Ad26.COV2.S vaccine and thromboembolic risks among premenopausal women: Exploring variation by combined hormonal medication use

Background: Exposure to both the Ad26.COV2.S vaccine and combined hormonal contraception have been independently associated with venous thromboembolism (VTE). However, no studies have evaluated their potential interaction on VTE risk.

Objectives: To estimate the association between the first dose of the Ad26.COV2.S (vs. BNT162b2 or mRNA-1273, hereafter referred to as mRNA) vaccine and VTE among women aged 18-49, stratified by exposure to combined hormonal medication in the six weeks before vaccine receipt.

Methods: We conducted a retrospective cohort study across five United States administrative claims databases between 2021 and 2023. We estimated the incidence of VTE, including pulmonary embolism or deep vein thrombosis, occurring in four periods following vaccine receipt: 1-14, 1-28, 1-42, and 1-90 days. Cox regression models with propensity score matching were fitted comparing Ad26.COV2.S to mRNA vaccines, generating hazard ratios (HR) and 95% Confidence Intervals (CI). Estimates were calibrated against a distribution of effect estimates from 93 negative control outcomes. All analyses were stratified by exposure to combined hormonal medications (i.e., including both an estrogen and progestin RxNorm ingredient) within the six weeks before vaccine receipt. For each database, results were subjected to a predetermined set of diagnostics, where any analysis that failed ≥1 diagnostic did not have results unblinded. HRs across databases were synthesized using Bayesian random-effects meta-analysis.

Results: Across databases, 517,918 Ad26.COV2.S-exposed individuals were matched with 6,632,337 mRNA-exposed, with an average of 17% exposed to combined hormonal medication. The incidence of VTE was < 7 events and < 5 events per 1,000 person-years in the Ad26.COV2.S and mRNA cohorts, respectively. In the stratum unexposed to combined hormonal medications, HRs from meta-analyses ranged from 1.20 to 1.51 across times at risk, with the 1-28 day window showing statistically significant results at α=0.05 (HR=1.48, 95% CI: 1.02, 2.14). In the stratum exposed to combined hormonal medications, 55% of estimates across databases and times at risk (and 100% of those in the 1-28 day window) failed diagnostics and results were blinded. Among those that passed diagnostics in the combined hormonal medication exposed stratum, meta-analytic HRs ranged from 0.92 to 1.11, and all 95% CIs included the null.

Conclusions: Evidence from comparisons of the Ad26.COV2.S with mRNA vaccine across combined hormonal medication exposure strata does not suggest that the risk of VTE differs between those exposed and unexposed to combined hormonal medications. These analyses highlight the challenges of evaluating the potential for effect measure modification in the presence of rare outcomes.