reviewers are in on the scam. Or maybe they're just 8ncompetent.
More to the point, if Pfizer had been running the wrongly designed HCQ trials, they could have still massaged or hidden all the raw data and declared success...
Brian, on page 80 there is a line item: "Had other important protocol deviations on or prior to 7 days after Dose 2". This was the line item that caused the most consternation because of the large difference between the arms (240 treatment 60 placebo). I didn't see a breakdown of the causes for this... did you?
Just the categories listed in p 79, which are quoted in the update to the original post: 203 improper administrations, of which 110 temperature errors and 76 dilution errors (so, 17 undefined). That leaves 37 non-administration "deviations" and the question of why the count dropped from December (302) to March.
Right, but those breakdowns pertain to the exclusions listed directly prior to the breakdown.. I did not see any breakdown for the exclusion line on page 80.
It's all continuous, one big Table 28. So there's continual redefinitions of the denominator over time and no idea given for which denominators to subtract excluded participants from. They could have hidden withdrawn participants anywhere, really.
Not to get too conspiratorial..... but the real problem is since I really don't trust them very much I wonder how much we can rely on any of these exclusions. How do they screw up the dosage and expiration - these should be performed at testing sites that are very careful and diligent. If there were reasons why people in the treatment arm didn't go to dose 2 (e.g. serious AE), they might have balanced this by somehow deliberately causing a matching protocol violation in the placebo arm, by failing to dose them on the proper schedule for instance, or just declaring it "uncertain/unreliable". I get the overall impression that these guys are experts at gaming trials, they've been doing it for years.
I'm not really familiar with the standard practice for handling exclusions. Given that the placebo arms are not identical in number, I assume that they don't just equalize them as a matter of course by dropping placebo participants. My model of the concept is that legitimate exceptions should be rare and could happen in both arms for similar reasons and at similar rates: protocol violations shouldn't differ substantially between the two arms, or there is something wrong that needs to be examined, that might also violate the blinding.
In terms of table 28 and the last exclusion listing that is out of whack, it's still weird to me that they enumerated the exclusions for the first three exclusion sets, but not this one. I'm not sure I understand your response that they could be hidden anywhere? The question is why were they excluded. IIRC the Pfizer protocol required that participants do not seek medical attention except through the trial personnel, which could give them some leeway to exclude people that happen to have serious events, and go to the emergency room.
We know Pfizer has done this kind of exclusion shenanigan before,
Kind of like the "errors" in hcq tests where they overdosed and declared it dangerous, and started too late and declared it ineffective. It's easy to perpetrate frauds when your reviewers are in on the scam. Or maybe they're just 8ncompetent.
More to the point, if Pfizer had been running the wrongly designed HCQ trials, they could have still massaged or hidden all the raw data and declared success...
Brian, on page 80 there is a line item: "Had other important protocol deviations on or prior to 7 days after Dose 2". This was the line item that caused the most consternation because of the large difference between the arms (240 treatment 60 placebo). I didn't see a breakdown of the causes for this... did you?
Just the categories listed in p 79, which are quoted in the update to the original post: 203 improper administrations, of which 110 temperature errors and 76 dilution errors (so, 17 undefined). That leaves 37 non-administration "deviations" and the question of why the count dropped from December (302) to March.
Right, but those breakdowns pertain to the exclusions listed directly prior to the breakdown.. I did not see any breakdown for the exclusion line on page 80.
It's all continuous, one big Table 28. So there's continual redefinitions of the denominator over time and no idea given for which denominators to subtract excluded participants from. They could have hidden withdrawn participants anywhere, really.
Not to get too conspiratorial..... but the real problem is since I really don't trust them very much I wonder how much we can rely on any of these exclusions. How do they screw up the dosage and expiration - these should be performed at testing sites that are very careful and diligent. If there were reasons why people in the treatment arm didn't go to dose 2 (e.g. serious AE), they might have balanced this by somehow deliberately causing a matching protocol violation in the placebo arm, by failing to dose them on the proper schedule for instance, or just declaring it "uncertain/unreliable". I get the overall impression that these guys are experts at gaming trials, they've been doing it for years.
Right, I assume placebo exclusions were "paired" as a matter of course, hence the similar numbers across the board except for this one item.
I'm not really familiar with the standard practice for handling exclusions. Given that the placebo arms are not identical in number, I assume that they don't just equalize them as a matter of course by dropping placebo participants. My model of the concept is that legitimate exceptions should be rare and could happen in both arms for similar reasons and at similar rates: protocol violations shouldn't differ substantially between the two arms, or there is something wrong that needs to be examined, that might also violate the blinding.
In terms of table 28 and the last exclusion listing that is out of whack, it's still weird to me that they enumerated the exclusions for the first three exclusion sets, but not this one. I'm not sure I understand your response that they could be hidden anywhere? The question is why were they excluded. IIRC the Pfizer protocol required that participants do not seek medical attention except through the trial personnel, which could give them some leeway to exclude people that happen to have serious events, and go to the emergency room.
We know Pfizer has done this kind of exclusion shenanigan before,
https://aaronsiri.substack.com/p/fda-buries-data-on-seriously-injured
What is a placebo these days. It's no longer just a saline solution. In fact it can even be another vaccine.
https://theconversation.com/coronavirus-vaccine-why-its-important-to-know-whats-in-the-placebo-146365