Introduction
Patients with end-stage kidney disease (ESKD) undergoing hemodialysis (HD) have a substantially higher risk of severe complications and death from severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infection than the general population [
1,
2]. This elevated risk is primarily attributable to impaired immune responses, a higher burden of comorbid conditions, and frequent health care encounters that increase the likelihood of exposure to SARS-CoV-2 [
3]. Although the initial two-dose messenger RNA (mRNA) coronavirus disease 2019 (COVID-19) vaccination regimen provided substantial protection, the emergence of the highly transmissible and immune-evasive Omicron variant has raised concerns regarding sustained vaccine effectiveness, particularly among vulnerable populations such as individuals with ESKD [
4].
Previous studies evaluating COVID-19 vaccine effectiveness in HD patients were conducted predominantly during periods of Alpha and Delta variant predominance and demonstrated significant reductions in SARS-CoV-2 infection, hospitalization, and mortality following a two-dose vaccination regimen [
5,
6]. During the Omicron surge, a Canadian study reported additional protection associated with a third vaccine dose [
7]. However, comprehensive population-based evidence regarding the effectiveness of third-dose vaccination during the Omicron-dominant period, as well as comparative effectiveness between different mRNA vaccine types, remains limited.
In this nationwide retrospective cohort study, we compared the clinical outcomes of two-dose vs. three-dose mRNA COVID-19 vaccination among Korean patients undergoing HD during the Omicron-dominant period. We evaluated the risks of SARS-CoV-2 infection, hospitalization, and mortality, and examined outcomes according to the mRNA vaccine product administered as the third dose.
Methods
Study population and design
This retrospective nationwide cohort study included Korean adults aged ≥18 years who had been undergoing HD for at least 3 months. The index date was January 16, 2022, when the Omicron variant became dominant in South Korea [
8]. Patients with ESKD undergoing HD were identified using a combination of 1) International Classification of Diseases, 10th Revision (ICD-10) diagnostic codes for chronic kidney disease (N18, N18.1–N18.4, N18.9, and N19), 2) HD-related procedure codes, and 3) special exemption codes issued by the National Health Insurance Service. Baseline characteristics included age, sex, health insurance status, residential area (metropolitan vs. nonmetropolitan), detailed COVID-19 vaccination status, and comorbid conditions identified using ICD-10 codes recorded within 1 year before the index date (
Supplementary Table 1, available online).
Participants who received only a single vaccine dose, were unvaccinated, or received the Ad26.COV2.S vaccine (Janssen/Johnson & Johnson) were excluded. COVID-19 vaccination status was categorized as receipt of two or three doses of the BNT162b2 (Pfizer-BioNTech) or mRNA-1273 (Moderna) vaccine as of the index date. Patients diagnosed with COVID-19 within 14 days of vaccination at the index date were excluded to allow sufficient time for an adequate immune response. Patients with missing information were excluded when essential data required for cohort construction or exposure classification, such as vaccination date, vaccine product, dose sequence, or key baseline linkage variables, were unavailable. Participant selection and exclusion criteria are shown in
Fig. 1.
Prior SARS-CoV-2 infection was defined as a polymerase chain reaction (PCR)-confirmed SARS-CoV-2 infection recorded in the Korea Disease Control and Prevention Agency–National Health Insurance Service (K-COV-N) database before the index date.
The primary outcome was the first diagnosis of COVID-19 infection confirmed by reverse transcription (RT)-PCR after the index date. Secondary outcomes included COVID-19–related hospitalization and all-cause mortality. Hospitalization was defined as admission within 14 days of a positive RT-PCR test result, and all-cause mortality was defined as death occurring within 30 days of a positive test result being recorded. Censoring was performed when a patient was confirmed to have the study outcomes or reached the study end date (July 9, 2022). Because detailed clinical severity indicators such as pneumonia, respiratory complications, oxygen requirement, and intensive-care admission were not available in the database, hospitalization and mortality were used as clinically relevant severe outcomes available in the nationwide database.
This study was conducted in accordance with the Declaration of Helsinki. The K-COV-N database used in this study was fully anonymized to protect patient privacy. The Institutional Review Board of Hallym University Kangnam Sacred Heart Hospital waived the requirement for informed consent (No. 2022-07-022).
Data sources
This nationwide population-based study used the K-COV-N–linked database, which contains comprehensive nationwide COVID-19 diagnosis data from South Korea. The database integrates information on medical service utilization, government-administered health insurance claims, detailed COVID-19 case data (including date of diagnosis and COVID-19–related death), and fully anonymized patient-level information. Data quality is maintained through standardized government verification procedures and routine audits.
Statistical analysis
Patients were categorized according to vaccination status (two vs. three doses) as of the index date. Continuous variables are presented as mean ± standard deviation, and categorical variables as number (percentage). Categorical variables were compared using the chi-square test or Fisher exact test, as appropriate, and continuous variables were compared using independent-samples t tests. The risk of SARS-CoV-2 infection was evaluated using Cox proportional hazards models adjusted for age, sex, residential area, health insurance status, receipt of subsequent COVID-19 vaccine doses during follow-up, and comorbidity burden, as measured by the Charlson Comorbidity Index (CCI). Receipt of subsequent COVID-19 vaccine doses during follow-up was defined as any additional COVID-19 vaccine dose administered after the index date, including a third dose among baseline two-dose recipients or a fourth dose among baseline three-dose recipients. Results are presented as adjusted hazard ratios (aHRs) with corresponding 95% confidence intervals (CIs).
Cox proportional hazards models were also used to assess the risks of all-cause mortality and COVID-19–related hospitalization according to vaccination status among patients with SARS-CoV-2 infection. These analyses were interpreted as conditional analyses among patients with confirmed infection. Multivariable analyses for hospitalization and mortality were adjusted for age, sex, residential area, health insurance status, receipt of subsequent COVID-19 vaccine doses during follow-up, and CCI. Prior infection was not included as a separate covariate in the original multivariable Cox models because none of the participants with prior infection became infected with COVID-19 during the study period. Kaplan-Meier curves were generated to describe the overall time-to-event patterns for SARS-CoV-2 infection and COVID-19 mortality according to vaccination status, and between-group differences were assessed using the log-rank test. All statistical analyses were performed using R software (version 4.0.5; R Foundation for Statistical Computing), and a two-sided p-value <0.05 was considered statistically significant.
Discussion
This study demonstrated that receipt of a third mRNA COVID-19 vaccine dose was associated with significantly lower risks of SARS-CoV-2 infection, COVID-19–related hospitalization, and death among patients undergoing HD. Third-dose vaccination was associated with a 42% reduction in the risk of SARS-CoV-2 infection, a 35% reduction in COVID-19–related hospitalization, and a 65% reduction in 30-day mortality following COVID-19 diagnosis. In addition, receipt of mRNA-1273 as the third dose was associated with lower observed risks of infection and hospitalization than receipt of BNT162b2. However, this vaccine type comparison should be interpreted cautiously because of heterogeneous prior vaccine sequences, imbalance in group size, possible center-level vaccine allocation, and residual selection bias. These findings support the clinical importance of third-dose vaccination strategies in reducing severe COVID-19 outcomes in the HD population while underscoring the need for cautious interpretation of observational vaccine platform comparisons.
A previous meta-analysis demonstrated that patients undergoing HD have a lower overall immunogenic response to COVID-19 vaccination after the first and second doses compared with individuals not receiving HD [
9]. Patients on HD exhibit reduced seroconversion rates to anti-spike protein receptor-binding domain (RBD) antibodies, as well as diminished neutralizing antibody activity and T-cell responses following vaccination, relative to the general population [
9–
11]. However, accumulating evidence indicates that a third vaccine dose substantially enhances both humoral and cellular immunity. After administration of a third dose, anti-spike immunoglobulin G antibody levels increase markedly in patients undergoing HD, reaching levels comparable to those observed in healthy individuals [
12]. This immunologic enhancement is particularly important given that antibody titers decline more rapidly over time in patients undergoing HD, thereby increasing susceptibility to SARS-CoV-2 infection. In addition, a third vaccine dose has been shown to induce a significant expansion of RBD-specific memory B cells in SARS-CoV-2–naïve patients undergoing HD, suggesting the development of more robust and durable immune memory [
13].
Beyond immunologic markers, a third dose of the COVID-19 vaccine has been associated with meaningful clinical benefits. In a study of patients undergoing HD during the Omicron-dominant period, receipt of a third vaccine dose was associated with a 65% lower risk of all-cause mortality compared with receipt of only two doses [
14]. In that study, higher circulating anti-spike protein RBD antibody levels were independently associated with a lower risk of all-cause mortality. Similarly, a Canadian cohort study of maintenance HD patients reported that three-dose mRNA COVID-19 vaccination was associated with lower incidence rates of SARS-CoV-2 infection and severe COVID-19–related outcomes during the Omicron period compared with two-dose vaccination [
7].
Although a third vaccine dose confers substantial benefits, certain limitations have been reported. Affeldt et al. [
15] showed that neutralizing activity against newer variants, such as Omicron BA.1, remains reduced even after a third dose, suggesting a potential need for variant-adapted vaccines. Although the timeframe of the present study coincided with the dominance of the Omicron BA.1 and BA.2 subvariants in South Korea, our findings are consistent with those of previous studies. Importantly, this large nationwide analysis provides robust evidence that third-dose vaccination offers substantial protection against COVID-19 in vulnerable populations, including patients undergoing HD.
In patients undergoing HD, both the BNT162b2 and mRNA-1273 COVID-19 vaccines elicit protective immune responses; however, prior studies have demonstrated differences in the magnitude and durability of these responses between the two vaccines. In a comparative study of vaccine immunogenicity among HD patients, those who received mRNA-1273 had significantly higher median anti-spike antibody titers (1,507 binding antibody units [BAU]/mL) than those who received BNT162b2 (676 BAU/mL), even after adjustment for patient characteristics [
16]. Another study reported that 6–7 weeks after vaccination, 95% of patients who received mRNA-1273 achieved antibody levels comparable to those observed in convalescent serum, whereas only 73% of BNT162b2 recipients reached similar levels. At 12 weeks, the durability of the antibody response remained greater in the mRNA-1273 group [
17]. Consistent with these immunogenicity findings, Alkadi et al. [
18] reported slightly higher vaccine effectiveness for mRNA-1273 than for BNT162b2 among patients undergoing HD, with effectiveness rates of 98.2% and 94.3%, respectively, in preventing SARS-CoV-2 infection more than 14 days after the second dose. In the present study, receipt of mRNA-1273 as the third dose was associated with lower observed risks of SARS-CoV-2 infection and COVID-19–related hospitalization compared with BNT162b2. However, given the substantial imbalance in the number of recipients between vaccine types and the heterogeneity of prior vaccination schedules, the observed association cannot be attributed solely to the third-dose vaccine platform itself. Therefore, this vaccine product comparison should be interpreted as an exploratory, model-based observational comparison rather than as evidence of causal superiority of one mRNA vaccine product over another. Further studies are warranted to confirm differences in effectiveness according to mRNA vaccine type.
This study has several limitations inherent to its retrospective observational design. First, although multivariable models were used to adjust for available covariates, residual bias may remain if the likelihood of SARS-CoV-2 testing differed according to vaccination status. Second, important factors such as detailed clinical information, socioeconomic status, functional status, dialysis-unit practices, and laboratory values were not available in the database, which may have resulted in residual confounding. Third, vaccination status was defined at the index date; therefore, this analysis could not fully account for changes in vaccination status or waning immunity during follow-up. The timing of vaccination and the duration of follow-up differed between the two-dose and three-dose groups. Accordingly, the observed associations may partly reflect differences in time since the most recent vaccination, waning immunity, follow-up duration, and vaccination behavior rather than the isolated effect of the third dose itself. Formal subgroup analyses according to time since vaccination or individual comorbid conditions were not performed, and residual confounding related to vaccination timing, comorbidity profiles, and unmeasured clinical factors cannot be excluded. Therefore, these findings should be interpreted as observational associations rather than definitive causal estimates. Fourth, although prior SARS-CoV-2 infection was uncommon in the overall cohort, it was more frequent in the two-dose group than in the three-dose group. Because infection-induced or hybrid immunity may influence subsequent infection risk and clinical outcomes, residual confounding related to prior infection cannot be completely excluded. Fifth, healthy-vaccinee bias may have influenced the estimated benefit associated with third-dose vaccination, because patients with better overall health status may have been more likely to receive additional vaccination. Sixth, hospitalization and mortality analyses among infected patients should be interpreted as conditional analyses after confirmed SARS-CoV-2 infection. Because vaccination affects the risk of infection, conditioning on infection may introduce selection bias and may not estimate the overall effect of vaccination on severe outcomes in the full cohort. In addition, hospitalization and mortality in patients undergoing HD may be influenced by causes other than COVID-19, and detailed indicators of COVID-19 severity, such as pneumonia, respiratory complications, oxygen requirement, and intensive care unit admission, were not available. Seventh, comparisons between BNT162b2 and mRNA-1273 should be interpreted with caution because they were based on an observational, model-adjusted analysis and were subject to substantial differences in group sizes, heterogeneous prior vaccination histories, and potential selection bias arising from vaccine availability and center-level vaccination practices. Eighth, patients with missing information were excluded, and this exclusion may have introduced selection bias if excluded patients differed systematically from those included in the final cohort. Finally, the present study focused on clinical effectiveness and did not evaluate vaccine safety. Detailed adverse event information, including local or systemic reactions and patient-reported side effects, was not available in the K-COV-N database. Therefore, the absence of safety data should be considered when interpreting the clinical implications of booster vaccination in patients undergoing HD.
In conclusion, this large nationwide study of a representative cohort of patients undergoing HD in South Korea showed that receipt of a third mRNA COVID-19 vaccine dose was associated with lower risks of SARS-CoV-2 infection and COVID-19-related mortality during the Omicron-dominant period. Given the substantial disease burden of COVID-19 among patients undergoing HD, future studies should continue to evaluate vaccine effectiveness and safety against emerging SARS-CoV-2 variants using designs that can better account for vaccination timing, prior infection, and time-varying exposure. Improved understanding of the public health benefits and limitations of vaccination is essential to inform evidence-based recommendations for individuals at increased risk of severe clinical outcomes.