Introduction: Early in the pandemic and the vaccination campaign, cases of immune thrombocytopenia (ITP) occurring after SARS-CoV-2 infection and COVID-19 vaccines have been reported. Both U.S. and European health authorities prompted heightened surveillance regarding the risk of ITP following vaccination. Although such cases were expected to be sporadic, the possibility of an increased risk at the population level remained uncertain to date.

Therefore, we aimed to measure the association between COVID-19 vaccines overall and by type (mRNA vaccines: Comirnaty Pfizer - BNT162b2; Spikevax Moderna - mRNA-12773; and adenovirus-based vaccines: Vaxzevria AstraZeneca -ChadOx-1-S, and JCovden Janssen AD26COV-2.S) and the occurrence of ITP at the nationwide scale in France. We also investigated the association between SARS-CoV-2 infection (defined by a SARS-CoV-2 positive test) and ITP.

Methods: We conducted a population-based study in France using the FAITH cohort (NCT03429660) built within the National Health Database (demographic data, in-hospital and out-hospital data for the entire country, 67 million inhabitants), linked with COVID-19 vaccination (VAC-SI) and SARS-CoV-2 diagnostic testing (SI-DEP) databases. All patients aged ≥18 years with a new diagnosis of ITP between January, 2021 and December, 2023 were included. Cases were identified using a validated algorithm combining ICD-10 codes. We evaluated the association between COVID-19 vaccination (overall and by type) or SARS-CoV-2 infection (defined by a RT-PCR or antigenic positive test) and ITP onset using three methods: two self-controlled designs (recommended for vaccine safety studies): a self-controlled case series (SCCS) and a case-crossover (CCO) designs; and a case-control design. In the SCCS analysis, we compared the incidence of ITP during risk periods following each vaccination (6 weeks) or SARS-CoV-2 infection (from 7 days before the positive test i.e. the first symptoms to 6 weeks after the positive test) to the incidence of ITP during other periods of time, excluding the 2 weeks prior vaccination to address the healthy vaccinee effect. Incidence rate ratios (IRRs) adjusted for age and seasonality were estimated using a Poisson regression. For the CCO analysis, we compared the frequency of exposure to vaccine or of SARS-CoV-2 positive test during a case period (the 6 weeks preceding immediately the diagnosis) to a control period (3 months before the diagnosis). Lastly, in the case-control analysis, ITP cases were matched with 5 controls from the general population for age, sex, insurance scheme and department of residency. Controls were assigned index dates corresponding to those of their matched cases. We compared the exposure to COVID-19 vaccine or SARS-CoV-2 positive test in the 6 weeks before the index date between cases and controls. For the CCO and case control analyses, odd ratios (OR) were estimated using conditional logistic regression models. For the three designs, sensitivity analyses were conducted using 4-, 8- and 12-week interest periods. Subgroups analyses were also conducted by ITP duration (<3 months or ≥3 months).

Results: We included 4,609 patients aged ≥18 years with a new diagnosis of primary ITP, with a median age of 62 years; 53% were male. Among them, 77% had received at least one dose of vaccine any time prior to ITP diagnosis; 29% had a positive SARS-CoV-2 test any time before ITP diagnosis.

All three designs yielded consistent results: no association was observed for COVID-19 vaccines overall (aIRR: 1.06, 95 % confidence interval: 95% CI: 0.96 - 1.17), nor for mRNA vaccine (aIRR for BNT162b2 vaccine: 1.00, 95% CI: 0.91 - 1.11 and aIRR for mRNA-1273 vaccine: 0.72, 95% CI: 0.45 - 1.17). However, an association was found for ChadOx1-S vaccine (aIRR: 2.74 95% CI 95%: 1.86 - 4.03). An association was also found for SARS-CoV-2 infection (aIRR: 2.50 95% CI: 2.22 - 2.81), particularly for ITP with a duration <3 months (aIRR: 4.23; 95% CI 3.57 - 5.03). The comparative analyses could not be performed for AD26COV-2.S due to too few patients with ITP exposed to this vaccine during the study period (n=83).

Conclusion: This nationwide population-based study using three different designs showed an association between SARS-CoV-2 infection and ChadOx1-S vaccine with ITP occurrence, while no association was found for mRNA vaccines.

This content is only available as a PDF.
2025
Sign in via your Institution