Editorial
March 19, 2024
Postmarketing Vaccine Safety Assessments: Important Work in Progress
Kathryn M. Edwards, Marie R. Griffin
JAMA. 2024;331(11):915-917. doi:10.1001/jama.2023.26630
After the initial randomized clinical trials and emergency use authorization of the COVID-19 vaccines by the US Food and Drug Administration (FDA) during the COVID-19 pandemic, the US Centers for Disease Control and Prevention (CDC) and the FDA undertook extensive postmarketing vaccine safety surveillance activities in the US, with close monitoring by an independent safety committee.1 Analysis of surveillance data identified several important adverse events associated with receipt of the COVID-19 vaccines.
When evolving data provided evidence for a potential causal relationship between vaccine receipt and a specific adverse event, the potential risk associated with the vaccine was quantified, other risk factors for the specific adverse event were identified (eg, age, sex), and the risks vs benefits were assessed.1 On January 13, 2023, the FDA and the CDC issued a joint public communication2 about the identification of a preliminary safety signal within the Vaccine Safety Datalink surveillance system for ischemic stroke within 21 days after receipt of a COVID-19 bivalent mRNA vaccine for individuals 65 years of age and older. This public communication2 also specified that the safety signal indicating a potential increased risk of stroke appeared greater when the COVID-19 bivalent mRNA vaccines (also called boosters) were given concomitantly with either a high-dose or adjuvanted influenza vaccine than when given alone.
To further investigate this safety signal, the FDA conducted an analysis and Lu et al3 report the results in this issue of JAMA. The analysis used data from Medicare beneficiaries aged 65 years or older and assessed the risk of stroke after receipt of either brand of COVID-19 bivalent mRNA vaccine alone or when given concomitantly with a high-dose or adjuvanted influenza vaccine. Lu et al3 used a self-controlled case series design in which individuals act as their own controls. The temporal association between a transient exposure and an event is evaluated and all time-invariant confounding is eliminated.4 Among more than 5 million recipients of either brand of COVID-19 bivalent mRNA vaccine, the study identified 11 001 (0.20%) with a cerebrovascular outcome (nonhemorrhagic stroke, transient ischemic attack, or hemorrhagic stroke). The study found no increased stroke risk associated with either brand of COVID-19 bivalent mRNA vaccine when administered alone. The results from this large cohort are reassuring and are consistent with those reported from France5 and Israel.6 In addition, no safety signal has been issued by the European Medicines Agency.7
Even though these studies provide reassurance of no increased stroke risk associated with either brand of COVID-19 bivalent mRNA vaccine when given alone, they raise the possibility that receipt of a high-dose or adjuvanted influenza vaccine was responsible for the safety signal. Among the COVID-19 booster recipients who had a cerebrovascular event, 34% to 45% had also concomitantly received a high-dose or adjuvanted influenza vaccine. In a secondary analysis of more than 9 million recipients of a high-dose or adjuvanted influenza vaccine, small statistically significant increased risks of cerebrovascular events were observed in several subgroups both when influenza vaccines were used alone and with COVID-19 boosters. However, in the study, the risks observed were inconsistent across the postvaccination risk windows, age groups, and type of cerebrovascular outcome so the findings provide insufficient evidence for a definitive conclusion about causation. If these risks are causal, the attributable risk would be 1 to 3 cerebrovascular outcomes per 100 000 vaccinated patients, emphasizing the remarkable ability of these surveillance systems to detect safety risks of even a small magnitude.
Several studies have addressed the risk associated with administration of a high-dose or adjuvanted influenza vaccine in older adults. A study8 using data from the Vaccine Adverse Event Reporting System (VAERS) documented nearly 12 000 reported adverse events in older adults during the first 8 years after licensure of a high-dose trivalent influenza vaccine in 2009; only 2 of these adverse events were cerebrovascular events. A similar recent study9 documented 2122 reported adverse events among older adults during the first year after licensure of a high-dose quadrivalent influenza vaccine; of these adverse events, 6 were cerebrovascular events. Likewise, a study10 using VAERS data documented 521 reported adverse events in older adults during the 2 years after licensure of an adjuvanted influenza vaccine in 2015; none were cerebrovascular events. Two recent studies11,12 of concomitant administration of COVID-19 and influenza vaccines provide information on safety and immunogenicity, but were too small (evaluating only 700 and 300 participants, respectively) to address rare adverse events.
From a population health perspective, a risk of serious outcomes of 1 per 100 000 vaccinated individuals would be more than balanced by the benefits of most recommended vaccines. For example, influenza virus results in thousands of potentially preventable illnesses, medical care visits, hospitalizations, and deaths among persons aged 65 years or older in the US annually13 and missed days from school and work in younger persons. However, some healthy older adults who are at an extremely low risk of serious influenza complications may not find the individual risk-to-benefit calculus favorable. Given that vaccines are recommended for healthy populations to prevent a possible future event (eg, during routine influenza seasons, 5%-10% of persons are infected with influenza annually), the bar for safety must be extremely high, especially for those who are unlikely to suffer serious complications (such as hospitalization or death) from the vaccine-preventable disease.
How can vaccine safety be monitored to ensure that individuals are well informed about their individual risks? First, standardized, consistent definitions of the adverse events are needed. The Brighton Collaboration14 has steadfastly worked to provide standardized definitions of adverse events with various gradations of certainty that can be effectively used across different studies. Second, large observational studies, such as the one conducted by Lu et al,3 should continue to assess adverse events associated with both existing and newly licensed vaccines.
Third, the FDA has recently endorsed “supporting infrastructure focused on facilitating larger, simpler, and more practical trials that are suitable for generating reliable pre-marketing and post-marketing evidence.”15 For example, a large pragmatic clinical trial was conducted in Denmark to compare the effectiveness of high-dose vs standard-dose influenza vaccine in older adults.16 To more specifically address the remaining questions raised in the study by the Lu et al,3 both methodological approaches could be used to evaluate differences in safety and effectiveness of a high-dose vs an adjuvanted influenza vaccine in older individuals as well as determine whether coadministration of multiple vaccines affects clinically important safety or effectiveness end points.
Public health recommendations generally suggest that multiple vaccines be given concomitantly to facilitate uptake. However, in some populations (such as in frail older adults) and when vaccines may be quite reactogenic, the tolerability and safety of concomitant administration should be evaluated. The COVID-19 vaccine experience has demonstrated that older adults will volunteer for clinical vaccine trials. The use of cluster randomization or other methods that have been used in pragmatic trials may facilitate enrollment of older adults, including those who are frail and vulnerable and who are generally excluded from traditional randomized clinical trials.
To develop safer vaccines, an understanding of the pathogenesis of the adverse events would be ideal. Formal, critical evaluations of data generated in preclinical studies, from animal models, and in human clinical trials might provide insight into the biological basis of adverse events. An emerging focus in the study of vaccines is evaluating whether adverse events after immunization might result from individual differences in innate immune responses, microbiomes, or immunogenetics.17 An example to support to this approach was seen in the outbreak of narcolepsy associated with receipt of the Pandemrix vaccine during the 2009 influenza pandemic; children with a specific human leukocyte antigen type (DQB1*06:02) developed narcolepsy.18 To better understand the pathophysiology of adverse events after COVID-19 vaccination, several consortia have been formed to foster research. The Global Vaccine Data Network19 and the International Network of Special Immunization Services17 have developed protocols to analyze the data from individuals experiencing adverse events to help inform the design of safer vaccines.
It is encouraging that the current US vaccine safety system can identify small vaccine risks on the order of 1 per 100 000. Importantly, public health professionals should be prepared to effectively communicate the level of certainty about potential risks. In addition, when risks of serious adverse events are confirmed, methods to communicate individual as well as population benefits vs risks should be developed. Vaccine manufacturers are required to conduct large randomized clinical trials before receiving a license for new vaccines. The pivotal clinical trials for licensure of COVID-19 and more recent respiratory syncytial virus (RSV) vaccines for adults in the US included 24 000 to 44 000 persons.20 However, even these large trials are unable to detect risks on the order of 1 per 100 000 with precision.
If certain adverse events occur more frequently among recipients of a vaccine vs a placebo, it may be difficult or impossible to establish causality, but their occurrence will generate a safety signal that should prompt follow-up if the vaccines are licensed. For example, in the recent trials of 2 RSV vaccines, several cases of Guillain-Barré syndrome occurred within the 21 days after vaccination. These events were in part responsible for the recommendation by the Advisory Committee on Immunization Practices supporting shared clinical decision-making for RSV vaccination rather than a universal recommendation for all those aged 60 years or older for whom the vaccines were licensed.
Although many clinicians find the explanations required for shared clinical decision-making to be time-consuming and difficult to implement, the current state of knowledge about these effective and potentially lifesaving RSV vaccines requires a nuanced approach. Better ways to communicate small risks that aid decision-making for clinicians and the public are needed. In the meantime, postmarketing studies of RSV vaccines are underway.
The study by Lu et al3 illustrates the value of a timely, well-designed analysis and has provided reassurance about the COVID-19 boosters. Ongoing monitoring of influenza vaccines marketed for older adults will provide additional data on stroke risk. This type of timely and transparent continuous assessment should be used to enhance vaccine safety and assure the public that vaccine safety is a priority of the FDA and the CDC.
