Clinical Utility of COVID-19 Vaccination in Patients Undergoing Hemodialysis

The coronavirus disease 2019 (COVID-19) pandemic has had a devastating effect on the CKD community, particularly for individuals receiving maintenance dialysis. During the early waves of the pandemic, nearly one in four patients who contracted severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) died from the illness (1). Many patients who survived their critical illness were left with long-term morbidity. This situation was particularly frightening for patients on dialysis with comorbid conditions who were not able to self-isolate because they required hemodialysis treatments in a congregate setting. Dialysis providers instituted strict precautions to prevent infections, but the eventual availability of effective vaccines hopefully offered more robust protection. Leaders of the community were successful advocates for prioritizing vaccination in the dialysis population, and providers facilitated the administration of the vaccine. For example, in Ontario, Canada, 90% of the maintenance dialysis population had received two doses of a COVID-19 vaccine by September 2021, and 78% had received three doses of a vaccine by January 2022 (Ontario Renal Network reporting). However, serology studies demonstrated muted antibody responses in patients on dialysis compared with healthy controls, which raised concerns about clinical efficacy (2). These results were not surprising given the suboptimal responses of patients on dialysis to other vaccines, such as hepatitis B, and the prevalence of immunocompromising conditions. How this suboptimal response would or would not translate into protection against acquiring SARS-CoV-2 infection was unclear, but new research, including the study from Ashby et al. (3) in this issue of CJASN, has emerged to answer this critical question.

Determining the clinical utility of COVID-19 vaccination is complex and depends on many factors, including the population under study; the waves of the pandemic; the circulating variants; the type, number, and timing of vaccination; how the infection was detected; and particularly, the analytic methods used. Vaccine efficacy is determined in clinical trials, whereas vaccine effectiveness (VE) is determined in real-world settings. Studying a large generalizable dialysis population in Ontario, Canada, our group found that two doses of mRNA vaccine reduced the risk of infection by 69% (4). Sibbel et al. (5) studied patients treated at DaVita dialysis centers across the United States and found 79% and 73% reduced infection risks for the BNT162b2 (Pfizer) and the mRNA-1273 (Moderna) vaccines, respectively. We used a cohort design and modeled vaccine receipt as a time-varying covariate. This method allowed inclusion of the entire dialysis population, increasing power, but it also required accounting for competing risks, such as non–COVID-19 death and transplant over time, which were more likely in unvaccinated patients. Cohort studies also require modeling of time-varying covariates, such as vaccine exposure and, ideally, background community rates of infection. Time-varying covariates also limit the use of the Fine and Gray model to account for competing risks, which are common in dialysis studies. Sibbel et al. (5) used a propensity score design to match vaccinated patients with unvaccinated controls on the date of vaccination and by region, but this requires a larger dialysis population because not all vaccinated patients can be matched to controls.

An alternative method to estimate VE is the test-negative design (TND) (6). In its traditional form, this analysis is restricted to individuals presenting for testing with typical symptoms of infection, such as fever, cough, sore throat, and myalgia. Participants are divided into those who test positive for SARS-CoV-2 (cases) and those who test negative (controls). The odds of testing positive are calculated by vaccine status, adjusting for baseline factors and the period of testing. The TND approximates VE estimates from randomized trials because patients are selected on the basis of similar health-seeking behaviors in that they are presenting themselves for testing. It is also simpler because logistic models estimate odds ratios rather than more complex survival models, which have to account for changes over time in vaccination status and competing events. Unfortunately, reliance on symptoms, a cornerstone of this design, is problematic in COVID-19 dialysis studies. First, information on symptoms at the time of SARS-CoV-2 testing is often lacking in large databases (6). Second, we previously demonstrated that 6% of patients on dialysis could not reliably answer symptom questionnaires due to language and cognitive barriers and that 17% had chronic background symptoms, such as cough, dyspnea, and fatigue, that are often indications to test for COVID-19 (7). Third, infection control experts often recommend testing patients on dialysis without any symptoms during surveillance for outbreaks (8). For these reasons, studies may use a pseudo-test-negative design, where all SARS-CoV-2–tested patients are selected without knowledge of symptoms. Butt et al. (9) used this approach and estimated the VE of mRNA COVID-19 vaccines to be 68% in patients on dialysis treated in the US Veterans Administration network. Importantly, 73% of participants had a prior solid organ transplant, which was an exclusion criterion in other studies.

Currently available VE studies demonstrate that COVID-19 vaccines provide moderate protection against contracting SARS-CoV-2 infection, but the VE for preventing severe outcomes is clinically more important for patients on dialysis because their risk of these events is high. Severe outcomes are defined as SARS-CoV-2–related hospitalization or death. Our study estimated that two doses of an mRNA vaccine reduced severe outcomes by 83%—a greater benefit than for infection prevention (4). This estimate of VE combined two elements: protection from acquiring infection and protection from severe outcomes in those who acquire an infection. An alternative approach is to focus on protection from severe outcomes in those who have an infection. In this issue of CJASN, Ashby et al. (3) analyzed the outcomes of 1323 patients who acquired SARS-CoV-2 infections, representing 24% of the 5500 patients on hemodialysis at risk. Patients were tested on a weekly basis or if they were symptomatic or had recent contacts, which may explain their very high rate of infection over a 9-month study (3). The dominant variant during the early part of the study was Alpha (B.1.1.7), and in the later part, it was Delta (B.1.617.2).

A significant number of infections occurred before vaccinations were widely administered, which may have biased the VE because treatment of COVID-19 improved over time from clinical experience and the availability of new treatments (e.g., remdesivir and tocilizumab). Because outcomes were analyzed within 28 days of the index infection, Ashby et al. (3) used a mixed logistic regression model with fixed effects to estimate the benefit rather than survival models, which are more complex. Prior vaccinations with either BNT162b2 or AZD1222 (ChAdOx1 nCoV-19) significantly reduced the risk of severe outcomes. The risk of hospitalization in those infected was reduced by 75%, and the risk of death was reduced by 88%. The need for oxygen and ventilation was also halved. Ashby et al. (3) demonstrated that of the 172 deaths occurring within 28 days, 90% were deemed to be actually caused by COVID-19, which is a unique strength of the study because other studies consider all deaths within 30 days to be COVID-19 related. Sibbel et al. (5) also reported a reduction in severity with infection and showed that vaccination with BNT162b2 reduced the risks of hospitalization and death by 50% and 71%, respectively. For mRNA-1273, the reductions were 32% for hospitalization and 65% for death. Other risk factors for severe outcomes included older age, diabetes, immunosuppression, and infection when the Delta variant was dominant.

Overall, these studies show that COVID-19 vaccination in the maintenance dialysis population provides moderate protection against acquiring SARS-CoV-2 infection but is highly protective against severe outcomes. However, this effectiveness is still substantially less than the efficacy among healthier participants for mRNA vaccines, where the seminal randomized studies reported a VE of 94%–95% for preventing symptomatic infection and 89% for severe infection (10,11).

Although these encouraging results prove the VE of COVID-19 vaccines for patients on dialysis, there is still much work to be done. The COVID-19 pandemic required the CKD community to react quickly and offer vaccination to patients without high-level evidence of benefit, and this was clearly justified. However, we must look back and carefully evaluate the clinical effect of these decisions. The SARS-CoV-2 virus continues to mutate, serology studies show waning, and third and fourth doses of vaccine have been administered to patients. New research is required to determine the VE of COVID-19 vaccines in these new scenarios, including in earlier stages of CKD. Although the COVID-19 pandemic is ever changing, making vaccine studies challenging, it also provides new opportunities to examine VE from many different angles. We should conduct these studies rigorously and expeditiously to bolster the case for prioritizing vaccination in the dialysis population in the past and to improve protection of our patients in the future.

Disclosures

P.G. Blake is a contracted medical lead at Ontario Renal Network, Ontario Health. He has received honoraria from Baxter Global for speaking engagements and serves on the editorial board of American Journal of Nephrology. M.J. Oliver is a contracted medical lead at Ontario Renal Network, Ontario Health. He is the owner of Oliver Medical Management, Inc., which licenses Dialysis Management Analysis and Reporting System software. He has received honoraria for speaking from Baxter Healthcare and participated on advisory boards for Amgen and Janssen. He also reports patents or royalties from Oliver Medical Management, Inc. and is co-owner of a Canadian patent for Dialysis Measurement Analysis and Reporting systems.

Funding

M.J. Oliver received funding from Public Health Agency of Canada grant COVID-19 Immunity Task Force/2122-HQ-000071.

Author Contributions

M.J. Oliver wrote the original draft, and P.G. Blake reviewed and edited the manuscript.