Key Points

QuestionÌý What is the risk of uveitis recurrence after COVID-19 vaccination in individuals with a history of uveitis?

FindingsÌý In this cohort study including 473 934 individuals with history of uveitis and COVID-19 vaccination, the 3-month and 1-year cumulative incidences of postvaccination uveitis were 8.6% and 16.8%, respectively. Risk varied with different vaccine types and was higher between the first andf second vaccination doses, decreasing after subsequent vaccinations.

MeaningÌý These results support consideration of monitoring for uveitis recurrence, particularly for the first month following vaccination in individuals with a history of uveitis.

ImportanceÌý Understanding the potential risk of uveitis recurrence after COVID-19 vaccination in individuals with a history of uveitis is crucial for vaccination strategies and clinical monitoring.

ObjectiveÌý To investigate the risk of uveitis recurrence after COVID-19 vaccination in a cohort of individuals with a history of uveitis.

Design, Setting, and ParticipantsÌý This retrospective population-based cohort study included individuals diagnosed with uveitis between January 1, 2015, and February 25, 2021, in South Korea. After excluding individuals without COVID-19 vaccination or with SARS-CoV-2 infection, individuals with a history of uveitis who had received at least 1 dose of a messenger RNA (BNT162b2 [Pfizer-BioNTech] or mRNA-1273 [Moderna]) or adenovirus vector-based (ChAdOx1 [AstraZeneca] or Ad26.COV2.S [Janssen]) COVID-19 vaccine were included. Data were analyzed from February 26, 2021, to December 31, 2022.

ExposureÌý Demographic and clinical data, along with vaccination details, were retrieved from the Korean National Health Insurance Service and Korea Disease Control and Prevention Agency databases.

Main Outcomes and MeasuresÌý Outcomes of interest were incidence and risk of postvaccination uveitis in association with different COVID-19 vaccines and periods before and after COVID-19 vaccination. Uveitis was categorized by onset (early, within 30 days, or delayed) and type (anterior or nonanterior). Hazard ratios (HRs) with 95% CIs were calculated to evaluate the risk of uveitis following COVID-19 vaccination, stratified according to vaccine type and vaccination period.

ResultsÌý Of 543 737 individuals with history of uveitis, 473 934 individuals (mean [SD] age, 58.9 [17.4] years; 243 127 [51.3] female) had documented COVID-19 vaccination and were included in analysis. The cumulative incidence of postvaccination uveitis was 8.6% at 3 months, 12.5% at 6 months, and 16.8% at 1 year, predominantly of the anterior type. Variations in the risk of postvaccination uveitis were observed across different vaccines and intervaccination periods. The risk of early postvaccination uveitis was increased for individuals receiving the BNT162b2 (HR, 1.68; 95% CI, 1.52-1.86), mRNA-1273 (HR, 1.51; 95% CI, 1.21-1.89), ChAdOx1 (HR, 1.60; 95% CI, 1.43-1.79), and Ad26.COV2.S (HR, 2.07; 95% CI, 1.40-3.07) vaccines. The risk of uveitis was higher particularly between the first and second vaccination doses (HR, 1.64; 95% CI, 1.55-1.73).

Conclusions and RelevanceÌý These findings suggest that there was an elevated risk of uveitis following COVID-19 vaccination, with the vaccine type and period mediating this risk. For individuals with a history of uveitis, clinicians should consider the potential risk of uveitis recurrence in vaccination strategies and clinical monitoring.

The introduction of COVID-19 vaccines marked a milestone in the global endeavor to combat the ongoing pandemic. These vaccines, including recombinant messenger RNA (mRNA) vaccine, adenovirus vector-based vaccine, and inactivated virus, received emergency use authorization from regulatory agencies worldwide.¹ Despite widespread vaccine deployment, continuous, large, epidemiologic monitoring for both short-term and long-term adverse events are warranted.

Reports of various ocular adverse events, although relatively rare, have emerged. These events have included facial nerve palsy, uveitis, and retinal vascular occlusion.^2-5 Uveitis has been reported following several different vaccinations,^2,3,6 with temporal associations that do not necessarily establishing a causal link.^3,7

For patients with a history of uveitis, understanding the potential impact of COVID-19 vaccination on its course, particularly its potential to induce uveitis recurrence vs recurrences coincidental to widespread vaccinations.^8,9 Anterior uveitis has been associated with administration of various COVID-19 vaccines,^2,10,11 including a recent study reporting estimated crude rates per million doses administered of 0.57 for BNT162b2 (Pfizer-BioNTech), 0.44 for mRNA-1273 (Moderna), and 0.35 for Ad26.COV2.S (Janssen).⁸

While these associations do not necessarily prove causation, several questions remain. It is unclear whether COVID-19 vaccination is associated with a higher incidence of uveitis recurrence in patients with a history of uveitis. Furthermore, risk of uveitis recurrence may be associated with other factors, such as specific dose administered (first, second, or third) or type of vaccine used, warranting further investigation.⁹ Timing of vaccine administration and postvaccination uveitis is of interest for a comprehensive understanding of the temporal association between COVID-19 vaccination and uveitis.

To address these questions, we assessed the Korean National Health Insurance Service (NHIS) and Korea Disease Control and Prevention Agency (KDCA) databases.¹² These databases offer a comprehensive dataset covering approximately 500 000 patients with history of uveitis (NHIS database) and their vaccination details (KDCA database), facilitating a detailed evaluation of the association between COVID-19 vaccination and uveitis.

This cohort study was conducted in accordance with the Declaration of Helsinki and approved by the institutional review board of Hanyang University Hospital. The need for informed consent was waived because of the retrospective nature of the study and deidentification of the data. This report follows the Strengthening the Reporting of Observational Studies in Epidemiology () reporting guideline for cohort studies.

In this study, we examined the entire South Korean population diagnosed with uveitis between January 1, 2015, and February 25, 2021, totaling 543 737 individuals from the Korean NHIS database. We used diagnostic codes for uveitis identification, as outlined in previous reports.^13,14 Codes were derived from the Korean Standard Classification of Diseases, 8th Revision (KCD-8), as slightly modified from the International Statistical Classification of Diseases and Related Health Problems, Tenth Revision (ICD-10). Details, including the list of diagnoses, are provided in eTable 1 in Supplement 1. The KDCA offered additional details on SARS-CoV-2 infection and COVID-19 vaccinations, including types and timing of COVID-19 vaccine doses administered. To analyze occurrence of uveitis before and after vaccination, we excluded 37 637 individuals who did not receive any vaccinations during the vaccination period in South Korea (from February 26, 2021) until the end of the study (December 31, 2022) and 1503 individuals for whom the date of vaccination was not recorded. We further excluded 30 663 individuals with documented SARS-CoV-2 infection during the observation period based on information from the KDCA database to exclude potential confounding effects of SARS-CoV-2 infection on uveitis. The final study population comprised 473 934 individuals with at least 1 COVID-19 vaccination and a history of uveitis, for whom we had records of vaccination dates and types used. A flowchart of inclusion and exclusion criteria and respective numbers of individuals accordingly is provided in eFigure 1 in Supplement 1.

In this study, uveitis was categorized according to its onset and type. We defined early-onset postvaccination uveitis as uveitis occurring within 30 days of the most recent vaccination; uveitis occurring after this period was classified as delayed-onset postvaccination uveitis. Uveitis was separated into 2 types, anterior and nonanterior, using KCD-8 and ICD-10 codes, as outlined in previous reports^13,14 (eTable 1 in Supplement 1).

Incidence of uveitis was assessed before and after the first vaccination. Cumulative incidences were calculated for different postvaccination periods: 3 months, 6 months, and 1 year. To evaluate risk of uveitis, hazard ratios (HRs) and corresponding 95% CIs were calculated. The analyses were conducted by stratifying data into subgroups based on the specific type of COVID-19 vaccine administered: BNT162b2, ChAdOx1, mRNA-1273, and Ad26.COV2.S. Additionally, we examined incidence during each period between vaccinations, including the first intervaccination period (between the first and second vaccination doses), the second intervaccination period, and the subsequent period following the third dose.

Descriptive statistics were used for demographic and clinical information, such as medical diagnoses and details of the vaccines used. Categorical variables are expressed as frequencies and percentages, while continuous variables are presented as mean and SD or median and IQR.

Cumulative incidence was determined using Kaplan-Meier curves. The rates of prevaccination and postvaccination uveitis were computed by person-months. HRs were compared using Cox proportional hazards models. HRs for postvaccination uveitis were obtained using prevaccination uveitis as a reference. All P values were 2-sided but not adjusted for multiple analyses. Statistical analyses were performed with SAS Enterprise Guide version 7.1 (SAS Institute) and R version 4.0.3 (R Project for Statistical Computing). Data were analyzed from February 26, 2021, to December 31, 2022.

The total study population included 473 934 individuals (mean [SD] age, 58.9 [17.4] years; 243 127 [51.3%] female) (Table 1). Most patients were aged 60 to 79 years (207 106 individuals [43.7%]) followed by 40 to 59 years (143 768 individuals [30.3%]). A comorbidity analysis indicated high prevalences of hypertension (216 768 individuals [45.7%]), diabetes (159 726 individuals [34.0%]), and rheumatic diseases (351 900 individuals [74.3%]) in the study population. Receipt of a first COVID-19 vaccine dose was a required inclusion criterion; 469 911 individuals (99.2%) received a second dose, and 396 091 individuals (83.6%) received a third dose, and 186 079 individuals (39.3%) received a fourth dose. By type of vaccines received, 171 928 individuals (36.3%) receiving ChAdOx1 (AstraZeneca), 242 463 individuals (51.2%) receiving BNT162b2, 48 666 individuals (10.3%) receiving mRNA-1273, and 9652 individuals (2.0%) receiving Ad26.COV2.S (Table 1).

Uveitis occurred in 40 815 individuals (8.6%) during the 3-month period after COVID-19 vaccination, with 32 844 individuals (6.9%) experiencing anterior uveitis and 7744 individuals (1.6%) experiencing nonanterior uveitis (Table 2). Higher cumulative incidences of postvaccination uveitis were observed at 6 months (59 126 individuals [12.5%], including 48 071 individuals [10.1%] with anterior type) and 1 year (79 821 individuals [16.8%], including 65 611 individuals [13.8%] with anterior type) (Table 2). The ratio of anterior to nonanterior uveitis was 4.3 at 3 months, 4.4 at 6 months, and 4.8 at 1 year, indicating a similar distribution of uveitis types among the periods.

During the prevaccination phase (reference period), the baseline uveitis rate stood at 0.02 per person-month. By contrast, there was an increased risk of uveitis in the postvaccination period overall (HR, 1.21; 95% CI, 1.19-1.24) (Table 3). Further exploration into early- and delayed-onset postvaccination uveitis revealed higher risk in the early-onset period (HR, 1.64; 95% CI, 1.53-1.76), but maintaining an increased risk in the delayed-onset period compared with the prevaccination period (HR, 1.18; 95% CI, 1.15-1.21), suggesting a sustained risk over time. eTable 2 in Supplement 1 presents incidence rates and HRs of prevaccination and postvaccination uveitis in different time periods, indicating higher risk in all of the postvaccination periods compared with prevaccination, particularly in the earlier period.

There were associations between anatomic types (anterior vs nonanterior) and disease processes (infectious vs noninfectious) between prevaccination and postvaccination uveitis (eTable 3 in Supplement 1). The types or disease processes of prevaccination uveitis were associated with those of postvaccination uveitis (eTable 3 in Supplement 1). Among the patients with infectious uveitis in the prevaccination period, more than half (53.7%) had noninfectious uveitis, whereas most individuals with noninfectious uveitis before vaccination had noninfectious uveitis after vaccination.

Patients who received the BNT162b2 mRNA vaccine had increased risk of uveitis in the postvaccination period compared with prevaccination (HR, 1.23; 95% CI, 1.19-1.27), with a more pronounced increase in the early-onset period (HR. 1.68 (95% CI, 1.52-1.86) compared with the delayed-onset period (HR, 1.19; 95% CI, 1.15-1.23) (Table 4). After the administration of mRNA-1273, the rate of uveitis increased after the first vaccination (HR, 1.22; 95% CI, 1.13-1.31). Additionally, there was an increased risk of uveitis in the early-onset period after mRNA-1273 vaccination (HR, 1.51; 95% CI, 1.21-1.89) (Table 4).

Compared with the prevaccination period, patients had increased risk of uveitis after vaccination with the ChAdOx1 adenovirus vector-based vaccine overall (HR, 1.19; 95% CI, 1.15-1.23) and in the early-onset period postvaccine period (HR, 1.60; 95% CI, 1.43-1.79) (Table 4). Following administration of Ad26.COV2.S, another adenovirus vector-based vaccine, the rate of uveitis incidence remained at 0.03 (HR, 1.16; 95% CI, 1.01-1.33). Notably, there was substantially elevated risk of uveitis in the early-onset after Ad26.COV2.S vaccination period (HR, 2.07; 95% CI, 1.40-3.07), indicating an increased risk of uveitis within the initial 30 days following vaccination with Ad26.COV2.S (Table 4).

Risk of postvaccination uveitis varied across different postvaccination time periods (Figure). Risk was highest for the period between the first and second vaccination. Similarly, between the second and third vaccinations, the risk of postvaccination uveitis remained elevated (Figure). For the period after the third vaccination, the risk was only slightly increased compared with the reference prevaccination period (Figure). These findings suggest that the risk of recurrent uveitis decreased from the first to subsequent doses.

eTable 4 in the Supplement provides a detailed overview of the incidence rates and HRs associated with postvaccination uveitis based on the administered vaccines, as separated into different intervaccination periods. The greatest risk of postvaccination uveitis was observed after the first vaccine dose, regardless of which vaccine was used. In the overall postvaccination period, the BNT162b2 , mRNA-1273, ChAdOx1, and Ad26.COV2.S vaccines were associated with similarly increased risk of uveitis (eFigure 2 in the Supplement 1). The early postvaccination and delayed postvaccination periods showed similar trends, with a comparable but slightly higher HR of uveitis following the BNT162b2 dose than following mRNA-1273 or ChAdOx1 (eFigure 2 in the Supplement 1). These results underscore the variability of uveitis risk as associated with different COVID-19 vaccines and intervaccination periods.

The comparison of postvaccination uveitis between patients with and without systemic autoimmune diseases is presented in eTable 5 in the Supplement. There was a higher risk of uveitis for individuals with systemic diseases only during the early postvaccination period.

This cohort study aimed to address several clinical questions related to the risk of uveitis recurrence following COVID-19 vaccination in patients with a history of uveitis. The introduction of COVID-19 vaccines was a crucial step in the global fight against SARS-CoV-2 infection.¹⁵ Because these vaccines have been administered widely, concerns have arisen regarding their potential adverse effects, including rare ocular events that might be addressed with large epidemiologic investigations.

Our study used data from the NHIS database, which provides a comprehensive dataset covering a large population with a history of uveitis in South Korea. By examining the entire South Korean population diagnosed with uveitis between January 1, 2015, and February 25, 2021 (the day before the first COVID-19 vaccination in South Korea), we were able to evaluate the association between COVID-19 vaccination and uveitis and to explore uveitis incidences and risk of recurrence in different time frames. For example, we have provided the risk of postvaccination uveitis with reference to the prevaccination period. Intervaccination periods may better indicate uveitis recurrence after COVID-19 vaccination, as the periods of vaccination varied among the study population; thus, a specific, fixed time point (eg, 3 months or 6 months after the first vaccination) may lead to heterogeneous vaccine statuses among the population. However, the common time points used for outcomes in the field of ophthalmology may also be useful for understanding incidences, for which reason we have also provided 3-month, 6-month, and 1-year incidence rates together with HRs, as well as cumulative incidences of postvaccination uveitis for those periods.

Our study offers several findings. First, the cumulative incidences of uveitis were 8.6% and 16.8% during the 3-month and 1-year periods, respectively, signifying that more than half of the incidences occurred during the early period. Moreover, we observed higher HRs for early-onset postvaccination uveitis, signifying a particularly increased risk within the initial 30 days after vaccination. The close temporal relationship between COVID-19 vaccination and incidence of uveitis supports their association. These findings also highlight the clinical implications of close monitoring for uveitis during the early period after vaccination in patients with a history of uveitis.¹⁶

Additionally, our analysis considered the postvaccination period and revealed notable variations in the risk of uveitis among the intervaccination periods. Specifically, the risk was higher between the first and second vaccination doses. Intriguingly, the risk decreased following subsequent vaccinations. In a recent study on the BNT162b2 mRNA vaccine and noninfectious uveitis, the standardized incidence ratio following vaccination was 1.41 after the first dose and 1.31 after the second dose,¹⁶ indicating a greater risk of uveitis after the first dose. This result is in line with our findings that all 4 of the vaccines in this study showed a common trend: a greater risk after the first dose than after subsequent doses. The increased immune response following the initial dose might activate inflammatory pathways, resulting in conditions like uveitis, particularly in individuals prone to autoimmune reactions or with a uveitis history. We hypothesize that the decline in risk with subsequent doses may stem from the immune system adapting to the vaccine antigen, resulting in a more controlled immune response that mitigates inflammatory side effects,^17,18 a possibility that should be validated by future studies. Furthermore, individuals who experienced flare-ups with earlier vaccine doses, particularly those with a previous diagnosis of uveitis, might have received prophylactic therapy or their treatment may have been intensified before subsequent doses, which possibilities, in any case, could not be fully evaluated in our study. Further investigations are required to elucidate the underlying mechanisms for the observed trend.

Our analysis also provides data on the associations of different COVID-19 vaccines with the risk of uveitis. All 4 vaccines were associated with an increased uveitis risk after the first dose. BNT162b2 exhibited a higher HR for postvaccination uveitis compared with other vaccines, with minimal systematic differences observed between mRNA vaccines and adenovirus vector-based vaccines. In terms of the onset of postvaccination uveitis, early-onset uveitis comprised 9.9%, 10.4%, 9.3%, and 11.7% among the patients receiving BNT162b2, mRNA-1273, ChAdOx1, and Ad26.COV2.S vaccines for the first vaccination, respectively, indicating no significant difference in onset between vaccine classes. Among the vaccine types, a large-scale database analysis based on 1094 patients with uveitis from 40 countries showed estimated crude reporting rates of 0.57, 0.44, and 0.35 per million doses for BNT162b2, mRNA-1273, and Ad26.COV2.S, respectively.⁸ Such a trend is also in agreement with our findings, indicating a greater risk associated with BNT162b2, followed by mRNA-1273 and Ad26.COV2.S in the overall postvaccination period. Whether or why the BNT162b2 or mRNA-1273 vaccine was associated with a higher incidence of uveitis relative to the other COVID-19 vaccines remains unclear^8,16; however, our data may be helpful in estimating the risk of postvaccination uveitis for each vaccine and assisting clinical decision-making regarding the choice of COVID-19 vaccines in patients with a history of uveitis.

Nevertheless, the lower bounds of the 95% CIs of many of our findings are close to 1.0, within a range where confounding factors could contribute to the associations identified. It remains unclear whether COVID-19 vaccination increases the risk of uveitis recurrence, as not every study to date has demonstrated an elevated risk of flares following vaccination. A 2024 study from the United Kingdom showed no increase in the reporting rate of uveitis following COVID-19 vaccination compared with the range of incidence in the overall United Kingdom population.¹⁹ Furthermore, a matched cohort and self-controlled case series analysis did not detect any overall increased noninfectious uveitis risk after COVID-19 vaccination in individuals without a history of uveitis.²⁰ However, a 2023 study performed in New Zealand showed that chronic or recurrent uveitis was associated with increases risk for uveitis after COVID-19 vaccination,⁹ and another population-based study showed increased standardized incidence ratios of noninfectious uveitis following the first or second dose in individuals with previously known uveitis.¹⁶ Our study, by comparing the incidence rates and calculating HRs for postvaccination uveitis with reference to prevaccination uveitis, found that an existing diagnosis prior to COVID-19 vaccination was associated with increased incidence rates and HRs, as is consistent with the previous findings on risk factor analysis.^9,16

The limitations of our study include the reliance on data from the Korean NHIS database, which may not capture all uveitis cases due to potential underreporting, leading to underestimation of our outcomes. Additionally, the potential inaccuracy of diagnostic codes in the database might lead to misclassification bias. While we have investigated uveitis in patients with a history of uveitis, postvaccination uveitis may occur in those without any history of uveitis. Although our study design for analysis of individuals with a history of uveitis could not evaluate the incidence of such de novo uveitis, which is another important limitation of our study, this should be explored and compared with the recurrence rate in those with a history of uveitis in future studies. Also, the NHIS database lacks detailed clinical information, making it challenging to fully characterize the severity or specific features of uveitis cases. Moreover, the absence of certain clinical details limited our ability to explore potential interactions or stratifications that could refine our understanding of the observed associations. Additionally, the retrospective study design introduces inherent biases, including reliance on existing data, and the relatively limited follow-up duration might not have captured long-term effects or rare events that may further increase recurrence rates. Furthermore, we did not adjust for multiple analyses, wherein we might have considered the lower bounds of a 99% CI to determine the strength of all of the associations evaluated.

The inclusion of populations receiving multiple types and sequences of vaccines posed challenges in attributing observed outcomes to a specific vaccine type. However, only slight differences in postvaccination HRs were observed between the patients who had received only BNT162b2 and those who had received a combination of different vaccines including BNT162b2. Nevertheless, we separated our analysis into intervaccination periods according to the vaccine used most recently to analyze the outcomes separately by vaccine and intervaccination period. Moreover, whereas we adjusted for several potential confounding variables, such as demographic and comorbidity factors by self-controlled study design methods, there may yet be unmeasured or residual confounders that could have influenced the observed associations.¹³ Notably, influence of medications used during both the prevaccination and postvaccination periods remains an important consideration, as anti-inflammatory medications may decrease the rate of postvaccination uveitis. Furthermore, external validity and the generalizability of our findings to other populations should be interpreted cautiously.²¹

This cohort study provides insights into the associations between COVID-19 vaccination and uveitis. Although uveitis following vaccination is rare, our findings support an increased risk after COVID-19 vaccination, particularly in the early postvaccination period. The type of vaccine administered also was associated with risk of uveitis. These results emphasize the importance of vigilance and monitoring for uveitis in the context of vaccinations, including COVID-19 vaccinations, particularly in individuals with a history of uveitis.

Accepted for Publication: February 20, 2024.

Published Online: April 25, 2024. doi:10.1001/jamaophthalmol.2024.0973

Corresponding Author: Seong Joon Ahn, MD, PhD, Department of Ophthalmology, Hanyang University Seoul Hospital, 222-1 Wangsipli-ro, Seongdong-gu, Seoul 04763, Korea (ahnsj81@gmail.com).

Author Contributions: Dr Ahn had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Concept and design: Kim, Ahn.

Acquisition, analysis, or interpretation of data: All authors.

Drafting of the manuscript: Ahn.

Critical review of the manuscript for important intellectual content: All authors.

Statistical analysis: All authors.

Obtained funding: Kim, Ahn.

Administrative, technical, or material support: Kim, Ahn.

Supervision: Ahn.

Conflict of Interest Disclosures: None reported.

Funding/Support: This work was supported by the National Research Foundation of Korea (grant No. NRF-2021R1G1A1013360) funded by the Korean Government MSIT and by the research fund of Hanyang University (grant No. HY-202300000003500).

Role of the Funder/Sponsor: The funders had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

Data Sharing Statement: See Supplement 2.

Additional Contributions: Ji Hong Kim provided technical assistance in support of this study.

Additional Information: This study used the database of the Korea Disease Control and Prevention Agency and the Korean National Health Insurance Service for policy and academic research. The research number of this study is KDCA-NHIS-2023-1-469.