Robert F. Kennedy, Jr., chairman and chief legal counsel for Children’s Health Defense, is asking Dr. David Kessler, one of the newly named co-chairs of President-Elect Biden’s Transition COVID-19 Advisory Board, to consider the long-overdue need for a comprehensive, high-integrity system to monitor adverse outcomes following vaccination.
In a letter this week addressed to Kessler, Kennedy wrote:
“ … [as] former FDA Commissioner, a professor with triple expertise in pediatrics, epidemiology, and biostatistics—and as someone who has repeatedly called attention to the structural deficiencies of our nation’s drug adverse event reporting systems — you are uniquely qualified to articulate the system’s existing shortcomings and recommend badly needed improvements.
COVID-19 vaccines are intended for more Americans than any prior vaccine, Kennedy wrote. As a result, shortcomings in the existing vaccine injury surveillance system will be even more obvious.
Regulatory officials can’t count on post-marketing surveillance to reveal COVID-19 vaccine injuries because the Vaccine Adverse Event Reporting System (VAERS) is broken, Kennedy said.
Read the letter here:
To: David Kessler, MD, JD
Date: December 18, 2020
Re: Post-Marketing Surveillance of COVID-19 vaccines
Dear Dr. Kessler,
I am writing on behalf of Children’s Health Defense to request that you consider the long-overdue need for a comprehensive, high-integrity system to monitor adverse outcomes following vaccination. This request has urgent significance in light of this month’s FDA decision to award Emergency Use Authorization (EUA) to the Pfizer COVID-19 vaccine and FDA’s imminent issuance of EUA for Moderna’s COVID-19 vaccine following abbreviated pre-approval clinical trials. As one of the newly named co-chairs of President-Elect Biden’s Transition COVID-19 Advisory Board, a former FDA Commissioner, a professor with triple expertise in pediatrics, epidemiology, and biostatistics—and as someone who has repeatedly called attention to the structural deficiencies of our nation’s drug adverse event reporting systems—you are uniquely qualified to articulate the system’s existing shortcomings and recommend badly needed improvements.
1. Regulatory officials cannot count on post-marketing surveillance to reveal COVID-19 vaccine injuries because VAERS is broken.
Since vaccine companies are immune from liability for injuries caused by their products, and therefore have no incentive to make them safe, our public health regulators have an amplified responsibility to monitor adverse events. There is widespread agreement, even among physicians, that vaccine adverse events—especially those that are relatively rare—can be difficult to detect pre-licensure. COVID vaccine advocates argue that post-licensing surveillance will remedy the deficiencies in the pre-licensing clinical trial data collection system. As you know, this is not true. The Vaccine Adverse Event Reporting System (VAERS), a voluntary reporting system co-administered by FDA and CDC since 1990, is a well-documented public failure.
You have said yourself that the vaccine injury reports received by FDA “represent only a fraction of the serious adverse events” that occur. For twenty years, thoughtful public health advocates—including the United States Congress—have reiterated your concerns. A 2010 federal study commissioned by HHS and performed by Harvard consultants on behalf of the Agency for Healthcare Research and Quality (AHRQ) found that “fewer than 1% of vaccine adverse events” are ever reported to VAERS.
Back in 1993, after recognizing that adverse event reports “are essential to ensure the safety of drugs, biologicals, medical devices, and other products regulated by the FDA once they are introduced into the US market,” and acknowledging that even “large, well-designed clinical trials . . . cannot uncover every problem that can come to light once a product is widely used,” you introduced an “overhaul” of the system for reporting problems with drugs and medical devices, called MEDWatch. However, writing about the new system in the Journal of the American Medical Association (JAMA), you noted that MEDWatch would not affect VAERS nor attempt to remedy underreporting of adverse events associated with vaccines. At the time, even vaccine manufacturers were estimating (confidentially) “a fifty-fold underreporting of [vaccine] adverse events.”
Since then, critics have called the government’s attention to the shortcomings of VAERS time and time again., Although the reporting barriers that you outlined in 1993 addressed drugs and medical devices only, those barriers apply as much or more to vaccines. Obstacles include health professionals’ (and vaccine recipients’) inability to recognize (or even suspect) a possible relationship between a given vaccine and the adverse outcomes; the powerful disincentives for physicians to acknowledge or report vaccine injuries; uncertainty surrounding which adverse events ought to be reported; a medical “culture” that has not made adverse event reporting an “ingrained practice”; and logistics barriers such as cumbersome forms or clinical time constraints. Some of the provider uncertainty about VAERS reporting is likely related to HHS’s spartan Vaccine Injury Table. Notwithstanding the CDC’s addition of a large number of vaccines to the childhood and adult vaccine schedules since 1986 (following passage of the National Childhood Vaccine Injury Act), HHS has made almost no revisions to the table, largely limiting table injuries to a small number of acute reactions occurring within the first 48 hours post-vaccination. In contrast, pre-licensing trials and post-marketing reports have produced documentation (in vaccine package inserts) of nearly 400 different types of adverse events following vaccination.
Another important VAERS limitation is that the system does not routinely follow up on the reports it receives by contacting the reporting health professional, six to 12 months later, to determine if the individual fully recovered from the reaction or further deteriorated. Without this longitudinal information, neither the CDC nor doctors can have accurate or meaningful safety data on vaccines. The Vaccine Injury Table again illustrates the fallout; despite numerous published reports linking vaccines to a wide range of autoimmune conditions—and despite latency periods for autoimmune reactions known to range from weeks to years—the only autoimmune-like conditions listed in the table are chronic arthritis (in association with rubella-containing vaccines), thrombocytopenic purpura (measles-containing vaccines), and Guillain-Barré syndrome (seasonal influenza vaccines).
2. The Department of Health and Human Services has ignored numerous opportunities to strengthen VAERS.
The 2010 AHRQ study—which analyzed several years of data (2006-2009) on 1.4 million doses of 45 different vaccines administered to over 376,000 individuals—found at least one adverse event (AE) for every 39 vaccines given, representing a shocking injury rate of approximately 2.6%. Before seeing these data, HHS had purposed to develop an effective “AI” or “machine counting” system to replace the “designed to fail” VAERS system; AHRQ tested its pilot AI system, called ESP-VAERS (Electronic Support for Public Health-Vaccine Adverse Event Reporting System), on the Harvard Pilgrim health maintenance organization (HMO). However, when CDC saw the alarming levels of vaccine injury exposed by AHRQ’s system, CDC killed the pilot study and refused to take phone calls from the AHRQ consultants. As discreetly stated in the final report, “Unfortunately, there was never an opportunity to perform system performance assessments because the necessary CDC contacts were no longer available and the CDC consultants responsible for receiving data were no longer responsive to our multiple requests to proceed with testing and evaluation.”
One of the AHRQ consultants, affiliated with both Harvard Pilgrim Health Care and Brigham and Women’s Hospital, went on to co-author a 2015 study that put ESP-VAERS to the test in a large health network in Ohio (a system with approximately one million patient encounters annually) between 2012 and 2013. The goal was “to facilitate automated AE detection and reporting” using electronic health records (EHRs), prospectively monitoring the records for “new diagnoses, changes in laboratory values and new allergies for up to 6 weeks following vaccinations” and sending messages to clinicians when the system flagged “suggestive events.” The system also automatically reported “high probability AEs following vaccination” to VAERS, even if the clinician chose to take no action. Over eight months and almost 92,000 vaccinations, the ESP-VAERS system generated from zero to eight alerts per clinician per month (mean=0.4) and captured adverse outcomes including seizures, rashes and other allergic reactions, Bell’s palsy, pleural effusion, lymphocytopenia, and hypothyroidism. At the study’s conclusion, the researchers reported that “The odds of a VAERS report submission during the pilot period were 30.2 . . . times greater than the odds during the comparable pre-pilot period.”
A similar “proof-of-concept” study focusing on use of EHR-based systems to automatically send electronic reports of adverse drug events to the FDA “in real time,” carried out at Brigham and Women’s Hospital in 2008-2009 and published in 2010, found that such systems are not only “efficient and acceptable to clinicians,” but can provide detailed clinical information and have “the potential to greatly increase the number and quality of spontaneous reports submitted to the FDA.” As an added bonus, the researchers found that it took clinicians, on average, 53 seconds to complete and submit the requisite form.
In 1990, an FDA-commissioned study in Rhode Island showed that even a simple intervention to educate physicians about reporting adverse drug reactions resulted in a 17-fold increase in submitted reports compared to the prior year, including significant increases in the number of severe reactions reported.
3. Although new post-licensure surveillance efforts have emerged over the past decade, they, too, have extensive limitations.
In 2019, researchers from the CDC’s Immunization Safety Office published a lengthy commentary in Expert Review of Vaccines examining “challenges in evaluating post-licensure vaccine safety” and reiterating the need for more “real-time” surveillance of vaccine safety concerns. Despite the longstanding critiques of both VAERS and the CDC-led Vaccine Safety Datalink (VSD), the authors put forth VAERS, VSD and the clinician-oriented Clinical Immunization Safety Assessment (CISA) project—established by CDC in 2001 to address “unmet vaccine safety clinical research needs”—as an adequate and suitably “harmonized and supportive approach” for studying post-vaccination adverse events.
Somewhat surprisingly, the 2019 commentary made no mention of a system called PRISM (Post-Licensure Rapid Immunization Safety Monitoring), launched a decade earlier as a demonstration project to monitor the safety of H1N1 influenza vaccines. After the 2009 rollout, FDA folded PRISM—which draws on data from national health insurance plans and state and local immunization registries—into its “Mini-Sentinel” project (with Harvard Pilgrim Health Care serving as the government’s lead collaborator) and then, in 2016, into the Sentinel System, now billed as the world’s largest “multisite distributed database . . . dedicated to medical product safety.” In 2017, FDA also initiated the Biologics Effectiveness and Safety (BEST) system as part of Sentinel, encompassing not just vaccines but other biologics, to fulfill the 2007 FDA Amendments Act’s “requirements for an active postmarket risk and analysis system covering at least 100 million persons.”
PRISM’s proponents praise the system as a tool for supplementing “preexisting and other vaccine safety monitoring systems” (such as VSD) and strengthening the federal vaccine safety “enterprise.” However, a November 2020 Congressional Research Service report asserts that while PRISM and VSD both permit similar types of “population-level scientific analyses of adverse events following vaccination,” VSD is more versatile due to its ability to conduct “more rapid analyses than PRISM.” The PRISM/Sentinel studies published to date also suggest that, although rates of adverse outcomes based on PRISM’s algorithms are “generally higher than rates described in the literature,” the system’s allowable scope of investigation may be relatively narrow. First, notwithstanding the 400 different types of vaccine-related adverse events documented in package inserts, PRISM appears intended to focus on a subset of just 25 outcomes deemed “relevant to vaccine safety monitoring.” Second, published PRISM/Sentinel studies reflect methodological decisions about risk intervals and other variables that could easily constrain researchers’ ability to detect outcomes of interest. For example, a 2018 study of the notoriously problematic quadrivalent human papillomavirus (HPV) vaccine looked at adverse events occurring less than two months (56 days) after the first dose—using this evidence to declare a stellar safety record—even though case reports have flagged fatal reactions following the second and third doses. Third, PRISM/Sentinel studies have largely been authored by the same small pool of insiders, raising questions about whether PRISM, like the VSD before it, will be largely impenetrable to independent investigators. Describing the VSD in 2005, the Institute of Medicine admitted that external researchers confronted a “limited ability . . . to conduct high-quality corroboration studies or studies of new hypotheses,” including studies that “members of the public consider to have high priority.”
4. Because COVID-19 vaccines are intended for more Americans than any prior vaccine, COVID-19 vaccination promises to bring existing surveillance systems’ shortcomings into even sharper relief.
An October commentary in JAMA emphasized that adverse event monitoring of COVID-19 vaccines will go hand in glove with tracking of vaccination coverage. Highlighting the anticipated reliance on VAERS and VSD (as well as on data from the Centers for Medicare & Medicaid Services and the Veterans Administration), the authors stated: “[S]everal large systems are developing capabilities for enhanced passive surveillance” by “leverag[ing] systems designed to trace COVID-19 vaccine administration”—including databases shepherded by the National Healthcare Safety Network, the Department of Defense, and the Indian Health Service—and “pair[ing] them with standardized reporting mechanisms for potential adverse events, such as VAERS.” The commentary’s authors were mute on the topic of PRISM, despite the lead author’s involvement with the majority of published PRISM/Sentinel studies and her prior stint as Associate Director of the Mini-Sentinel Project. Others writing about post-approval safety surveillance of COVID-19 vaccines have also declined to mention PRISM., Although PRISM and Sentinel appear in the Congressional Research Service’s November 2020 overview of postmarketing surveillance, the report indicates that PRISM is less well suited to “real-time” analyses than VSD.
Over the past three decades, the National Vaccine Injury Compensation Program (NVICP) has awarded over $4.4 billion for vaccine injuries and deaths. Peer-reviewed studies show that many of these grave injuries have long diagnostic horizons and chronic consequences poorly captured by pre-licensing studies and flawed post-licensure surveillance systems. Relying on VAERS underreporting, among other factors, public health officials continue to make the improbable argument that “bona fide vaccine injuries are rare.”
Despite the evidence that there are feasible and acceptable ways to strengthen vaccine adverse event reporting, HHS still does not appear to have the technology and systems in place to collect and efficiently process the high volume of COVID-19 vaccine adverse reactions expected by European and other regulators. October’s JAMA commentary flagged likely vaccine reactions that include “allergic, inflammatory, and immune-mediated reactions, such as anaphylaxis, Guillain-Barré syndrome, transverse myelitis, myocarditis/pericarditis, vaccine-associated enhanced respiratory disease, and multisystem inflammatory syndrome in children.” The commentary’s authors also pointed out the surveillance challenges of assessing variations in reactions by type of vaccine (such as mRNA versus viral vector), by population subgroup (such as older adults, pregnant women, or children), or by latency period.
As you aptly noted almost three decades ago, lack of reporting delays problem detection. CDC researchers agree that surveillance systems that “detect adverse events as soon as possible after the introduction of new vaccines” are essential to “speed the recognition of increased rates of adverse events” and “inform public policy.” VAERS’s 30-year-old deficits and the nontransparency of systems like the Vaccine Safety Datalink and PRISM/Sentinel are likely to hinder HHS’s ability to rapidly identify emerging safety issues with COVID-19 vaccines. These lacunae represent, therefore, a direct threat to patient life and public health. Your committee should not allow further rollout of COVID-19 vaccines until FDA’s capacities for monitoring long-term vaccine safety are significantly improved.
Robert F. Kennedy, Jr., Chairman
Children’s Health Defense