@Onthebeach the product date code you suggested checks with 10 random reports I pulled as well as your examples. I tried to download a copy of VARES data but it is not currently available. So I made a list of decoded lot dates. This is something I will modify if I can get actual production dates for the lots, so consider it for what it is (incomplete list of fatalities sorted by a part of the lot code).

I discarded unknown codes and consolidated the similar appearing codes with hyphens. I converted A20 to 1 on the graph ... M20 would be 13 representing December 2020, A21 = 14 = Jan 2021 ...

The first public jabs were in late December 2020 (13 on the timeline) and 26 is where we are now. If the Lot code conversion is correct, there have been 0 identified lots in the last 6+ months. The total unknown lots are less than 40.