Thromboembolic Events after COVID-19 Vaccination: An Italian Retrospective Real-World Safety Study

The study addresses whether COVID-19 vaccines are associated with thromboembolic risks in real-world use and how these risks compare across vaccine platforms. After several vaccines received EMA authorization (mRNA: Pfizer-BioNTech BNT162b2, Moderna mRNA-1273; adenoviral vector: Oxford-AstraZeneca ChAdOx1 nCoV-19, Janssen Ad26.COV2.S; protein-based: Novavax NVX-CoV2373, Sanofi Vidprevtyn; inactivated: Valneva VLA2001), spontaneous reports of thromboembolic events—particularly with adenoviral vector vaccines—raised safety concerns, leading to temporary suspensions and ongoing regulatory review. Prior population-based studies suggested increased risks of certain thrombotic events after Oxford-AstraZeneca vaccination, while mRNA vaccines showed mixed or limited associations. This study aims to estimate the incidence (reporting rates) of thromboembolic events after COVID-19 vaccination in the Campania Region (Italy) and to compare the risk profile of Oxford-AstraZeneca with other vaccines using real-world data.

The discussion reviews prior evidence with mixed findings. VAERS analyses and a Danish retrospective cohort reported no significant increase in thromboembolic or thrombocytopenic events after mRNA vaccination. A large multi-state US health system study found cerebral venous sinus thrombosis (CVST) to be rare and not significantly associated with COVID-19 vaccination. International network cohort analyses found increased risk of thrombocytopenia with Oxford-AstraZeneca compared with Pfizer-BioNTech. Danish and Norwegian cohorts identified elevated standardized incidence ratios for venous thromboembolism and thrombocytopenia within 28 days of Oxford-AstraZeneca vaccination, largely driven by CVST, with no increased arterial thromboembolism. Scottish case-control work found increased idiopathic thrombocytopenic purpura (ITP), arterial thromboembolism, and hemorrhagic events after the first dose of Oxford-AstraZeneca; other UK studies noted associations with thrombocytopenia, venous thromboembolism, and CVST after Oxford-AstraZeneca, and arterial thromboembolism or CVST after Pfizer-BioNTech in some subgroups. Additional studies reported increased pulmonary embolism after Pfizer-BioNTech and elevated risks of pulmonary embolism and thrombocytopenia after Oxford-AstraZeneca, with immune thrombocytopenia risk for both Oxford-AstraZeneca and Pfizer-BioNTech. Mechanistic hypotheses include vaccine-induced immune thrombotic thrombocytopenia (VITT) mediated by anti-PF4 antibodies, adenoviral vector interactions with platelets and PF4, and sex-related differences in platelet activation.

Design: Retrospective real-world safety study utilizing two data sources: (1) the Italian Pharmacovigilance database (Rete Nazionale di Farmacovigilanza, RNF) for Individual Case Safety Reports (ICSRs) of suspected adverse reactions; and (2) the Campania Region health information system (SINFONIA) for regional vaccination data. Population and period: Vaccinations administered in the Campania Region from the start of the vaccination campaign to 27 September 2022. Data sources and extraction: From SINFONIA, anonymized aggregated data on doses administered by age, sex, vaccine type, and dose number were extracted using ML-assisted selection and organization workflows (Python/Spyder; results organized in Tableau). From RNF, ICSRs that listed a COVID-19 vaccine as suspected and at least one vascular event (MedDRA SOC: Vascular disorders, v25.1) were retrieved; data included age, sex, event(s), and seriousness (ICH E2D criteria). Vaccines considered: Pfizer-BioNTech, Moderna, Oxford-AstraZeneca, Janssen, Novavax. Outcomes: Thromboembolic events identified within vascular events from ICSRs; seriousness classification per ICH E2D; event counts by vaccine and dose. Measures: Reporting rate (RR) of thromboembolic events per 10,000 doses for each vaccine (events from RNF divided by doses administered in SINFONIA) with 95% CI. Comparative analysis: Odds ratios (ORs) comparing the odds of exposure to Oxford-AstraZeneca vs. other vaccines among cases (with thromboembolic events) vs. controls (without such events). Statistical analysis: ORs and 95% CIs computed; significance threshold 5%; analyses performed in R (v3.2.2). Ethics: Retrospective pharmacovigilance study using anonymized data; per national rules, no informed consent or ethics committee approval required for anonymized data; use of SINFONIA data approved by the relevant Ethical Committee (Protocol 00000926, 11 January 2022).

- Vaccination coverage: 12,692,852 doses administered in the Campania Region; 51.28% to females. Vaccine distribution: Pfizer-BioNTech 65.05%, Moderna 24.31%, Oxford-AstraZeneca 9.71%, Janssen 0.91%, Novavax 0.02%. - Safety reports: 641 ICSRs with vascular events identified; 70.67% female; 2451 total events (mean 3.8 per ICSR); 292 thromboembolic events reported. Most ICSRs concerned first doses (51.32%); 40.72% included at least one serious event. - Reporting rates (RRs) of thromboembolic events per 10,000 doses: Oxford-AstraZeneca 4.62 (95% CI: 3.50–5.99); Janssen 3.45 (95% CI: 0.94–8.82); Pfizer-BioNTech 0.76 (95% CI: 0.59–0.98); Moderna 0.71 (95% CI: 0.45–1.08). - Comparative odds: Thromboembolic events were associated with higher odds of exposure to Oxford-AstraZeneca compared with Pfizer-BioNTech (OR 6.06; 95% CI: 4.22–8.68) and Moderna (OR 6.46; 95% CI: 4.00–10.80); no significant difference compared with Janssen. - Event types: Venous thromboembolism events were more frequently reported than arterial events; commonly reported preferred terms included thrombosis, venous thrombosis, deep vein thrombosis, and elevated D-dimer.

Findings indicate higher reporting rates of thromboembolic events with adenoviral vector vaccines (Oxford-AstraZeneca and Janssen) relative to mRNA vaccines in a large regional cohort. The elevated odds of Oxford-AstraZeneca exposure among thromboembolic cases versus Pfizer-BioNTech and Moderna align with regulatory assessments and several population-based studies that identified increased risks of thrombocytopenia and venous thromboembolic events after adenoviral vector vaccination. Literature comparisons reveal heterogeneous results for mRNA vaccines, with some studies showing no overall increased risk and others identifying specific associations (e.g., pulmonary embolism or arterial events) in certain subgroups. Potential biological mechanisms include VITT mediated by anti-PF4 antibodies, adenoviral vector interactions with megakaryocytes and platelets leading to thromboinflammation, and sex-related differences in platelet activation that may contribute to the observed higher reporting in females. The study underscores the importance of continuous pharmacovigilance and contextually weighing rare thromboembolic risks against the substantial benefits of COVID-19 vaccination in preventing severe disease.

The study identified higher reporting of thromboembolic events following viral-vector COVID-19 vaccines (Oxford-AstraZeneca and Janssen) and greater odds of Oxford-AstraZeneca exposure among thromboembolic cases compared with mRNA vaccines (Pfizer-BioNTech and Moderna). While these events are rare, ongoing pharmacovigilance is essential to refine risk assessment and management. Leveraging Big Data resources like SINFONIA and ML-assisted analytics can enhance detection and evaluation of vaccine-related adverse events. COVID-19 vaccination remains the most effective strategy to mitigate the pandemic, and safety concerns should be balanced against vaccination benefits. Future work should refine event phenotyping, assess dose- and time-specific risks, and incorporate individual-level risk factors to improve causal inference.

- Use of spontaneous reporting data (RNF) subject to underreporting and variable data quality, potentially underestimating true event rates and limiting clinical detail (e.g., risk factors, precise timing from vaccination to event onset). - Aggregated vaccination data from SINFONIA prevented linkage at the individual level and limited analyses by dose or time window for specific vaccines. - Pharmacovigilance reports indicate suspected associations rather than confirmed causality, which may introduce reporting biases. - Limited numbers for specific thromboembolic phenotypes restricted analyses to aggregated thromboembolic events rather than detailed subtype-specific risk estimates.