Dorney et al.1 found that mRNA COVID-19 vaccinations did not increase the risk of retinal vein/artery occlusions (RVO) whereas Li et al.2 found that all COVID-19 vaccines increased the risk of RVO. Although it may appear that our studies have differing conclusions, we aim to show where and why our studies are in agreement. Our analysis specifically evaluated mRNA vaccinations and the risk of developing RVO 21 days after the first COVID-19 vaccinations from 2020 to 2022 compared to patients receiving the influenza and Tdap vaccination in 2018 to 20191. Li et al. stratified by COVID-19 vaccination brand and observed the risk of RVO development at both 12 weeks and 2 years after vaccination compared to unvaccinated patients. At the 12 weeks, neither study showed an association between COVID-19 vaccinations and RVO (Fig. 1). Previous literature regarding RVO after COVID-19 vaccination reported development within several days to a maximum of 6 weeks post-vaccination, making the clinically relevant timeline a scale of days to weeks, not months to years1,3. This was further supported by a recent review of ocular adverse events after vaccination where the average time between vaccination and adverse events was 10.8 days4.

This forest plot compares the risk of retinal vascular occlusions at 3 and 12 weeks after mRNA COVID-19 vaccination. *Dorney et al. Study1 (evaluated 3 weeks after COVID vaccination). + Li Study2 (evaluated 12 weeks after COVID vaccination).

Li’s study had several additional selection biases and uncontrolled confounders compared to our study including a restrictive exclusion criteria, removing patients on anticoagulants, antiplatelet, contraceptives, diuretics as well as other commonly prescribed medications. Given that many American patients are on at least one of these classes of medications, this greatly limits the generalizability of this study. Furthermore, the comparison of vaccinated individuals to unvaccinated individuals raises concerns as these two groups may differ fundamentally in access to care and health behaviors, which could artificially minimize reported rates of RVO in the unvaccinated control group5.

While we commend their attempts to exclude COVID-19 positive patients and only include COVID-19 negative patients, there are several flaws with the methodology used. In the study, 562,700 patients with documented COVID-19 infections were excluded out of the 7.3 million study eligible patients, which is likely a high underreporting of COVID-19 cases. This analysis only excludes documented positive COVID-19 tests at a TriNetX affiliated institution, missing the asymptomatic proportion of patients, patients taking at-home tests, or tests at other healthcare institutions. The CDC reported 146.6 million Americans likely had COVID-19, equating to 43.9% of the population6. Therefore, approximately 3.2 million patients within this cohort would have likely had COVID-19 infections but were still included. COVID-19 is a hyper-thrombotic condition7 with a known risk of RVO1,2. Thus, including patients with only one documented negative COVID-19 test is insufficient to accurately capture a 2 year time period where 43.9% of patients were likely to have at some point been infected with COVID-19. This would bias both vaccinated and especially unvaccinated patient groups who had no vaccine protection against contracting COVID-19.

In addition, although patients were required to have a negative COVID-19 test to be in the study; this negative result could have changed at any time throughout the 2 year research time period. Thus, it is likely that many patients included in the analysis contracted COVID-19 at some point within the 2 years, which could have also influenced the increased associated risk between COVID-19 vaccines and 2 years after the vaccine reported in Li et al.’s study. Our analysis had the advantage of using influenza and Tdap vaccinations in 2018 and 2019 as a control, thus eliminating COVID-19 infection as a potential confounder in our control group. Lastly, this inclusion criteria also overlooks that patients with a documented negative COVID-19 test would only be receiving the test if they had a known exposure or COVID-19 like symptoms. Thus, this inclusion criteria drastically reduced the patient pool from 95 million patients to a mere 6.7 million, resulting in an inclusion of only 739,066 vaccinated individuals, while also failing to include individuals who were COVID-19 negative with no known exposures or symptoms.