Exploring Temporal and Spatial Characteristics of Serious Adverse Event Reports Following COVID-19 Bivalent Vaccines

Yiming Li, Wei Tao, Yifang Dang, Yong Chen, Cui TaoResearch Square 2025 · DOI

TL;DR

This study provides a comprehensive temporal and spatial analysis of serious adverse events (SAEs) following COVID-19 bivalent vaccinations in the United States. By integrating VAERS reporting data with CDC administration records and using NLP embedding models, the researchers found that while SAEs are rare and their clinical patterns are remarkably consistent across the country, reporting rates fluctuated in alignment with variant waves and regional reporting behaviors.

What Problem Does This Paper Address?

While the safety of monovalent COVID-19 vaccines has been extensively studied, the bivalent boosters—designed to target both the original strain and Omicron subvariants—required dedicated post-licensure surveillance. Most existing studies focus on absolute counts of adverse events over short periods.

This paper addresses the gap in understanding the "when" and "where" of serious adverse events. Specifically, it seeks to determine if SAE patterns vary over time (temporal) or across different U.S. regions (spatial), and whether these reports correlate with the actual volume of vaccines administered. This level of granularity is essential for public health officials to distinguish between genuine safety signals and reporting artifacts.

Methods

The researchers employed a multi-pronged analytical approach:

  • Descriptive Analysis: They categorized VAERS reports by gender, age (six brackets), and manufacturer (Pfizer vs. Moderna).
  • Reporting Rate Calculation: Unlike studies that only look at raw report counts, this team calculated the reporting rate by dividing the number of reported SAEs by the total number of vaccinations administered (sourced from the CDC).
  • Temporal Analysis: They tracked the top 10 reported SAEs weekly from September 2022 to September 2023 to identify peaks and trends.
  • Spatial & Semantic Analysis: To compare regions, they used the BioWordVec_PubMed_MIMICIII embedding model. This NLP technique transformed complex symptom compositions into mathematical vectors, allowing the team to calculate the "semantic similarity" of adverse event profiles across different Standard Federal Regions.

Datasets

The study integrated two primary data sources:

  1. VAERS (Vaccine Adverse Event Reporting System): Used to extract demographic data, vaccination dates, and specific adverse event descriptions (Preferred Terms from MedDRA).
  2. CDC COVID Data Tracker: Used to obtain the denominator—the total number of bivalent vaccine doses distributed and administered across the U.S. during the study period.

Key Insights and Ideas

  • The "Denominator" Effect: The study highlights that peaks in SAE reports often coincided with the rollout of vaccination campaigns or the emergence of new variants (like the XBB omicron variant in early 2023). This suggests that reporting frequency is heavily influenced by public awareness and vaccine uptake rates.
  • Biological Consistency: One of the most striking findings was the 98%+ similarity in SAE profiles across all U.S. regions. Using the BioWordVec model, the researchers showed that the "symptom fingerprint" of a serious reaction in the Northeast is nearly identical to one in the West, suggesting a consistent biological response to the vaccine regardless of geography.
  • Demographic Vulnerability: Serious reports were disproportionately higher in the senior population (65+), who accounted for nearly 55% of all SAEs. This aligns with the higher baseline medical complexity and higher booster uptake in this age group.
  • Regional Variance: While the types of symptoms were similar, the rate of reporting was not. Northern and middle regions (e.g., South Dakota, West Virginia) showed significantly higher reporting rates than the Southeast or Southwest, likely reflecting differences in regional healthcare reporting systems or population density.

Key Results

  • Scale of SAEs: Out of 37,441 total reports, only 1,062 (2.84%) were classified as serious adverse events.
  • Common Symptoms: The most frequently reported SAEs were death (N=250), dyspnea (shortness of breath, N=115), and COVID-19 infection (N=107).
  • Manufacturer Split: Pfizer accounted for 59.42% of serious reports, while Moderna accounted for 40.58%, roughly mirroring their respective market shares.
  • Temporal Peaks: Four distinct peaks in reporting occurred in December 2022, February 2023, March 2023, and April 2023.
  • Geographic Leader: South Dakota (SD) had the highest serious reporting rate at 83.9 per million doses.

Limitations and Future Work

The authors acknowledge several inherent limitations:

  • VAERS Limitations: As a passive surveillance system, VAERS is subject to underreporting, unverified information, and "reporting bias" where media coverage influences the number of reports filed.
  • Data Gaps: Some CDC administration data was unavailable, and the study excluded patients who received "mixed" doses (e.g., a Pfizer primary series followed by a Moderna bivalent booster) due to tracking complexities.
  • Causality: The study identifies associations but cannot definitively prove that the vaccine caused the reported events, particularly in cases like "death" or "COVID-19," which may be coincidental in high-risk populations.

Future Work: The team intends to apply more advanced statistical methodologies to uncover deeper correlations between specific vaccine components and rare conditions like Guillain-Barre syndrome (26 cases identified) or autoimmune hepatitis (4 cases identified).

Summary updated

Abstract

Overall, our research underscores the ongoing need for vigilant post-licensure vaccine monitoring, emphasizing the continuous surveillance and analysis essential for upholding the safety and effectiveness of COVID-19 bivalent vaccines.
BibTeX
@inproceedings{li2025exploring,
  author = {Yiming Li and Wei Tao and Yifang Dang and Yong Chen and Cui Tao},
  title = {{Exploring Temporal and Spatial Characteristics of Serious Adverse Event Reports Following COVID-19 Bivalent Vaccines}},
  booktitle = {Research Square},
  year = {2025},
  doi = {10.21203/rs.3.rs-6096098/v1},
}
Figure