Problems with Pfizer's vax trials?

Just saw this:

Some significant problems at a Pfizer subcontractor site that was involved with the COVID-19 vaccine trials. The broad deployment and success of the Pfizer vaccine among the general public suggests that these problems didn’t dramatically affect the results of the trial, but I’m still surprised this is not bigger news.

I’d say the major reason is the one you mentioned – there have been millions of doses given to people, with no significant problems reported.

But also, it’s a year-old report from a fired employee, and some of the ‘integrity issues’ seem more like minor issues than something that will affect the trials:

  • used needles discarded in a bio-hazard bag rather than a sharps container – maybe dangerous for the janitor there, but unlikely to sway the results of the trial.
  • data inquiry responses not replied to for over 3 days, when the objectives say they should get a reply in 24 hours – again, seems unlikely to sway the results of the trial.
  • drug information included in patient records folders, so that vaccinators could have looked in the patient records to see if patient received vaccine or a placebo – ‘could have’, no allegation that it actually happened.

So while these items should be corrected or future studies, there isn’t evidence that they affected the results of the study. And the fact that millions of people (including me) have received this vaccine without bad results means that whining about poor procedures is only of interest to people who run studies. No reason it should be bigger news to the rest of us.

No, that’s not what the problem was.

Early and inadvertent unblinding may have occurred on a far wider scale. According to the trial’s design, unblinded staff were responsible for preparing and administering the study drug [these were the “vaccinators”] (Pfizer’s vaccine or a placebo). This was to be done to preserve the blinding of trial participants and all other site staff, including the principal investigator. However, at Ventavia, Jackson told The BMJ that drug assignment confirmation printouts were being left in participants’ charts, accessible to blinded personnel.

My bold. The text after the bold makes it clear that staff who should have been blinded to the treatment were potentially unblinded. In a clinical trial this is a big issue. Trials are set up to be cheat-proof - you take that away and you have a problem.

Mitigating points:

  • This issue only affected a couple of percent of the total study population.
  • There’s scope for a post-hoc analysis with these subjects excluded

Regarding the second point, unscheduled post-hoc analyses are not good, but in this case I think it may prove to be something useful to fall back on. Yes, the real-world experience of the vaccine has established its value, but I’d still like to have this option to fall back on (if needed). It wasn’t the real world experience that was used to licence the vaccine.