Background: Vaccination is an important tool in the fight against pandemics. However, the associated adverse events (AEs) may negatively impact the public perception of vaccines, therefore leading to decreased vaccination willingness. Interestingly, pharmacovigilance data of the three COVID-19 vaccines with a two-dose schedule approved in the EU (AstraZeneca, BioNTech and Moderna) already revealed country-specific differences in their safety profile early on (as of February 2021), at a time when the accumulated occurrence of specific AEs was not yet known. In the safety outcome assessment presented here, we aimed to establish whether these country-specific differences in pharmacovigilance data could be explained by differences in the frequency of AEs as reported in the respective approval studies of each vaccine.
Methods: A systematic search was performed to identify all publications regarding the randomized controlled trials (RCTs) of two-dose vaccines approved in the EU (AstraZeneca, BioNTech and Moderna), including regulatory reports and journal articles. All obtained safety data was manually entered into an SQL database. In order to enable the comparability among the data, the solicited AEs for all vaccines (i.e. those AEs actively sought after vaccination) were investigated. The data was standardized to promote comparability and overcome data heterogeneity and complexity.
Findings: Twelve documents regarding the RCTs for the three COVID-19 vaccines with a two-dose schedule approved in the EU (AstraZeneca, BioNTech and Moderna) were included in the safety outcome analysis. The entire safety data compiled in the SQL database amounted to 66 different study arms. The data structure revealed 13 different age thresholds or ranges and three different data sets regarding doses (first dose vs. second dose vs. all doses). After standardization and identification of subgroups, the analyses demonstrated that the highest rates of AEs occur after the first dose with the AstraZeneca vaccine, whereas with Moderna and BioNTech most AEs occur after the second dose. Astonishingly, the absolute frequencies of each AE after the first AstraZeneca dose correspond to those of the second dose of the mRNA vaccines (BioNTech and Moderna). Reversely, the absolute frequencies of the same AEs after the second AstraZeneca dose correspond to those of the first dose with the mRNA vaccines. The most common AEs with any vaccine were fatigue, headache and myalgia. Moreover, middle-aged subjects (18 to 55 years) had more side effects than older individuals (> 55 years), an observation that persisted among vaccines.
Interpretation: This is the first indirect comparison of these vaccines that uses all available RCT data. The absolute frequency of each AE is similar between the first AstraZeneca dose and the second dose of BioNTech or Moderna; their occurrence was thus independent of platform (vector or mRNA) or the vaccine itself. This assessment demonstrates that the varying frequencies of AEs reported in early pharmacovigilance data for the vaccines in distinct countries, at a time when the accumulated occurrence of specific AEs with certain vaccines was not yet known, cannot be explained by different frequencies being reported in the respective RCTs.
Conclusion: The approach presented here could help to objectify future discussions on vaccine preferences. Therefore, it may serve as basis for future public awareness campaigns and may also allow the comparison of vaccine performance in different subgroups (e.g. virus variants, high-risk patients). This approach may also be applied to a broad range of other challenges across the R&D process and various disease categories.
Copy-and-patch compilation: a fast compilation algorithm for high-level languages and bytecode
