(350) Pilot study on the application of rapid cycle analysis method for post-marketing active vaccine safety surveillance using COVID-19 vaccines in Japan: the VENUS Study

Background: COVID-19 vaccines have been administered in Japan as part of the mass vaccination program since their approval in February 2021. The Japanese surveillance system for vaccine safety mainly relies on passive surveillance, which is based on reports from medical facilities or manufacturers. However, near real-time active safety surveillance is performed through the Vaccine Safety Datalink in the United States.

Objectives: To evaluate the possibility of applying active surveillance method in Japan and identify relevant challenges.

Methods: We used data from five municipalities in the Vaccine Effectiveness, Networking, and Universal Safety (VENUS) study. This included register data, claims data, COVID-19 vaccination records, and the incidence of COVID-19 cases between January 2015 and February 2023. The surveillance duration was from February 2021 to February 2023 and was updated monthly. We assessed all doses of mRNA COVID-19 vaccines (Pfizer/BioNTech) and used a rapid cycle analysis (RCA) method to analyze 24 pre-specified outcomes of interest using historical comparators. We defined outcomes (e.g., acute myocardial infarction, pulmonary embolism, stroke) with risk intervals using ICD-10 codes in inpatient settings. The historical comparators were the 2019 VENUS data, which determined the background rate of the outcomes. We used the conditional maximized sequential probability ratio test to estimate the relative risk (RR) for each outcome. When the test statistics of the log-likelihood ratio exceeded the critical value, we considered that a signal was detected. The analyses continued until a signal was detected or until the end of the surveillance period. The analyses were conducted using R Sequential package.

Results: The numbers of Pfizer/BioNTech vaccine doses (original monovalent) delivered for the first to fifth vaccinations were 382,961 (mean age, 71.1), 382,308 (71.2), 251,873 (73.7), 218,390 (76.9), and 33 (77.9), respectively. The historical control group was 648,584 (mean age, 65.6). We could not assess rare outcomes, due to the small sample size. Throughout the surveillance duration, we did not identify any signals for the other outcomes (myocardial infarction [N = 139; RR = 0.85]; pulmonary embolism [N = 61; RR = 0.88]; stroke, ischemic [N = 752; RR = 0.90]; stroke, hemorrhagic [N = 194; RR = 0.83]; and venous thromboembolism [N = 103; RR = 0.83]).

Conclusions: We applied RCA method to Japanese database from the VENUS study. We evaluated the frequency outcomes, mainly in older adults—however, an approach to assess rare outcomes in children and younger adults should also be developed. Further consideration is warranted to develop a real-time monitoring system in Japan.