Vaccine safety research relies on being able to identify the exposure to a vaccine. To detect very rare events that may be associated with vaccines we need very large populations because:
- To identify a two-fold risk of a vaccine-associated event that occurs in once in 100,000 people, we need to compare 2.35 million people who have received the vaccine with 2.35 million people who have not.1
- There is wide variability in the types and manufactures of vaccines used in different countries.
- We need to explore the risk in subgroups such as older people and children.
Collections of routinely collected health data already exist in many countries. These collections are diverse, containing useful information such as hospitalisations, pharmaceuticals including vaccines, pregnancy and birth over time. Collaborations between organisations holding different data within a country, for example the Vaccine Safety Datalink (VSD) in the U.S., or across countries within a continent such as the VAccine monitoring Collaboration for EUrope (VAC4EU), are examples of the successful use of big data for meaningful advances in the knowledge of vaccine safety and effectiveness as well as vaccination coverage and cost effectiveness.
How big data can help assess vaccine safety
Randomised controlled trials (RCT) are usually used to assess vaccine safety and efficacy. However compared to RCTs, big data can provide a picture of vaccine safety and effectiveness across large and diverse groups of people and over time. This is because some groups of people are usually left out of prelicensure trials. For example, people with particular health conditions, ethnic minority groups, and pregnant women. Additionally, big data can capture information over longer periods of time than RCTs. These features of big data mean identifying rare events and longer term effects is possible.
1. Strom, BL. Appendix A: Sample size tables. In: Strom BL, Kimmel SE, Hennessy S, editors. Pharmacoepidemiology. 6th ed. West Sussex: John Wiley & Sons; 2019. pp. 1123-40.