Selection Effects and COVID-19 Mortality Risk after Pfizer vs. Moderna Vaccination: Evidence from Linked Mortality and Vaccination Records

Abstract

Prior research generally finds that the Pfizer-BioNTech (BNT162b2) and Moderna (mRNA1273) COVID-19 vaccines provide similar protection against mortality, sometimes with a Moderna advantage due to slower waning. However, most comparisons do not address selection effects for those who are vaccinated and with which vaccine. We report evidence on large selection effects, and use a novel method to control for these effects. Instead of directly studying COVID-19 mortality, we study the COVID-19 excess mortality percentage (CEMP), defined as the COVID-19 deaths divided by non-COVID-19 natural deaths for the same population, converted to a percentage. The CEMP measure uses non-COVID-19 natural deaths to proxy for population health and control for selection effects. We report the relative mortality risk (RMR) for each vaccine relative to the unvaccinated population and to the other vaccine, using linked mortality and vaccination records for all adults in Milwaukee County, Wisconsin, from 1 April 2021 through 30 June 2022. For two-dose vaccinees aged 60+, RMRs for Pfizer vaccinees were consistently over twice those for Moderna, and averaged 248% of Moderna (95% CI = 175%,353%). In the Omicron period, Pfizer RMR was 57% versus 23% for Moderna. Both vaccines demonstrated waning of two-dose effectiveness over time, especially for ages 60+. For booster recipients, the Pfizer–Moderna gap is much smaller and statistically insignificant. A possible explanation for the Moderna advantage for older persons is the higher Moderna dose of 100 μg, versus 30 μg for Pfizer. Younger persons (aged 18–59) were well-protected against death by two doses of either vaccine, and highly protected by three doses (no deaths among over 100,000 vaccinees). These results support the importance of a booster dose for ages 60+, especially for Pfizer recipients. They suggest, but do not prove, that a larger vaccine dose may be appropriate for older persons than for younger persons.

Vaccinating against a Novel Pathogen: A Critical Review of COVID-19 Vaccine Effectiveness Evidence

“…However, differences between vaccines seen for primary vaccination may diminish after a booster. A U.S. study compares BNT162b2 and mRNA1273 and finds a BNT162b2/mRNA1273 odds ratio for death of 2.40 (z = 4.26) after waning for primary vaccination, but only 1.25 (insignificant, z = 0.50) after a booster dose [ 24 ]. Whether larger differences between vaccines after a booster dose would emerge over a longer time period is unknown.…”

“…For example, test-negative designs, which were the principal designs used in the observational studies that we reviewed, suffer from healthy vaccinee bias [ 82 , 83 ], in which healthier people, who face lower background risk of severe disease, are more likely to become vaccinated or boosted. This bias can be substantial, especially for VE against hospitalization or death [ 13 , 24 ]. The implication is that many studies overestimate VE (underestimate RR).…”

“…The included studies are consistent with gender having only modest effects on VE and age having only modest effects on VE against the infection endpoints (e.g., [ 22 , 23 ]). More recent studies provide evidence that waning VE against hospitalization and death occurred principally for ages 60+ [ 11 , 24 , 25 ]. Most studies report VE for the whole adult population, without controlling for prior infection.…”

“… Table sources: Panel A [ 63 , 64 , 65 , 66 , 67 , 68 , 69 , 70 ]. Panel B [ 24 , 66 , 69 , 71 , 72 ]. Panel A reports point estimates for risk reduction for vaccinated plus boosted people versus people vaccinated without booster, to nearest percent, for endpoints with available data.…”

COVID-19 Vaccination Effectiveness in the General Population of an Italian Province: Two Years of Follow-Up

“…The comparison of different vaccines’ effectiveness provided mixed results, making it complex to draw clear conclusions. While several studies reported a higher effectiveness of mRNA-1273 compared to BNT162b2 against infection and severe COVID-19 outcomes both as a primary and booster vaccination [ 51 , 52 , 53 , 54 , 55 , 56 ], in the present cohort we observed a significantly higher protection of mRNA-1273 only against SARS-CoV-2 infections or after a single dose, and the subjects who received three or more doses of mRNA-1273 showed a higher likelihood of death, as compared to the recipients of BNT162b2 only. In contrast, we observed a greater protection conferred by heterologous booster (mixed) vaccination compared with a homologous booster with BNT162b2 only, in line with the findings of immunogenicity [ 57 ] and other population-based [ 20 , 58 , 59 ] studies.…”