The University of Oslo, as part of the European VAC4EU network (https://vac4eu.org/), contributes to generating real-world evidence about the safety of mRNA-based COVID19 vaccines. This is possible by capitalizing on our national health registries data covering the entire population and capturing a vast array of information, e.g., COVID19 vaccination, positive PCR tests for COVID19, hospitalizations, deaths.
VAC4EU (Vaccine monitoring Collaboration for Europe)brings together many institutions across Europe, each acting as “data access provider (DAP)” of the corresponding country or region. The idea is that each DAP transforms its source data according to a common structure, so that analytical scripts for the needed analyses are developed centrally, then distributed to and executed by each DAPs. Summary results from all DAPs are pooled into unique effect estimates. This is an exciting and unique collaborative approach with high impact for public health. Indeed, VAC4EU generates real-world safety data about the COVID19 vaccines in a timely manner and across all Europe. The huge sample size achieved by combining all DAPs in Europe allows identification of small risks and rare adverse events. This would not have been possible without the IT-support provided by USIT and, especially the TASK team. Indeed, lots of IT challenges were met as soon as the analytical phase of the project got started.
One of the challenges was that the development of the code did not take into account differences in IT-infrastructure, including system specifications. For instance, the code was not pressure tested on cloud-servers such as TSD, where I/O (input/output) is vastly different from non-cloud infrastructures. Concretely, parallelizing tasks related to I/O of a database using SQLite (a database engine) resulted in database locking issues. Furthermore, due to the way the code is developed, it is not possible to capture the differences and intricacies of each DAPs datasets, which can lead to runtime errors or even information being left out in the result. In both these cases, it means that the DAPs to the best of their abilities have to come up with suggestions to “optimize” the code, which USIT have been involved in. However, not all DAPs have access to expertise in the programming language R, meaning that they are entirely dependent on the code developers to deal with such issues. At USIT and in the TASK team, we are lucky to have such expertise.
