Abstract
Vaccines help to protect individuals and populations against harmful diseases. However, vaccines, as any other medicines, can trigger side effects or adverse events that range from mild to severe symptoms or diseases. Because severe adverse events are rare, they may go undetected during the clinical development phases which are conducted
... read more
on a limited number of participants. Therefore, assessing risks associated to vaccines after their approval for use is necessary to provide reassurance on benefit-risk profile of vaccines and to maintain public confidence in vaccine programmes.
The monitoring of benefit-risk profiles of vaccines is a process that starts at early clinical development phases and includes passive, enhanced and active surveillance activities. Over the last years, the use of real-world data (RWD) and existing healthcare data sources has grown to assess vaccine safety post-licensure. RWD are, for instance, data electronically collected by physicians during routine clinical practice or disease diagnosis or procedures at hospital level or pharmacy claims. With collaboration initiatives at global level, the implementation of multi-data sources studies using RWD became a standard in vaccine safety assessment, which necessitates the development of methods such as common analytical approaches to overcome the observed heterogeneity across data sources.
In the thesis, in first instance, we emphasize the increasing use of multi-data sources studies and the use of distributed data networks to generate data on vaccine safety. The main advantage of multi-data sources studies is that they allow to gain statistical power to study rare outcomes by increasing study population size, and therefore maximizing the likelihood to detect and assess rare adverse events that may occur following vaccination. However, the applicability of multi-data sources studies has limitations related to the observed heterogeneity across data sources. Because a variety of types of data source exists which includes inpatient and/or outpatient medical diagnoses from hospitalization data sources, medical records from general practitioners or family pediatricians’ data sources and record-linkage data sources that link hospitalization data and general practitioners’ data or link data from registries, it is of importance to consider the data provenance and the setting where diseases are typically diagnosed for a correct data interpretation.
Second, we discuss and provide methodological considerations in vaccine safety signal evaluation studies, which are based on the experience from the bivalent HPV vaccine post-authorization safety studies. We highlight that harmonized clinical case definition can minimize bias linked to heterogeneity across studies, misclassification of exposure can be overcome by using active comparators and tailored statistical methods should preferably be used when assessing rare adverse events.
Third, we discuss future perspective for the study of vaccines in the European context with the development of EU DARWIN and the conduct of fit-for-purpose data sources exercises.
To conclude, this thesis underlines the methodological improvements in vaccine safety assessment which should be maintained globally to ensure reproducibility and comparability of study results.
show less
