Cases of acquired hemophilia A following COVID-19 vaccines: Cogent safety signal or possible reporting/detection bias? Preliminary evidence from Tuscany, Italy

Purpose: Several case reports of acquired hemophilia A (AHA) following COVID-19 vaccines were recently published. A possible increased incidence of AHA during the COVID-19 vaccination campaign was also suggested. We aimed at generating evidence for the preliminary assessment of the association between AHA and COVID-19 vaccination through an ecological study in one Italian region, Tuscany. Methods: An ecological study was performed using the population-based administrative data source of Tuscany. Per each year between 2017 and 2021, we included patients aged 5+ and active into the database as of January 1. Temporal patterns of annual incidence of possible AHA cases and AHA-tested patients were respectively observed. The rates of possible AHA cases per AHA-tested patients were calculated in 2021 and 2017-2019, respectively (calendar year 2020 was excluded because non-representative of the pre-pandemic era). Age-sex standardization was applied. Poisson's 95% confidence intervals (95% CI) were estimated. Statically significant differences were defined as absence of 95% CI overlap. Results: In 2021, standardized incidence of both possible AHA cases (5.6/million subjects/year; 95% CI = 3.4-8.7) and AHA-tested patients (60.7/1000 subjects/year; 95% CI = 60.4-60.9) showed the lowest point estimates, though only the latter was statistically different compared to previous calendar years. The standardized rate of possible AHA cases per AHA-tested patients was 9.2/100000 (95% CI = 5.6-14.3) in 2021 and 12.5/100000 (95% CI = 8.2-18.1) during 2017-2019. Conclusions: These preliminary findings do not support the hypothesis of an increased incidence of AHA cases during the COVID-19 vaccination campaign. However, in 2021, the still ongoing healthcare access restrictions might have contributed to the low incidence of AHA and laboratory tests observed. Therefore, large-scale multi-database studies are warranted.