The spectrum of non-fatal immune-related adverse events following COVID-19 vaccination: The population-based cohort study in Seoul, South Korea

Data source

This from the Korean National Health Insurance Service (NHIS) database on 1, January 2021 enrolled randomly extracted 50% of individuals residing in Seoul, South Korea. We randomly selected 50% of the residents living in Seoul as of January 1, 2021, and obtained their diagnostic records from 2020 to 2021. The International Classification of Diseases, 10th revision (ICD-10), was adopted by the NHIS to classify disease diagnoses. The data included the primary diagnosis, secondary diagnosis, and dates of hospital visits. This population-based cohort study was conducted by the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines.²¹

Study population

A total of 4 348 412 individuals living in Seoul, constituting 50% of the population, were included and investigated as of January 1, 2021. Individuals aged under 20 years were excluded, leaving 4 203 887 individuals for analysis. In this study, only individuals who had received two doses of COVID-19 vaccine were included in the vaccinated group. In this cohort study, the index date, which is the date on which individuals started participating in the study, was set differently for the vaccinated and non-vaccinated groups. For the vaccinated group, the index date was set as the date of the second vaccine dose administered before September 30, 2021. On the other hand, for the non-vaccinated group, the index date was set as October 1, 2021. The history of COVID-19 vaccination was investigated as of January 1, 2021. On January 1, 2021, the vaccinated group included 3 839 014 individuals, whereas the non-vaccinated group included 364,873 individuals. Individuals who received a dose of vaccine before January 1, 2021, and did not receive a second dose between January 1, 2021, and September 30, 2021, were excluded. The vaccinated group included 2 154 389 individuals, whereas the non-vaccinated group included 350,953 individuals.

The non-fatal irAEs included gynecological (endometriosis, and menstrual disorders [polymenorrhagia, menorrhagia, abnormal cycle length, oligomenorrhea, and amenorrhea]), hematological (bruises confined to nontender and yellow-colored on especially extremities), dermatological (herpes zoster, alopecia, and warts), ophthalmological (visual impairment, and glaucoma), otological (tinnitus, inner ear, middle ear, and outer ear disease), and dental problems (periodontal disease) as reported by the Vaccine Adverse Event Reporting Center. To investigate the causal relationship between vaccine administration and irAEs, the diagnostic records for a year before the index date were traced. Individuals with any target disease as a primary or secondary diagnosis during this period were excluded from the study. The occurrence of the target disease was defined as receiving a primary diagnosis of the disease from the day after the index date.

Outcome measurements

The primary outcome measure was cumulative incidence rates (cIRs) of irAEs per 10,000 population between the vaccinated and non-vaccinated subjects. The cIRs of the irAEs were measured at one week, two weeks, one month, and three months. The secondary outcome measures were the odd ratios (ORs) and hazard ratios (HRs) of irAEs. Furthermore, subgroup analyses were also conducted based on gender, the number of COVID-19 vaccine doses, the vaccine type (mRNA vaccine, cDNA vaccine, and heterologous vaccination), health insurance level, presence of diabetes mellitus (DM), hypertension (HTN), hyperlipidemia, and chronic obstructive pulmonary disease (COPD).

Age, gender, insurance level, Charlson’s comorbidity index (CCI), presence of DM, HTN, hyperlipidemia, and COPD, and prior COVID-19 infection history were extracted using their ICD-19 codes, which were suggested by Sundararajan et al.²² The presence of comorbid diseases (i.e. DM, HTN, hyperlipidemia, and COPD), items in CCI, and the prior COVID-19 infection history was determined as a primary or secondary diagnosis 2 or more times within 1 year before the index date. The NHI premium was used as a proxy measure of income because it is proportional to monthly income, including earnings and capital gains. The income quantiles of the enrolled individuals were categorized into three groups (low-, middle- and high-income groups in medical aid enrollees and the 0–33, 34–66, and 67–100 centiles of NHI enrollees).

Statistical analysis

Statistical analysis was performed using SAS Enterprise Guide (version 8.3., SAS Institute, Cary, NC, USA). A normal distribution was confirmed with the Kolmogorov–Smirnov test. Baseline patient characteristics and comorbidities were reported as means ± standard deviation for continuous variables and frequency (percentage, %) for categorical variables. Student’s t-test was performed for continuous variables, and the chi-square test or immune-mediated adverse events associated with COVID-19 vaccination were assessed using Student’s t-test for continuous variables, and the chi-square test or Fisher’s exact test for categorical variables. The cIR was calculated per 10,000 populations. To identify the association between COVID-19 vaccination and irAEs, a multiple logistic regression model was used for ORs, corresponding to 95% CIs. Cox proportional hazards regression was used to estimate the HRs and 95% CIs. Two-sided P values of 0.05 or less were considered to indicate statistical significance

The participants’ characteristics

In total, 1 748 136 subjects were included in this study. Among them, 289 579 (16.57 %) had not received the Covid-19 vaccine (i.e. non-vaccinated subjects), whereas 1 458 557 (83.43 %) were vaccinated against COVID-19 (i.e. vaccinated subjects) (Figure 1). The baseline characteristics of the vaccinated and non-vaccinated groups are shown in Table 1.

• Download figure
• Open in new tab

Figure 1. Flowchart of this study

The cIRs per 10,000 of the non-fatal irAEs following the COVID-19 vaccination

Among the non-fatal irAEs in this study, the cIRs at three months following COVID-19 vaccination were higher in vaccinated subjects than in non-vaccinated subjects, except for endometriosis. The highest cIR of the non-fatal irAEs in vaccinated subjects was observed in other ear diseases (cIR, 51.78%; 95% CI, 50.61-52.94) followed by inner ear diseases (cIR, 47.10%; 95% CI, 45.99-48.21), herpes zoster (cIR, 45.08%; 95% CI, 43.99-46.17), menstrual disorders (cIR, 44.43%; 95% CI, 43.35-45.51), and glaucoma (cIR, 39.42%; 95% CI, 38.40-40.43). Among the non-fatal irAEs, 50% exhibited a significant difference in cIRs one-week post-vaccination. Menstrual disorders and visual impairments were noted from one-month, whereas alopecia, warts, and periodontal disease were observed from two weeks onwards (Table 2).

When stratified by gender, the cIRs of irAEs showed a similar pattern to that of the overall population. At three months post-vaccination, menstrual disorders presented the highest cIRs in females (cIR, 87.54%; 95% CI, 85.41-89.66) followed by inner ear diseases in females (cIR, 62.82%; 95% CI, 61.02-64.62), other ear diseases in females (cIR, 55.27%; 95% CI, 53.58-56.96), and herpes zoster in females (cIR, 53.13%; 95% CI, 51.48-54.79). When stratified by vaccine type, heterologous vaccination increased the cIR of menstrual disorders to 78.96% (95% CI, 73.45-84.48). The detailed cIRs stratified by gender or vaccine type were presented in Supplement tables S1 and S2.

The risks of non-fatal irAEs following the COVID-19 vaccination

In the Cox proportional hazard model in this study, COVID-19 vaccination significantly increased the risks of non-fatal AEs except for visual impairments (HR, 3.935; 95% CI, 0.943-16.410), with the highest level of alopecia (HR, 2.397; 95% CI, 1.896-3.029) followed by inner ear diseases (HR, 2.368; 95% CI, 2.153-2.604), and herpes zoster (HR, 2.337; 95% CI, 2.122-2.573) (Figure 2a). In the multivariate logistic model in this study, the COVID-19 vaccination was associated with a significant increase in the risk of most non-fatal irAEs, indicating potential influence in the early time point (one-week after COVID-19 vaccination). At three months post-vaccination, the COVID-19 vaccination significantly increased the risk of endometriosis (OR, 1.631; 95% CI, 1.313-2.025). Furthermore, visual impairment at three months (OR, 3.956; 95% CI, 0.949-16.496), tinnitus at one-week (OR, 1.840; 95% CI, 0.981-3.451), and two weeks (OR, 1.513; 95% CI, 1.000-2.287), and periodontal diseases at one-week (OR, 0.965; 95% CI, 0.366-2.546), two weeks (OR, 1.771; 95% CI, 0.760-4.126), and one-month (OR, 1.662; 95% CI, 0.955-3.041) showed no statistical differences of ORs between two groups (Figure 2b). The detailed data for risks of non-fatal irAEs following the COVID-19 vaccination were described in Supplement Tables S3 and S4.

• Download figure
• Open in new tab

Figure 2. The risk for non-fatal autoimmune-related adverse events (irAEs) by COVID-19 vaccination.

a. Cox proportional hazard model, which is presented as a forest plot (hazard ratio [red circle] with 95% confidence interval [bar]). If clinical significance, it is marked as red-bar. b. Multivariate logistic regression model along with time point, which presented as forest plot (odd ratio [red circle] with 95% confidence interval [bar]).

The risks of non-fatal irAEs according to the COVID-19 vaccine type

Both the multivariate logistic regression model and the Cox proportional hazard model were used to assess the risk factors according to the COVID-19 vaccine type (Figure 3). In the Cox proportional hazard model, heterologous vaccination was more increased the risks of gynecological problems including endometriosis (HR, 2.784; 95% CI, 2.083-3.722]) and menstrual disorders (HR, 2.837; 95% CI, 2.583-3.117]), hematological problem including bruise (HR, 1.891; 95% CI, 1.208-2.962), dermatological problems including herpes zoster (HR, 2.894; 95% CI, 2.153-2.604) and alopecia (HR, 3.413; 95% CI, 2.520-4.623), ophthalmological problem including glaucoma (HR, 1.828; 95% CI, 1.596-2.093), and periodontal diseases (HR, 3.560; 95% CI, 2.179-5.818) compared to other types of vaccination (Figure 3a). In the multivariate logistic regression model, most trends of risks were shown a similar pattern to that of Cox proportional hazard models. According to time points, alopecia and periodontal diseases were associated with the highest risk outcomes following heterologous vaccinations compared to other vaccination methods. However, vaccination using cDNA only was observed to notably increase risks of bruise at one-week (OR [95% CI] = 5.767 [2.010-16.543]), two weeks (HR, 6.260; 95% CI, 2.280-14.621), and one-month (HR, 4.996; 95% CI, 2.659-9.388]) (Figure 3b). The detailed data was presented in Supplement tables S5 and S6.

• Download figure
• Open in new tab

Figure 3. The risk for non-fatal autoimmune-related adverse events (irAEs) of vaccinated subjects according to vaccine type.

a. Cox proportional hazard model, which is presented as a forest plot (hazard ratio [red circle] with 95% confidence interval [bar]). If clinical significance, it is marked as red-bar. b. Multivariate logistic regression model along with time point was presented as the odd ratio of a heatmap. If clinical significance, it is marked as *.

Strengths and Limitations

Our study has several strengths and potential limitations for non-fatal AEs following COVID-19 vaccination. First, many non-fatal irAEs have been reported as case reports or case series so it is the first study to investigate non-fatal irAEs following COVID-19 vaccination.^{14 15 20 29 31} For diverse manifestations including post-vaccination glaucoma, ear diseases, and alopecia, these irAEs shared potential pathophysiological mechanisms as expression of spike proteins, which dysregulate immunity, lead to molecular mimicry phenomena, and activate the pro-inflammatory cytokines.^{13 15} With the suggested mechanisms in the literature, our studies are consistent with this hypothesis because the types of vaccinations have higher HRs in heterologous or cDNA vaccination compared to mRNA-only vaccination. Second, the current studies have been demonstrated that the COVID-19 vaccine effect on T cell mediated immune response in multiple sclerosis, which lead to autoimmunity.¹⁹ Our findings for the spectrum of non-fatal irAEs strengthens their studies sharing the similar pathophysiological mechanisms and hypothesis such as role of spike proteins and autoimmune diseases triggered by vaccines. Last, the non-fatal irAEs after COVID-19 vaccination are relatively common and can be affected by various factors such as vaccination methods.¹⁹ Furthermore, our findings also suggest that different types of vaccinations exhibit distinct activation patterns at various sites, which will need to be studied in future. Even in case, the warts on cheek was developed with positive for spike IgG and negative for nucleocapsid IgG after mRNA-based vaccination, suggesting that the mRNA-LNP triggers autoimmune response.¹⁸ Therefore, the COVID-19 vaccines may not be fatal, but vaccinated people with predisposition may more susceptible for occurrence of AEs. Furthermore, our study indicate that it is essential to consider potential side effects that persist beyond three months. As the toxicity of COVID-19 diminishes and a significant portion of the worldwide population acquires natural immunity, it is important to designate as vaccine recipients those for whom the benefits of vaccination outweigh the potential side effects for ongoing vaccination.

Conclusions

The three-month risks of incidental non-fatal irAEs are substantially higher in the COVID-19 vaccinated subjects than in non-vaccinated controls. Our findings suggested that vaccinated subjects with predisposition are potentially vulnerable to the occurrence of diverse irAEs although the COVID-19 vaccines may not be fatal. Consequently, clinicians should maintain closed observation of a range of non-fatal irAEs after vaccination, given that these manifestations might emerge post-vaccination.

What is already known on this topic

• Although the COVID-19 vaccines have been shown to have favorable safety and efficacy, they have also been associated with adverse events (AEs) stemming from immune-related responses.

• Much of the current literature on COVID-19 vaccination focused on serious AEs but the various non-fatal immune-related adverse events (irAEs) have yet to be comprehensively elucidated.

What this study adds

• In a population-based cohort study of 1,748,136 participants in Seoul, South Korea, the three-month cumulative incidence rates of non-fatal irAEs were significantly higher in vaccinated subjects than in non-vaccinated subjects. The vaccination significantly increased the risks of non-fatal irAEs following COVID-19 vaccination.

• The findings of this study suggest that clinicians should maintain closed observation of a range of non-fatal irAEs after vaccination, given that these manifestations might emerge post-vaccination.

How this study might affect research, practice or policy

• This population-based study suggested that vaccinated subjects with predisposition are potentially vulnerable to the occurrence of diverse irAEs although the COVID-19 vaccines may not be fatal.