The complicated task of monitoring vaccine safety

Abstract

Vaccination is an essential component of modern public health programs and is among our most cost-effective medical interventions. Yet despite vaccines' clear effectiveness in reducing risks of diseases that previously attacked large proportions of the population, caused many deaths, and left many people with permanent disabilities, current vaccination policies are not without controversy. Vaccines, like all other pharmaceutical products, are not entirely risk-free; while most known side effects are minor and self-limited, some vaccines have been associated with very rare but serious adverse effects. Because such rare effects are often not evident until vaccines come into widespread use, the Federal government maintains ongoing surveillance programs to monitor vaccine safety. The interpretation of data from such programs is complex and is associated with substantial uncertainty. A continual effort to monitor these data effectively and to develop more precise ways of assessing risks of vaccines is necessary to ensure public confidence in immunization programs.

Full text

The Complicated Task of Monitoring Vaccine Safety
Susan S. Ellenberg, Ph.D. and Robert T. Chen, M.D., M.A.
SYNOPSIS
Vaccination is an essential component of modern public health programs, and is among our
most cost-effective medical interventions. Yet despite their clear effectiveness in reducing risks
of diseases that previously attacked large proportions of the population, caused many deaths,
and left many people with permanent disabilities, current vaccination policies are not without
controversy. Vaccines, like all other pharmaceutical products, are not entirely risk-free; while
most known side effects are minor and self-limited, some vaccines have been associated with
very rare but serious adverse effects. Because such rare effects are often not evident until
vaccines come into widespread use, the Federal government maintains ongoing surveillance
programs to monitor vaccine safety. The interpretation of data from such programs is complex
and is associated with substantial uncertainty. A continual effort to monitor these data
effectively and to develop more precise ways of assessing risks of vaccines is necessary to
ensure public confidence in immunization programs.
Text
At the beginning of the 20th century, the most serious threat to human life and well-being was
infectious disease. Outbreaks of diseases such as diphtheria and pertussis were common, and
mortality was significant; 160 of every 1000 children born at the turn of the century died of an
infectious disease before the age of 5. Today parents in most developed countries no longer
fear these diseases; a small group of parents, however, question whether the very vaccines
that have prevented so much morbidity and mortality now cause more harm than the benefits
they provide.
Vaccination is surely among the most significant public health interventions of all time.
Nevertheless, the concept of vaccination--essentially, the introduction of foreign material into
healthy individuals--has provoked controversy at times. Pasteur's first administration of rabies
vaccines to humans was strongly protested by physicians and the public, and efforts to
immunize British troops against typhoid at the turn of the century were bitterly opposed despite
the encouraging results of earlier immunization efforts and the serious risk of typhoid faced by
troops serving in the Boer War.
As the effectiveness of vaccines against diphtheria, pertussis, tetanus, typhoid, rabies, and
other diseases became incontrovertible and vaccines came into more widespread use, the
incidence of these diseases rapidly declined. In addition, advances in vaccine technology
brought substantial reductions in the incidence and severity of side effects associated with
rabies and typhoid vaccines, and reduced the risk of manufacturing problems such as
contamination or incomplete inactivation of virus. With the clear evidence of both protection
from disease and high levels of safety came increased acceptance of vaccination as a
valuable safeguard of individual and public health. Support for routine vaccination was no
longer controversial in the medical community by the mid 20th century.
Pa
g
e 1 of 14Date: 10/31/96

Public concerns waned as well. By the mid-1950s, especially after the Salk polio vaccine
became available, the benefit to any individual child of being vaccinated was clear: eliminating
the possibility of contracting disease far outweighed the tiny risk of a serious adverse side
effect from the vaccine itself. Today, in developed countries, diseases such as pertussis,
diphtheria, measles and rubella are rare; polio has essentially been eradicated from the
Western Hemisphere.
The success of vaccination in reducing the risk of disease has led in recent years to renewed
public interest and concern about vaccine safety; in the context of minimal risks of contracting
vaccine-preventable diseases, the risks of side effects, especially the very small chance of
serious adverse effects, take on greater weight and need to be continually reevaluated.
A dramatic example of the need to balance adverse outcomes of vaccination against
protection from disease is the case of polio. In 1996 it is almost a certainty that no child born in
America will contract wild-virus polio, yet eight to ten people each year will develop paralytic
polio as a result of vaccination with (or contact with someone vaccinated with) oral polio
vaccine. This situation has led to recommendations for replacing some of the oral polio
immunizations with inactivated vaccine, which will reduce the risk of vaccine-induced paralytic
polio.
In the late 1970s and early 1980s substantial public attention in the United States as well as in
other developed countries was given to the safety of whole-cell pertussis vaccines. A few
parents who believed their children had been seriously injured as a result of vaccination
brought their concerns to the public through the media. Vaccine coverage plummeted, with the
consequent return of epidemic pertussis disease in Japan, the United Kingdom, and Sweden.
,,
In the United States, while public acceptance of pertussis vaccine generally remained high,
numerous lawsuits were filed against vaccine manufacturers. This resulted in major increases
in prices and decisions by several companies to discontinue manufacture of pertussis
vaccines, resulting in temporary shortages. These events contributed to the passage of the
National Childhood Vaccine Injury Act (NCVIA) in 1986. The NCVIA mandated important new
initiatives relating to vaccine safety; it established the National Vaccine Injury Compensation
Program, provided for an independent review of the available scientific evidence on adverse
events attributable to vaccination,
,
and mandated physician reporting of certain vaccine-
associated adverse events to the Secretary of the Department of Health and Human Services.
The Act also created a unified national system to help identify rare vaccine reactions. This
system, initiated in 1990 and jointly managed by the Food and Drug Administration (FDA) and
the Centers for Disease Control and Prevention (CDC), is called the Vaccine Adverse Event
Reporting System (VAERS). VAERS receives reports of adverse events following vaccination
from vaccine manufacturers, private practitioners, state and local public health clinics, and
vaccinees themselves (or their parents or guardians). It is similar in intent and operation to
surveillance systems for other types of pharmaceutical products, such as the MedWatch
system maintained by the FDA, and to safety surveillance programs in other countries. Such
systems are essential to the discovery of potential rare adverse consequences of
pharmaceutical products that may not become evident until millions of people have been
exposed to these products. But these surveillance systems have important limitations that
complicate the interpretation of the data they accumulate.
The remainder of this paper will address the ways in which vaccine risks are assessed, some
specific issues that have been raised by concerned members of the public, and the potential of
Pa
g
e 2 of 14Date: 10/31/96

VAERS and other vaccine safety surveillance efforts to address these and other important
issues relating to vaccine safety.
Safety Assessment during New Vaccine Development
Like other pharmaceuticals, vaccines must go through extensive clinical testing before they are
marketed. Yet unlike other pharmaceutical products, vaccines are primarily targeted to healthy
people, especially children; virtually every child will receive some vaccinations. Because
healthy people are less willing to accept risk than people who need treatment for illness and
because society is unwilling to impose unnecessary risks on healthy infants and children,
vaccine developers must be particularly sensitive to the risks of adverse effects.
New vaccines are initially tested in small groups of adult volunteers following a series of
laboratory and animal tests that establish, to the extent possible, the safety of delivery to
humans. These adult studies are followed by larger but still preliminary studies in children. In
all studies at the early stage of vaccine development, subjects are monitored intensively for
any potential adverse effects. Rates of common expected reactions, such as swelling and
fever, can begin to be estimated at this stage.
Manufacturing processes are also reviewed carefully by regulators to ensure appropriate
conformance with good manufacturing practices designed to assure consistency, prevent
errors, and avoid contamination with unwanted substances.
When preliminary studies indicate that the vaccine is both safe and producing the desired
immune responses associated with disease protection, larger-scale clinical trials, usually
randomized and placebo-controlled, are undertaken to provide definitive estimates of
protective efficacy and more precise estimates of rates of the more common adverse effects.
The control groups in these studies who do not receive the investigational vaccine are critical
in distinguishing between vaccine-induced effects and those unrelated to the vaccine but
occurring spontaneously in the population studied.
How Safe Is Safe?
The total number of people who have been exposed to a new vaccine by the time it is put on
the market ranges from several hundred to tens of thousands, depending on the intended use
of the vaccine. Even the largest of these pre-marketing studies, however, are inadequate to
assess the vaccine's potential to induce infrequent but serious reactions. With approximately
four million births a year in the United States, an adverse outcome occurring at a rate of one in
10,000 vaccinations--a rate far too low to be detected in conventionally sized clinical studies
performed prior to marketing--would affect 400 babies each year.
Despite the difficulty of assessing these infrequent reactions, the virtually universal exposure o
f
the population to vaccines makes it vitally important to understand even the very rare
complications of vaccination. For this reason, it is essential to continue to collect information on
vaccine-related adverse events even after the vaccine is approved for general use by the FDA.
Advent of a Unified Surveillance S
y
stem
Pa
g
e 3 of 14Date: 10/31/96

Prior to 1990, FDA and CDC each had their own reporting systems for vaccine-associated
adverse events. The FDA system received reports primarily from vaccine manufacturers as
they became aware of specific instances of adverse outcomes following vaccination with their
products; the CDC system, known as MSAEFI (Monitoring System for Adverse Events
Following Immunization), collected reports from state public health coordinators based on
events observed at public clinics for which CDC provided vaccines.
A
s noted earlier, these systems were unified following passage of the NCVIA in 1986. The goal
of a unified surveillance system was to increase the efficiency of the Public Health Service in
collecting and assessing these reports and in monitoring the overall safety of vaccines.
18
How VAERS Works
VAERS is a passive surveillance system, a repository for voluntarily submitted reports. (An
active surveillance system, in contrast, would follow all individuals in a defined population to
determine their responses to vaccination.) To encourage reporting of any possibly vaccine-
induced adverse event, the criteria for reporting to VAERS are unrestrictive; the system
accepts and includes any report submitted, no matter how tenuous the possible connection
with vaccination might seem. As noted earlier, the NCVIA does require physicians to report
(directly to VAERS, or the manufacturer) certain categories of serious outcomes occurring
within a short period of time following specified childhood vaccinations, so one might expect a
fairly complete reporting of such events; the lack of enforcement provisions or even any
monitoring of reporting practices, however, precludes any assumptions about the extent to
which such events are in fact reported. Thus, VAERS potentially suffers both from
underreporting--not all vaccine-induced events are reported--and overreporting--coincidental
events, not caused by vaccines, can be reported.
Approximately 10,000 reports per year are submitted to VAERS. About 15% of these describe
a serious event, defined for regulatory purposes as an event resulting in a death, life-
threatening illness, hospitalization, prolongation of existing hospitalization, or permanent
disability. These reports are entered into the system by a Federal contractor and are reviewed
by medical staff of the FDA's Center for Biologics Evaluation and Research and CDC. Most of
the approximately 85% of reports not classified as serious describe events such as local
reactions and fever occurring within a day or two of vaccination. Many of these events are
clearly caused by the vaccine; they have been shown during placebo-controlled clinical trials to
occur more frequently in vaccine recipients than in those receiving placebo. The serious
events, unfortunately, are much more difficult to evaluate with regard to their causal
association with vaccines. Such events are observed too infrequently, if at all, in clinical trials
to assess their relation to vaccine. Most of these tend to be of a type known to occur in the
absence of vaccines as well, so in an individual case it is almost never possible to assess
definitively the role of the vaccine.
Temporal versus Causal Associations
Because of the large number of vaccine exposures, it is clear that temporal associations with
adverse outcomes will occur even when there is no true causal association. With hepatitis B
vaccine now recommended for all newborns, and other childhood vaccines (DTP, OPV, Hib)
being administered to nearly all infants starting at two months of age, most health problems in
infancy (of which there are many), whatever their cause, will occur in children who have been
vaccinated. Some of these problems will by chance occur in recently vaccinated children.
Pa
g
e 4 of 14Date: 10/31/96

An adverse event can be causally attributed to a vaccine if: (a) it conforms to a specific clinical
syndrome (such as anaphylaxis immediately following vaccination); (b) a laboratory result
confirms the association (for example, isolation of vaccine strain mumps vaccine virus from
patient with aseptic meningitis); (c) the event recurs on re-administration of the vaccine
(positive rechallenge); or (d) a controlled clinical trial or carefully designed epidemiologic study
shows greater risk of adverse events among vaccinated than control groups. Because few of
the adverse events reported to VAERS meet any of the first three criteria and because clinical
trials are almost always too small to provide useful information on serious rare events,
epidemiologic evidence is the basis for assessing causality for most serious adverse events
that are investigated.
Strengths and Weaknesses of VAERS
As a data base for epidemiologic studies, VAERS has many weaknesses. One major problem
is that, since unvaccinated people experiencing adverse events are not reported to VAERS,
there is no control group to study. Thus, there is no way to assess whether the number of
reported events is different from the number that would have been observed in the absence of
vaccination.
The quality of the data is also less than optimal. Because reports are sent in by a wide variety
of individuals, few of whom are experienced in completing data forms for medical studies,
many reports omit important data and contain obvious errors. Given that VAERS receives over
10,000 reports annually, it is difficult to assure the accuracy and completeness of the database
with current resources, although checks and follow-up are performed for a few key data items
such as the type of vaccine administered and the severity of the event.
Finally, the administration of multiple vaccines at the same time, following currently
recommended vaccine schedules, further complicates the assessment of adverse outcomes
because there is usually no way to determine which of the vaccines (if any) was most likely to
cause the outcome.
This is not to diminish the value of such a system. While VAERS data can rarely provide
definitive evidence of causal associations between vaccines and particular reported outcomes,
this type of national reporting system can rapidly document possible effects, generating early
warning signals that can then be more rigorously investigated in focused studies. In a sense,
VAERS is the front line of vaccine safety surveillance, so sensitivity takes precedence over
specificity; reporting of all serious events following vaccination is encouraged, inevitably
resulting in large numbers of reports that do not represent vaccine-induced problems.
VAERS data are especially valuable in assessing the safety of newly marketed vaccines.
Careful review of reports coming in during the initial months of availability can provide
additional reassurance about the safety of the vaccine or rapidly identify potential problems not
observed during the investigational phase. For example, the recent review by Niu et al
provided reassuring confirmation of the safety of hepatitis B vaccines in infants. FDA and CDC
medical staff maintain ongoing intensive surveillance of the new varicella and hepatitis A
vaccines.
The Case of Sudden Infant Death Syndrome (SIDS).
The case of SIDS exemplifies the problem with interpreting VAERS data. About 200 deaths a
Pa
g
e 5 of 14Date: 10/31/96

year are reported to VAERS; most of these are of infants under one year of age; of these, most
are diagnosed as SIDS. The reported time from vaccination until death varies from a few hours
to many weeks or even months. In most cases multiple vaccines are involved, consistent with
recommended immunization schedules. Because SIDS is a well documented (if not well
understood) phenomenon that occurs both in the absence and presence of vaccination, one
cannot presume a causal connection if SIDS follows shortly after vaccination; in fact, one can
predict that such events would occur even in the absence of a causal connection because
virtually all infants receive vaccines and because SIDS occurs at the relatively high rate of
somewhat over one per thousand live births in the United States.
In response to public concerns arising in the early 1980s about the safety of DTP vaccines, the
National Institute of Child Health and Development conducted a large case-control study
directed specifically at the question of the association between SIDS and DTP vaccination.
This study did not support the hypothesis that DTP vaccine caused SIDS; in fact, it
demonstrated a lowered risk for SIDS in children receiving DTP vaccine. (The authors of the
report suggested that this lowered risk estimate was more likely the result of differences in
baseline health status between children who did and did not receive scheduled vaccinations
than to any protective effect of the vaccine against SIDS.)
While this and other studies with similar results resolved the issue to the satisfaction of the
scientific community, some members of the public have remained concerned about a possible
connection between DTP vaccine and SIDS, citing the SIDS cases regularly reported to
VAERS. In response to such concerns, FDA and CDC staff calculated the number of SIDS
cases expected to occur by chance within a fixed number of days following immunization,
accounting for the age-adjusted SIDS rate and the proportion of infants vaccinated at specific
intervals, and determined that the number of cases reported for each time interval is far lower
than would be expected to occur by chance alone. (Of course, these estimates may have been
artificially lowered by underreporting of SIDS occurring shortly after vaccination.)
Advocacy groups raising concerns about vaccine safety regularly point out that the reasoning
described above for SIDS is flawed; since nearly all children are vaccinated, how do we know
that the background SIDS rate is not partially or even largely caused by vaccination? It is true
that there is no satisfactory unvaccinated control group to turn to, since the small group of
children in the United States who go unvaccinated through the first year of life would almost
certainly differ in important ways from those who do receive vaccinations on schedule. Well-
designed studies to date, including the study described above, have used an alternative
approach based on the assumption that if immunization caused sudden infant deaths, it would
do so within a few days of immunization. This approach allowed researchers to compare
children who died of SIDS with age-matched controls with respect to time since vaccination.
Other risk factors for SIDS have been identified recently: these include prone sleep position;
the thermal environment, including use of heavy and confining bedclothes; and maternal
smoking.
,,
SIDS rates have decreased dramatically in several countries
,,,
in connection with
back to sleep initiatives. These findings, while providing no additional direct evidence on the
role of vaccines, do suggest a mechanism for SIDS that may be more related to the physical
environment than to systemic factors in the child that might be affected by vaccination. Overall,
the evidence continues to strongly negate any causal association between SIDS and
vaccination.
The Safet
y
of Vaccine Lots
Pa
g
e 6 of 14Date: 10/31/96

Vaccine advocacy groups, organized and led primarily by parents who believe their children
have died or suffered serious injury as a result of vaccination, have questioned whether
particular vaccine lots may be more likely to induce such injuries. These groups have reviewed
the publicly available VAERS database, which includes the vaccine lot number for most
reports, and have raised questions about the safety of particular lots that appear to be
associated with a higher number of reports.
Vaccines are manufactured in large lots from which vials for individual administration are
derived. Procedures for the manufacture and release of vaccine lots are strictly regulated; prior
to public release, each vaccine lot must undergo stringent testing to assure both the potency of
the vaccine and the lack of contamination. Because of these procedures, the likelihood that
there could be something wrong with any bulk lot is extremely low--but not zero.
VAERS can address the question of the safety of individual vaccine lots much more effectively
than the question of causality in individual cases of adverse events. Since 1993, FDA medical
officers have performed weekly reviews of lot-specific reporting. These reviews require much
more information than simply the number of reports submitted for each vaccine lot, however;
the numbers themselves are inadequate to support any conclusions about safety, for several
reasons.
First, as noted earlier, there are errors in the database. The vaccine lot number may be
miscopied by the reporter onto the VAERS form. This results in numerous lots in the database
with a single report or a very few reports, providing a misleading contrast with the numbers of
reports for valid lots.
Second, lot sizes can vary greatly. Clearly, the number of reports generated from a vaccine lot
containing half a million doses cannot be sensibly compared with the number of reports
generated from a lot only one-tenth as large. Data on lot size, although available to the FDA for
monitoring purposes, do not appear in the VAERS database because these data are legally
considered proprietary to the manufacturer and their release by FDA is prohibited.
Third, there is a time factor. Vaccine lots that have been on the market for only a few months
will be associated with fewer reports than a lot released several years earlier.
Finally, there will always be chance variability in reporting rates. Even when the numbers of
reports are standardized for lot size and length of time on the market (as they are for internal
FDA review of the database), there will always be one lot associated with the highest rate of
reports and one lot with the lowest. The more lots, the greater the difference between the
highest and lowest rates will be; the difference may be quite large for old and widely used
vaccines such as DTP even assuming all lots are equally safe.
When numbers of reports are compared between vaccines, there are further considerations.
The database will contain more reports for a vaccine administered five times to each individual,
such as DTP, than for a vaccine such as MMR that is administered only twice. There will be
more reports for a vaccine that has been on the market for many years (such as DTP) than for
a vaccine that has been available for only a few years (such as Haemophilus influenzae Type
B vaccine). More death reports will be seen for vaccines given in infancy, when the
background death rate is higher and SIDS is a factor, than for vaccines given later in
childhood.
Pa
g
e 7 of 14Date: 10/31/96

The FDA staff must take all of these factors into consideration when monitoring the database
for unusual patterns of reporting from specific lots. Computerized methods are in place to
identify lots with high reporting rates, accounting for lot size and time on the market. The
threshold for identifying these lots is set deliberately low to ensure the earliest possible signal
of a real problem.
Lots identified at this first screening stage are subject to additional scrutiny. The experience of
related or sister lots (those made from the same large batch of product) is considered, since
most problems should affect all lots from the same batch. The types of reports submitted are
carefully reviewed; a series of similar events would be more suggestive than a scattering of
events of different types with no unique syndrome evident. The results of the lot's initial safety
testing is also reviewed. Additional information may be requested from the manufacturer, and
in some cases the safety testing might be repeated.
Over the three years during which these monitoring procedures have been in place, no lot has
been found to be unsafe. This result is not surprising given the stringency of the manufacturing
and testing requirements to which vaccines are subject. Nevertheless, because of the
possibility of such a problem arising, regular attention to lot-specific reporting will remain an
important aspect of FDA's program of vaccine safety monitoring.
Using VAERS Data to Identify Possible New Reactions
Several investigations of VAERS data have uncovered previously unrecognized problems that
may occur rarely in vaccine recipients. Beeler, Varricchio and Wise noted occasional instances
of life-threatening thrombocytopenias following the administration of MMR vaccine, a
previously unappreciated level of severity of a known side effect. Wise and Kiminyo
documented a series of cases in which hair loss followed immunizations (primarily hepatitis B),
a rare effect not previously reported. Braun and colleagues have identified a series of cases of
severe injuries resulting from vaccination-induced fainting, or syncope.
Sometimes VAERS data may provide the useful and reassuring information that new problems
have not been identified after additional experience with a vaccine, as in the previously noted
report of Niu et al with regard to hepatitis B vaccine in infants.
27
Using VAERS to Study Trends in Reporting of Adverse Events
VAERS data have also been used to compare reporting patterns over time and investigate
changes in reporting rates that might be due to changes in vaccine practices. For example,
CDC epidemiologists reviewed reports of fever, seizures, and hospitalizations following
administration of a newly licensed combination of diphtheria, tetanus and acellular pertussis
vaccine (DTaP). The rate of such reports was about one-third lower than the reporting rate
following the standard DTP vaccine, consistent with--and confirming in the context of general
practice--the safety findings of the prelicensure clinical trials
,
.
Ecologic studies--that is, comparisons of outcomes during different time periods--using data
from MSAEFI, the CDC surveillance that preceded VAERS, showed that a change from
separate to simultaneous immunization with DTP and MMR vaccines at 15 months of age did
not change the types of adverse events reported.
46
A combined analysis of MSAEFI and
VAERS data indicated that the rates of reported hospitalizations and deaths following DTP
Pa
g
e 8 of 14Date: 10/31/96

vaccine remained constant from 1985 to 1992, despite the addition of Haemophilus influenzae
Type B vaccine to the routine infant immunization schedule. These studies provided some
reassurance that by adding vaccines to the recommended immunization schedule gains in
protection from disease were not offset by an increased burden of adverse events caused by
the vaccines.
Reporting patterns have also been studied to estimate the extent of underreporting of adverse
events. Researchers used the known rates of a variety of vaccine reactions, as reported in
controlled studies, to estimate the reporting efficiency--the proportion of events actually
occurring that are reported--of VAERS and predecessor passive surveillance systems.
Reporting efficiencies were similar and, perhaps not surprisingly, were the highest for serious
events such as vaccine-associated paralytic polio (about 70%) and lowest for minor events
such as rashes following measles vaccination (less than 1%). This study provides suggestive
evidence that for the events of greatest concern, VAERS reports represent substantially more
than the tip of the iceberg.
Newer Approaches to Vaccine Safety Monitoring
While VAERS is the initial safety screen, potentially providing the earliest signal of any new
vaccine reaction, a reporting system of this kind has major limitations, including
underreporting, lack of specificity, and lack of a natural control group. To compensate for these
limitations, other approaches to vaccine safety surveillance have been developed.
The increasing availability in recent years of computerized medical administrative data bases
for defined populations such as health maintenance organization (HMO enrollees) has
radically improved the ability to conduct pharmacoepidemiologic studies. In these databases,
vaccination records can be linked to records of hospitalizations and diagnoses of serious
conditions, minimizing the problem of underreporting of serious events encountered in VAERS.
The enrollment in individual HMOs can number in the millions, permitting study of rare events.
This capability can be further enhanced by collecting data prospectively from several HMOs
under a well-defined research protocol and aggregating the data. In principle, all of the
information necessary for rigorous epidemiologic analysis is available in such settings,
including numbers of doses administered, comparison groups, and potential confounders.
Since 1990, the CDC has worked with four HMOs to organize a Large Linked Data Base
(LLDB) for vaccine safety studies as part of the Vaccine Safety Datalink Project. In this project,
automated vaccination records on half a million children under six years of age (representing
about 2% of the U.S. population in this age group) are linked to their medical records; the
project will soon be expanded to include older age groups. This resource is being used to
examine particular associations identified by the Institute of Medicine in their study mentioned
earlier as requiring further investigation
14,15
(for example, seizures following immunization)
and is also regularly utilized to further evaluate potential associations identified through
VAERS. One recent investigation confirmed an association between seizures and DTP and
MMR vaccinations by comparing vaccine exposures within specified time periods (one day for
DTP, one week for MMR).
48
The Vaccine Safety Datalink Project also has its limitations, however. Because few
nonimmunized controls are available within the HMOs, the project relies predominantly on
comparison of incidence rates of adverse events between specified time periods following
vaccination. These studies are therefore limited in their ability to investigate the association
Pa
g
e 9 of 14Date: 10/31/96

between vaccination and events with delayed or insidious onset such as autism and learning
disability. Adverse events that do not result in a health care visit (or, more generally, data not
automated in the HMO) are also not easily studied. The geographical concentration of the
project's HMOs on the West Coast of the United States may also limit the generalizability of
results. Finally, even the large sample sizes available in this HMO consortium currently do not
provide enough power to study extremely rare events such as Guillain-Barré syndrome (GBS)
and encephalopathy in a timely manner; specialized studies are still required to address such
issues. For example, to evaluate the occurrence of GBS following the 1993-1994 influenza
vaccination season, a special study is being conducted by the CDC in which all cases of the
syndrome in four states are being sought through the centralized statewide hospital discharge
tapes.
Challenges for the Future
The continued development of new vaccines to prevent diseases such as chicken pox,
rotavirus, pneumococcal pneumonia, and respiratory syncytial virus will intensify the
challenges of vaccine safety monitoring. Concerns have been raised about reactions being
exacerbated when vaccines are combined; while current experience does not suggest that
there would be insurmountable safety problems with adding new vaccines to currently
available combinations, the possibility of increased reactogenicity is well recognized.
The public concerns about the safety of vaccines that are frequently mandated prior to entry
into public school, day care centers, universities, and workplaces are legitimate and important.
We would all like such products to pose zero risk of adverse effects. Unfortunately, this goal is
not achievable for any pharmacologically active product--if there is a beneficial effect, there will
be some risk, however tiny, of an adverse effect.
It must be recognized that protection of individuals from serious diseases depends not only on
their own immunization but on the immunization of others in their community; since vaccines
are not 100% effective, people's chance of disease is lower if those around them remain
healthy than if those around them carry the disease. This is why most communities require
children to be vaccinated against certain diseases, assuming no contraindication, rather than
leaving the choice to individual parents on the basis of their own risk-benefit assessments.
The overwhelming view of the medical/public health community is that the risks of vaccine
reactions, both the common mild reactions and the rare, more serious reactions, are very
much outweighed by the public health benefit conferred by current vaccination practices and
policies. Epidemiologists, pediatricians, statisticians, and others involved in vaccine safety
surveillance projects will continue to investigate new and improved methods to monitor vaccine
safety. The goal of understanding what events might be caused or promoted by certain
vaccines, and which individuals might be at high risk to experience such events, will remain a
challenging but extremely important one for the public health community.
ACKNOWLEDGEMENTS
Dr. Ellenberg is Director, Division of Biostatistics and Epidemiology, Office of Establishment
Licensing and Product Surveillance, FDA.
Pa
g
e 10 of 14Date: 10/31/96

Dr. Chen is Chief, Vaccine Safety and Development Activity, Epidemiolgy and Surveillance
Division, CDC.
We would like to acknowledge contributions from Carol Krueger RN,BSN, David Davis, Karen
Goldenthal MD, Peter Patriarca MD, MPH and Jerome Donlon MD, PHD of the FDA; Gina
Terracciano DO, Penina Haber MPH, and Tara Strine, of the CDC; and Geoffrey Evans MD of
the Health Resources and Services Administration, to the preparation of this manuscript.
Address correspondence to Susan Ellenberg, Ph.D. 1401 Rockville Pike, HFM-210, Rockville
MD 20852-1448
References
* Waldman RH, Kluge RM, editors. Textbook of infectious diseases.
New Hyde Park, New York: Medical Examination Publishing Co., Inc.,
1984; 4.
* Plotkin SL, Plotkin SA. A short history of vaccination. In: Plotkin SA,
Mortimer EA, editors. Vaccines. Philadelphia: WB Saunders, 1994;1-12.
* Sutter RW, Prevolts DR. Vaccine-associated paralytic poliomyelitis
among immunodeficient persons. Infect Med 1994;11:426:429-430,
435-438.
* U.S. Changes Policy on Polio Inoculation, Recommending Salk
Vaccine for First Doses. New York Times 1996 Sep 20, Sec.A:18.
* Coulter HL, Fisher BL. DPT, a shot in the dark. NewYork: Warner
Books, 1985.
* Kimura M, Kuno-Sakai H. Developments in pertussis immunisation in
Japan. Lancet 1990;336:30-32.
* Miller D, Madge N, Diamond J, Wadsworth J, Ross E. Pertussis
immunisation and serious acute neurological illnesses in children. BMJ
1993;307:1171-1176.
* Krantz I, Taranger J, Trollfors B. Estimating incidence of whooping
cough over time: a cross-sectional recall study of four Swedish birth
cohorts. Int J Epidemiol 1989;18:959-963.
* Hinman AR. DTP vaccine litigation. Am J Dis Child 1986;140:528-530.
* Mariner WK, Clark ME. Confronting the immunization problem:
proposals for compensation reform. Am J Public Health 1986;76:703-
708.
* CDC. Diptheria-tetanus-pertussis vaccine shortage-United States.
MMWR 1984;33:695-696.
Pa
g
e 11 of 14Date: 10/31/96

* National Childhood Vaccine Injury Act of 1986, at Section 2125 of the
Public Health Service Act as codified at 42 U.S.C.  300aa-
(Supp.1987).
* Flamberg G. An experiment in tort reform. Public Health Rep
1995;110:635-638.
* Howson CP, Howe CJ, Fineberg HV, editors. Adverse effects of
pertussis and rubella vaccines. Washington DC: National Academy
Press, 1991.
* Stratton KR, Howe CJ, Johnston RB Jr., editors. Adverse events
associated with childhood vaccines: evidence bearing on causality.
Washington DC: National Academy Press, 1994.
* CDC. National Childhood Vaccine Injury Act: requirements for
permanent vaccination records and for reporting of selected events after
vaccination. MMWR 1988;37:197-200.
* Chen RT, Rastogi SC, Mullen JR, Hayes S, Cochi SL, Donlon JA,
Wassilak SG. The Vaccine Adverse Event Reporting System (VAERS).
Vaccine 1994;12:542-550.
* Kessler DA. Introducing MEDWatch. JAMA 1993;269:2765-2768.
* Bankowski Z, Dunne JF (editors). Drug surveillance: international
cooperation past, present and future. Proceedings of the XXVIIth
Council for International Organizations of Medical Science Conference;
1993 Sep 14-15; Geneva, Switzerland. Geneva: CIOMS, 1994.
* Herrington DA. Initial clinical evaluation of new vaccine candidates. In:
Woodrow GC, Levine MM, editors. New generation vaccines. New York:
Marcel Dekker, 1990:43-49.
* Parkman PD, Hardegree MC. Regulation and testing of vaccines. In:
Plotkin SA, Mortimer EA, editors. Vaccines. Philadelphia: W.B.
Saunders, 1994;889-902.
* Clemens JD, Stanton BF. Longer term evaluation of vaccine efficacy.
In: Woodrow GC, Levine MM, editors. New generation vaccines. New
York: Marcel Dekker, 1990;51-68.
* Faich GA. Adverse drug reaction monitoring. N Engl J Med
1986;314:1589-1592.
* Stetler HC, Mullen JR, Brennan JP, Livengood JR, Orenstein WA,
Hinman AR. Monitoring system for adverse events following
immuniation. Vaccine 1987;5:169-174.
* Braun MM, Ellenberg SS. Descriptive epidemiology of adverse events
Pa
g
e 12 of 14Date: 10/31/96

following immunization: reports to the vaccine adverse event reporting
system (VAERS), 1991-1994. J Peds 1996 (in press).
* CDC. Recommended childhood immunization schedule. MMWR 1996;
44:940-943.
* Niu, MT, Davis, DM, Ellenberg, SS. Recombinant hepatitis B
vaccination of neonates and infants: emerging safety data from the
Vaccine Adverse Events Reporting System (VAERS). Pediatr Infect Dis
J. 1996;15:771-776.
* Hoffman HJ, Hunter JC, Damus K, Parker J, Peterson DR, van Bell G,
Hasselmeyer EG. Diphtheria-tetanus-pertussis immunization and
sudden infant death syndrome. Pediatr 1987;79:598-611.
* NBC Now. Adverse Reactions to Vaccines. 1995 Mar 2.
* Hoffman HJ, Hilman LS. Epidemiology of the sudden infant death
syndrome: maternal, neonatal, and postnatal risk factors. Clin Prenatol
1992;19:717-737.
* Ponsonby A-L, Dwyer T, Gibbons LE, Cochrane JA, Wang Y-G.
Factors potentiating the risk of sudden infant death syndrome
associated with the prone position. New Eng J Med 1993;329:377-382.
* Klonoff-Cohen, HS, Edelstein, SL, Lefkowitz, ES, Srinivasan, IP,
Kaegi, D, Chang, JC, Wiley, KJ. The effect of passive smoking and
tobacco exposure through breast milk on sudden infant death
syndrome. JAMA 1995;273:795-798.
* Wigfield RE, Fleming PJ, Berry PJ, Rudd PT, Golding J. Can the fall in
Avons sudden infant death rate be explained by changes in sleeping
positions BMJ 1992;304:282-283.
* Dwyer T, Ponsonby AL, Blizzard L, Newman NM, Cochrane JA. The
contribution of changes in the prevalence of prone sleeping position to
the decline in sudden infant death syndrome in Tasmania. JAMA
1995;273:783-789.
* Mitchell EA, Brunt JM, Evard C. Reduction in mortality from sudden
infant death syndrome in New Zealand. Arch Dis Child 1994;70:291-
294.
* Spiers, PS, Guntheroth WG. Recommendations to avoid the prone
sleeping position and recent statistics for sudden infant death syndrome
in the United States. Arch Pediatr Adolesc Med 1994;148:141-146.
* Willinger M. Sleep position and sudden infant death syndrome. JAMA
1995;273:818-819.
Pa
g
e 13 of 14Date: 10/31/96

* The National Vaccine Information Center (NVIC). Vaccine Adverse
Event Reporting System (VAERS)--DPT vaccine report: master list of lot
numbers. 1994.
* 21 CFR  610
* Beeler, J, Varricchio, F, Wise, RP. Thrombocytopenia following
immunization with measles vaccines. Pediatr Infect Dis J 1996;15:1019-
1030.
* Wise RP, Kiminyo KP. Postvaccinal alopecia: recognition through
passive surveillance [abstract]. Pharmacoepidemiology and Drug Safety
1995;4:S79.
* Braun MM, Patriarca P, Ellenberg SS. Syncope after immunization.
Arch Pediatr Adolesc Med. In press.
* Rosenthal S, Chen R, Hadler SC. The safety of acellular pertussis
vaccine versus whole cell pertussis vaccine: a post-marketing
assessment. Arch Pediatr Adolesc Med 1996;150:457-460.
* Centers for Disease Control and Prevention. Food and Drug
Administration approval of use of diphtheria and tetanus toxoids and
acellular pertussis vaccine. MMWR 1991;40:881-882.
* Centers for Disease Control and Prevention. Pertussis vaccination:
acellular pertussis vaccine for the fourth and fifth doses of the DTP
series: update to supplementary ACIP statement. MMWR 1992;41:(RR-
15);1-5.
* Chen RT, Haber P, Mullen JR. Surveillance of the safety of
simultaneous administration of vaccines: the CDC experience. Ann NY
Acad Sci 1995;754:309-320.
* Rosenthal S, Chen RT. Reporting sensitivities of two passive
surveillance systems for vaccine adverse events. Am J Public Health
1995;85:1706-1709.
* Strom BL, Carson JL. Use of automated databases for
pharmacoepidemiology research. Epidemiol Rev 1990;12:87-107.
* Chen RT, Glasser J, Rhodes P, Thompson RS, Mullooly JP, Black SB,
et al. The vaccine safety datalink project: a new tool for improving
vaccine safety monitoring in the United States. Pediatrics 1996 (in
press).
* Williams JC, Goldenthal KL, Burns DL, Lewis BP Jr., editors.
Combined vaccines and simultaneous administration: current issues
and perspectives. Ann New York Acad Sci 1995;754:XI-XV, 1-404.
Pa
g
e 14 of 14Date: 10/31/96