Invisible Reactions: How Passive Pharmacovigilance Infrastructure Erases Vulnerable Populations from Vaccine Safety Data
A Surveillance System Built for the Already-Visible
The United States maintains two primary mechanisms for monitoring vaccine adverse events: the Vaccine Adverse Event Reporting System (VAERS), co-administered by the Centers for Disease Control and Prevention and the Food and Drug Administration, and the Vaccine Safety Datalink (VSD), a collaborative network of integrated health systems. Both instruments have contributed meaningfully to post-licensure safety surveillance. Yet the architecture undergirding these systems reflects a foundational assumption — that the populations most at risk of adverse outcomes are also the populations most capable of navigating complex, English-dominant, digitally mediated reporting pathways. That assumption is epidemiologically untenable.
Passive surveillance systems, by design, depend on voluntary action. A clinician must recognize a potential adverse event, associate it with a prior vaccination, locate the appropriate reporting portal, and complete a detailed submission. A patient or caregiver must similarly identify a potential connection, possess sufficient health literacy to interpret their symptoms as reportable, and access a platform that accommodates their language and technological capacity. For populations occupying the margins of the health care system — including limited-English-proficient individuals, rural residents with constrained provider access, and those managing chronic immunosuppression — each of these steps represents a potential point of failure.
Language as a Structural Barrier in Safety Monitoring
The VAERS online portal, the primary public-facing interface for adverse event submission, is available in English. While translated guidance documents exist in limited form, the submission workflow itself does not offer robust multilingual functionality. For the estimated 25 million people in the United States who speak English less than very well, according to American Community Survey data, this design constraint is not a minor inconvenience — it is a disqualifying barrier.
The implications extend beyond access. When non-English-speaking populations are systematically underrepresented in adverse event databases, the signal detection algorithms applied to those databases will reflect that absence. Disproportionate enrollment of English-proficient, digitally connected, urban-adjacent individuals introduces a selection bias that distorts population-level safety assessments. A safety signal occurring at elevated frequency in a Spanish-speaking immigrant community in South Texas or a Hmong-speaking community in Minnesota's Twin Cities metro area may never accumulate sufficient case volume to trigger algorithmic detection — not because the signal is absent, but because the infrastructure was never designed to receive it.
Researchers examining pharmacovigilance equity have noted that language-concordant provider relationships, where patients and clinicians share a primary language, are associated with higher rates of adverse event disclosure during clinical encounters. Yet even when disclosure occurs, the translation of that clinical interaction into a formal VAERS submission remains inconsistent. Reporting rates among non-specialist providers — including federally qualified health center clinicians who disproportionately serve linguistically diverse populations — lag behind those of academic medical centers with dedicated pharmacovigilance staff.
Rural Geographies and the Surveillance Gap
Geographic isolation compounds the equity problem in distinct ways. Rural populations in the United States face a convergence of factors that suppress adverse event reporting: lower rates of broadband internet access, reduced availability of specialist providers capable of attributing complex symptom presentations to vaccination history, and health care systems that often lack the administrative infrastructure to support systematic reporting workflows.
When a rural resident in West Virginia or eastern Montana experiences a delayed hypersensitivity reaction or an unexpected neurological symptom following vaccination, the clinical pathway to formal adverse event documentation is considerably longer than it would be for a patient embedded in an urban academic health system. The primary care provider may be a nurse practitioner managing a caseload that precludes detailed pharmacovigilance work. The nearest specialist may require a multi-hour drive. The patient may delay or forgo follow-up care entirely. The adverse event, whether or not it is causally attributable to the vaccine, disappears from the surveillance record.
This geographic attrition of safety data is not merely a technical inefficiency. It carries direct implications for equity-stratified safety assessments. If rural populations — who carry elevated burdens of chronic disease, nutritional deficiency, and limited prior health care engagement — respond differently to certain vaccine formulations than their urban counterparts, passive surveillance systems are structurally incapable of detecting that divergence at meaningful statistical power.
Immunocompromised Populations and the Limits of Passive Detection
Individuals living with conditions that alter immune function — including recipients of solid organ transplants, patients undergoing chemotherapy, those with autoimmune conditions managed by immunosuppressive agents, and people living with HIV — represent a clinically distinct population for whom vaccine safety profiles may differ from those observed in immunocompetent trial participants. Regulatory approval of vaccines typically rests on clinical trial data that either excludes or substantially underenrolls immunocompromised individuals, making post-licensure surveillance the primary mechanism for characterizing safety in this group.
Yet immunocompromised patients frequently receive care across fragmented multi-specialty networks, where attribution of adverse events to specific interventions requires coordinated clinical communication that passive reporting systems cannot enforce. A patient receiving immunosuppressive therapy for Crohn's disease who develops an unusual inflammatory response following vaccination may have that symptom managed by a gastroenterologist, assessed by a primary care physician, and documented in two separate electronic health record systems — none of which automatically generate a VAERS submission. The adverse event may be clinically recognized, carefully managed, and completely invisible to national surveillance.
The Absence of Equity Stratification in Safety Databases
Beyond access and reporting barriers, a more fundamental structural deficiency warrants attention: the absence of mandatory, standardized equity-stratified data collection within pharmacovigilance infrastructure. VAERS submissions include demographic fields, but race, ethnicity, and primary language data are inconsistently captured and frequently missing. Without complete demographic metadata, post-hoc equity analyses of adverse event patterns are methodologically compromised before they begin.
The contrast with clinical trial reporting standards is instructive. The FDA has progressively strengthened requirements for demographic subgroup reporting in pre-licensure trial data, reflecting recognition that safety and efficacy profiles may vary across population groups. No analogous mandate governs the demographic completeness of post-licensure adverse event submissions. The result is a surveillance architecture that applies rigorous equity standards to pre-approval data while permitting systematic demographic incompleteness in the post-approval monitoring that most directly affects the populations excluded from trials.
Toward a More Equitable Pharmacovigilance Architecture
Addressing these structural deficiencies requires interventions at multiple levels. Multilingual portal development, with full submission functionality in languages reflecting the linguistic demographics of vaccine-receiving populations, represents a minimum threshold for access equity. Integration of adverse event reporting workflows directly into the electronic health record systems used by federally qualified health centers, Indian Health Service facilities, and rural health clinics would reduce the administrative burden that suppresses reporting in under-resourced settings.
Active surveillance expansion through the Vaccine Safety Datalink and complementary data-linkage mechanisms offers a partial corrective, but only to the extent that the health systems participating in VSD serve demographically representative populations — a condition that does not uniformly hold. Mandatory demographic data completeness standards for VAERS submissions, modeled on existing clinical trial reporting requirements, would provide the equity-stratified foundation necessary for meaningful subgroup analysis.
The integrity of vaccine safety science depends on the integrity of the data that science draws upon. When surveillance infrastructure systematically excludes the populations most structurally vulnerable to gaps in health care engagement, the resulting evidence base does not merely underserve those populations — it actively misrepresents the safety landscape for all. Correcting that misrepresentation is not a secondary concern of equity advocacy. It is a prerequisite for rigorous public health science.
