Association of cerebral venous thrombosis with recent COVID-19 vaccination: case-crossover study using ascertainment through neuroimaging in Scotland

Abstract

Objectives To investigate the association of primary acute cerebral venous thrombosis (CVT) with COVID-19 vaccination through complete ascertainment of all diagnosed CVT in the population of Scotland. Design Case-crossover study comparing recent (1-14 days after vaccination) with less recent exposure to vaccination among cases of CVT. Setting National data for Scotland from 1 December 2020, with diagnosed CVT case ascertainment through neuroimaging studies up to 17 May 2021 and diagnostic coding of hospital discharges up to 28 April 2021 and with linkage to vaccination records. Main outcome measure Primary acute cerebral venous thrombosis Results Of 50 primary acute CVT cases, 29 were ascertained only from neuroimaging studies, 2 were ascertained only from hospital discharges, and 19 were ascertained from both sources. Of these 50 cases, 14 had received the Astra-Zeneca ChAdOx1 vaccine and 3 the Pfizer BNT162b2 vaccine. The incidence of CVT per million doses in the first 14 days after vaccination was 2.2 (95% credible interval 0.9 to 4.1) for ChAdOx1 and 1 (95% credible interval 0.1 to 2.9) for BNT162b2. The rate ratio for CVT associated with exposure to ChAdOx1 in the first 14 days compared with exposure 15-84 days after vaccination was 3.2 (95% credible interval 1.1 to 9.5). The 95% credible interval for the rate ratio associated with recent versus less recent exposure to BNT162b2 (0.6 to 95.8) was too wide for useful inference. Conclusions These findings support a causal association between CVT and the AstraZeneca vaccine. The absolute risk of post-vaccination CVT in this population-wide study in Scotland was lower than has been reported for populations in Scandinavia and Germany; the explanation for this is not clear. What is already known on this topic The risk of cerebral venous thrombosis (CVT) within 28 days of receiving the AstraZeneca ChAdOx1 vaccine has been estimated as 18 to 25 per million doses in Germany and Scandinavia, but only 5 per million doses in the UK based on the Yellow Card reporting scheme. Risk estimates based on adverse event reporting systems are subject to under-ascertainment and other biases. What this study adds All diagnosed cases of CVT in Scotland were ascertained by searching neuroimaging studies from December 2020 to May 2021 and linked to national vaccination records. The risk of CVT within 28 days of vaccination with ChAdOx1 was estimated as 3.5 per million doses with an upper bound of 6 per million doses, against a background incidence of about 12 per million adults per year. This indicates that the Yellow Card system has not seriously underestimated the risk in the UK; the explanation for higher risk in other European countries is not clear.