Michels also raised issues regarding total death reports and their timing.

Since the death total from both study groups, 38, appeared “surprisingly low” to study authors — particularly during a pandemic — they undertook their own analysis based on population mortality expectations at the time.

Assuming that age-adjusted death rates for the study subjects were similar to those of the general population, they estimated that 222 subjects should have died from July 27, 2020, to March 13, 2021. The reported number, 38, is just 18% of the expected number.

Michels explained this by the large number, 4.2% of “discontinued subjects.” The most concerning of these were subjects “lost to follow-up,” which means missing scheduled visits or other required activities.

Pfizer-BioNTech tried to reach these subjects via phone, certified mail or through their emergency contact but despite their efforts could not account for 395 subjects who had dropped out.

The authors wrote:

“These are not insignificant numbers and could easily account for the low number of deaths reported in this safety period of the trial. Given the importance of knowing the status of each trial subject, there should have been greater effort to locate these individuals.

“Additionally, Pfizer/BioNTech was responsible for oversight of the trial sites. Sites with excessive numbers of lost to follow-up should have been evaluated for performance.”

Michels was also concerned over how certain trial centers had many dropped-out subjects while others had none or just a few.

Ninety-six of 153 trial sites (63%) reported 0 or 1 subjects lost to follow-up and 34 (22%) reported 2-5 dropouts. But four sites reported more than 20 subjects lost to follow-up, amounting to about 5% of all trial subjects.

Since the vaccine makers were responsible for trial site oversight, the authors wrote, “Sites with excessive numbers of lost to follow-up should have been evaluated for performance.”

Finally, based on the data, it appears Pfizer-BioNTech was in no hurry to enter death reports before the EUA submission deadline, particularly for the BNT162b2 group.

Of the 38 reported deaths only one case was added on the day the subject died. Delays of 20+ and 30+ days were common.

One death took 72 days to find its way into the database, and all were entered as occurring on the reporting day, not on the actual date of death.

Of the eight subjects in the vaccine group that should have been reported by Dec. 10, 2020, the EUA application cutoff, the average reporting delay was 17.5 days for subjects in the vaccine group, but just 5.9 days for deaths among subjects in the placebo group.