Near Real Time Safety Surveillance of COVID-19 Vaccines in the UK

Background: The UK was the first country to introduce a COVID-19 vaccine. A proactive pharmacovigilance system, supported by near real time data, was vital to further reassure on safety and support public confidence but also to rapidly detect true rare safety concerns, informing decision-making.

Objectives: To continuously evaluate the rate of 19 pre-specified adverse events of special interest (AESI) with COVID-19 vaccines through comparative continuous safety surveillance of electronic healthcare records (EHR).

Methods: We conducted repeated weekly Maximised Sequential Probability Ratio Tests within the primary care data of the Clinical Practice Research Datalink (CPRD) to assess the rate of AESI in a risk window post-vaccination compared to a 2015/19 pre-pandemic period. A signal was triggered if the relative risk crossed a threshold based upon the expected number of doses, duration of monitoring, and the pre-pandemic rate. Analyses were carried out by vaccine brand, dose, and age group with sensitivity analyses allowing for changes in background rates through the pandemic and adjustments to follow up made to account for delays in recording of AESI in the primary care record.

Results: Within two months, 2.4m first doses of a COVID-19 vaccination had been captured in the CPRD. At four months, signals had been raised in group analyses across all age groups for 11 AESI-vaccine brand combinations. Of these, 4, Bell's palsy with both AstraZeneca and Pfizer vaccines, and Guillain-Barré Syndrome and immune thrombocytopenia with AstraZeneca, after evaluation alongside evidence from national and international spontaneous reports and epidemiological studies, have been included within UK vaccine product information as adverse reactions with further signals emerging as the campaign continued.

Conclusions: This is the first real time use of these methods exclusively within EHR in the UK and results have contributed to the body of evidence on the safety of COVID-19 vaccines, supporting more robust earlier decision-making than that based mainly on spontaneous reports. We demonstrated the value of the size, data quality, and daily data flow of CPRD in supporting near real time surveillance and highlighted important operational considerations for future implementations with a flexible approach and careful interpretation needed.