Skip Navigation
  • Text Size: A A A
  • Print
  • Email
  • Facebook
  • Tweet
  • Share
  • Print
  • Email
  • Facebook
  • Tweet
  • Share

The Vaccine Safety Data System

Marie C. McCormick MD, ScD
Sumner and Esther Feldberg Professor of Maternal and Child Health, Department of Social and Behavioral Sciences, Harvard School of Public Health
Professor of Pediatrics, Harvard Medical School
Senior Associate Director of the Infant Follow-up Program, Boston Children’s Hospital

The response to the pandemic of influenza due to H1N1 with a rapidly developed vaccine presented a substantial challenge for monitoring its safety.  This challenge required a major, rapid expansion of surveillance strategies.  Under the aegis of an interagency coordinating committee, surveillance techniques and analyses developed by the Centers for Disease Control and Prevention (CDC) and the Food and Drug Administration (FDA) were extended to data from the Department of Defense and the Department of Veterans Affairs, and the Indian Health Service, and a new network was developed to link vaccine registries with medical records obtained from insurance firms.  The success of this endeavor to provide timely oversight, and eventually, reassurance about the safety of the vaccine rested on the almost two decades of development and methodological investment in monitoring vaccine safety.

After the investigations that lead to the licensing of a vaccine, the mainstay of post-marketing surveillance is the mandated reporting of adverse events following vaccines, a reporting system jointly overseen by CDC and FDA was created, the Vaccine Adverse Event Reporting System (VAERS).  Because it is a passive reporting system, VAERS incurs a number of limitations including variable clinical information, potential underreporting and overreporting, vague descriptions of adverse events, uncertainty about the generalizability of the information, lack of data on concomitant exposures and supportive tests, and, perhaps most importantly, major limitations in assessing the causal relationship between vaccine exposure and the adverse event.  To address these deficiencies, CDC began to explore the use of large linked databases consisting of computerized medical records with information on both vaccines and health events.  This experience led to a partnership between CDC and several health maintenance organizations (HMOs) to create a mechanism for examining adverse medical events following vaccine administration, the Vaccine Safety Datalink, for both descriptive epidemiological studies of vaccine administration and assessments of causal links between vaccines and a variety of adverse outcomes.  Further work led to the development of analytic techniques for surveillance of adverse events rapidly.  

The experience with the H1N1 surveillance highlighted both the potential of the new surveillance techniques as well as some of the challenges of expanding coverage that point to the future.  First, the rapid implementation of analytic capacity in systems with electronic health records (EHRs) (both medical records and registries) points to the potential for broader surveillance with the expansion of EHRs more generally.  In particular, broadening the reach of the systems will permit better understanding of the risk of adverse events in smaller, potentially vulnerable populations. This approach is being developed in a pilot project, the FDA Mini-Sentinel .

However, as the H1N1 experience revealed, including populations that have not previously been covered requires additional methodological work to be assured of the validity of the diagnoses for adverse events related to vaccines rather than reflections of other, co-existing morbidity.  Second, the diagnostic coding system is currently less than optimal for examining vaccine safety on some groups, especially pregnant women, and additional strategies may be needed for this important subgroup, such as that developed for H1N1.  Moreover, timeliness remains an issue.  Although the vast majority of potential adverse reactions to the H1N1 vaccine were rapidly eliminated, assessing the causal connection in the remaining few was not completed until three years after the roll-out of the vaccine.  Improved timeliness of assessment may occur with some of the methodological work needed to refine the precision of detecting adverse events related to vaccines.  However, improved timeliness and accuracy of establishing vaccine-related adverse events will also require laboratory methods to validate the epidemiologic findings, and new characterizations at the genetic or immunologic levels of the risk for adverse events, as outlined in the National Vaccine Plan.