Does the analysis of Schmeling et al. suggest a batch-dependent safety signal for the BNT162b2 mRNA COVID-19 vaccine?

The short answer is no.

The conclusion of Schmeling et al. [1]

“In conclusion, the results suggest the existence of a batch-dependent safety signal for the BNT162b2 vaccine, …”

does not follow from their analysis.

Let me explain why.

They assign reported adverse effects (AEs) to vaccine batches, and make and interpret graphs of number of batch-assigned AEs versus number of vaccine doses per batch, delivered in Denmark between 27 December 2020 and 11 January 2022. Each point on such a graph (for a given AE type or severity) is for one of the 52 batches used in Denmark in this period.

They show the resulting graph only for all-AEs, irrespective of type or severity (their Figure 1). They admit “Compared to the rates of all SAEs, serious SAEs and SAE-related deaths per 1000 doses were much less frequent and numbers of these SAEs per 1000 doses displayed considerably greater variability between batches, with lesser separation between the three trendlines,” but they do not show those graphs.

That is the full extent of their analysis. In doing this, they exclude all the information about the subjects experiencing the AEs, such as age and sex; and all the information about the injections, such as the date of the injection. They are blind to virtually everything except what is needed to make the batch assignment for each AE.

This blindness by design is acknowledged as: “in the present study, … , demographics of SAE cases, … , were not examined.”

This is a recipe for studying artifacts arising from invisible factors.

link