Classification bias and impact of COVID-19 vaccination on all-cause mortality: the case of the Italian region Emilia-Romagna

Abstract

Real-world studies on vaccine effectiveness may suffer from several biases, typically distorting their results. A previous article on the population of an Italian province, correcting the "immortal time bias", showed worse results for the all-cause death of the vaccinated compared to the unvaccinated. This article highlights the "case counting window bias", that considers the vaccine recipients "unvaccinated" usually for 14 days, a time interval reputed necessary to express the vaccine immune response. We aim to document this bias in an Italian region, calculating the daily death incidence for each age class of vaccinated and unvaccinated and checking their all-cause mortality difference within the considered time window. Indeed, in this window the two groups showed huge differences in all-cause deaths, that cannot be attributed only to COVID-19 deaths (in the absence of reasons to expect significant vaccine effects on non-COVID-19 deaths). In conclusion, analyzing the data of an Italian Region, we found evidence of the 'case counting window bias', which artificially increases the 'unvaccinated' mortality and reduces the mortality in the vaccinated.