U.S. FOOD AND DRUG ADMINISTRATION
BLOOD PRODUCTS ADVISORY COMMITTEE
July 21, 2009
Hilton Washington DC North/Gaithersburg
620 Perry Parkway
Gaithersburg, Maryland 20877
“This transcript has not been edited or corrected, but
appears as received from the commercial transcribing
service. Accordingly the Food and Drug Administration makes
no representation as to its accuracy”
CASET Associates, Ltd.
Fairfax, Virginia 22030
TABLE OF CONTENTS
Opening Remarks, Statement of Conflicts of Interest 1
Recognition of Retiring Members 2
Summary of the April 30-May 1, 2009 Meeting of the DHHS 3
Advisory Committee on Blood Safety and Availability
Summary of the June 12, 2009 Meeting of the FDA 7
Transmissible Spongiform Encephalopathies
Advisory Committee – David Asher
2009 A/H1N1 Influenza Update
Joseph Bresee 38
Louis Katz 48
Topic III: Informational Session: Hemovigilance
Pilot Hemovigilance Module of the National Health Care 59
Safety Network – William Bower
FDA's Sentinel Initiative and CBER's Analytic Epidemiology 69
Branch – Robert Wise
P R O C E E D I N G S (9:00 a.m.)
Agenda Item: Opening Remarks, Statement of
Conflicts of Interest
DR. SIEGAL: Good morning. I rarely have any
opening remarks of note, but today since it is my last
official meeting on this committee and my last meeting as
Chair, I just want to thank the FDA and especially Jay
Epstein and Jesse Goodman, but also Bill Freas and Dawn Jane
and Pearlie Muckleby for the wonderful opportunity I have had
to be a part of this committee and this process, and what a
tremendous honor it was to have done so. So I want to thank
On that note, I'll turn the meeting over to Bill.
DR. FREAS: Good morning. I have a brief conflict
of interest statement announcement that pertains to today
only that I would like to read into the record.
This brief announcement is an addition to the
conflict of interest statement read at the beginning of the
meeting on July 20 and will be part of the public record for
the Blood Products Advisory Committee meeting on July 21,
2009. The committee will hear updates and informational
presentations on the April 30 and May 1 meeting of the DHS
Advisory Committee on Blood Safety and Availability. The
June 12, 2009 meeting of the FDA Transmissible Spongiform
Encephalopathy Advisory Committee and an overview of the
epidemiology and virology of the 2009 A/H1N1 influenza virus
and its impact on the U.S. blood system. The committee will
also hear informational presentations on recent hemovigilance
efforts, including the pilot hemovigilance module in the
National Health Care Safety Network.
These updates and presentations are not for
discussion by the committee members, and therefore the
committee members were not screened for financial conflicts
of interest relating to these presentations and informational
updates. The conflict of interest statements will be
available for review at the registration table.
Dr. Siegal, I'll turn it back over to you.
Agenda Item: Recognition of Retiring Members
DR. SIEGAL: The next item on the agenda is the
recognition of retiring members, I suppose.
DR. FREAS: Dr. Midthun, would you come and make
the presentations? Dr. Midthun is our Acting Center
DR. MIDTHUN: Good morning. We have many members
rotating off our Advisory Committee in September. This will
probably be the last meeting where many will be in
attendance. I know what a large amount of work this is to be
part of this committee. People really spend many, many hours
and days working for us and contributing their expert advice,
and we very much appreciate it. This was just a small token
of our appreciation, but nonetheless it is very important to
I would like to start by presenting plaques to some
of our departing members. If Dr. Ballow would please come up
so we can present him with his plaque.
If Dr. Cryer would please come up next.
If Dr. Kulkarni would please come up.
Then last but certainly not least, Dr. Siegal, if
you would please come on up. Dr. Siegal has been the Chair
of our Advisory Committee, and that really presents an added
burden, and we really wish to thank him for all his work.
Agenda Item: Committee Updates
DR. SIEGAL: Now it is time to get serious. Let's
start with the committee updates. The first update will be
provided by Richard Henry, Advisory Committee on Blood Safety
and Availability, summary of the April 30 to May 1, 2009
meeting of the DHHS Advisory Committee on Blood Safety and
Agenda Item: Summary of the April 30-May 1, 2009
Meeting of the DHHS Advisory Committee on Blood Safety and
DR. HENRY: Good morning. Thanks for having me. I
am Rich Henry from HHS Secretary's office. I will be
providing a brief summary of our May meeting of the Advisory
Committee on Blood Safety and Availability.
Before the committee met, we set out some goals.
We wanted the committee to comment on the elements that would
characterize a more robust and transparent decision making
process for transfusion safety policy. We also wanted the
committee to comment on whether use of formal tools of policy
analysis such as risk assessment and cost effectiveness or
cost utility modeling would be of value to enhance the
current decision making process, and how they might be
integrated into the current system.
We asked specific questions of the committee. What
are the advantages and disadvantages of comparative
effectiveness models? What next steps if any does the
committee recommend to enhance the quality and transparency
of federal decision making for transfusion and
transplantation safety policy?
The committee came back afterwards and sent several
messages to the Secretary. Decisions regarding transfusion
and transplantation safety are often complex and require
consideration of multiple features, including scientific,
medical, economic, social and political aspects of the issue
Medical and scientific considerations should be
paramount in guiding blood policy decisions on transfusion
and transplantation safety. However, the state of knowledge
is often incomplete or imperfect.
Two more statements to the Secretary. Ultimately
safety and not economic considerations should be the primary
feature considered. However, decisions should not be made in
an economic void. Safety decisions should be made in an
ethical transparent manner with adequate input from all
involved stakeholders, including end users, the public and
expert resources from diverse backgrounds relevant to the
question at hand.
After the statements to the Secretary, the Advisory
Committee made five specific recommendations. The first
one's theme is stabilization of the role of HHS in the
decision process. The recommendation was, recognizing the
importance of stability in staff and funding for transfusion
safety within HHS, the committee recommends establishment of
a line item or a budget item in the Secretary's budget for
support of these activities. Currently there is none.
The second theme is access to critical public
health information. Recognizing that a robust and sustained
biovigilance reporting network for monitoring transfusion and
transplantation safety is essential, the committee recommends
that the Department of Health and Human Services identify and
provide short term bridge funding to continue the necessary
pilot and recruitment phases of the National Biovigilance
Network with consideration of mandatory requirements in
participation and reporting.
Further, to convene a group of stakeholders to
define the optimal long term funding model for national
biovigilance, including a blood, tissues and organs program
whose implementation the Secretary will insure.
The third theme was prioritization of issues.
Whereas optimizing the safety of transfusion and
transplantation is a universally shared priority, policy
making without overt consideration of reimbursement and other
systemic implications may have adverse impacts on the
allocation of limited health resources and on patient health.
To address this issue, the committee recommends
ongoing assessment of transfusion risk and prioritization of
threats and opportunities.
The fourth theme was the alignment of funding. The
implementation of blood safety initiatives requires funding
that currently is not linked with policy decisions. This
disparity creates distortions to decision making, adverse
tradeoffs and possible delays of implementation of safety
measures within the blood system.
The committee then recommends, when policy
decisions are reached, there must be linkage of funding to
insure system wide implementation of desired safety
Transparency. The committee recommends the
rationale for blood safety enhancements must be communicated
to all stakeholders, including end users, the public and
treating physicians. Where needed, adequate support should
be provided to meet educational needs of these stakeholders.
The committee also recommends convening a workshop,
including requisite legal, ethical, economic and consumer
representative to explore public understanding and tolerance
of transfusion related risk and its relationship to
mechanisms of compensation for harm.
Our next advisory committee will be on October 1
and 2. By then we should have our new Advisory Committee
members seated. There are approximately five seats that we
are trying to fill at the moment. It will be held at the
university, the same place as last time, just north of here.
We do not have an agenda or a theme for the meeting yet, but
it will take place on October 1-2.
DR. SIEGAL: Thank you. Next we will hear from
David Asher from OBRR, FDA, a summary of the June 12, 2009
meeting of the Transmissible Spongiform Encephalopathies
Advisory Committee, TSEAC.
Agenda Item: Summary of the June 12, 2009 Meeting
of the FDA Transmissible Spongiform Encephalopathies Advisory
DR. ASHER: Thank you, Dr. Siegal. On the 17th of
February the United Kingdom's Health Protection Agency
announced that evidence of infection with the agent causing
variant Creutzfeldt-Jakob disease, as evidenced by the
detection of abnormal prion protein in the spleen, had been
found postmortem in a person who had died with hemophilia.
The patient was over 70 years old and died of a condition
unrelated to variant CJD with no symptoms of that or any
other neurological condition prior to death. The abnormal
protein was identified during a very extensive postmortem,
including more than 20 sections taken from the spleen.
The conclusion was that what had until then been a
theoretical risk associated with a human plasma derived
coagulation factor must now be considered an actual risk, at
least to certain individuals.
The Center for Biologics Evaluation and Research is
committed to re-evaluate the risk of spongiform
encephalopathies to the blood supply and supply of plasma
derivatives periodically when important new information
becomes available. This finding appeared to us to constitute
such an event. We have not had a meeting of the Spongiform
Encephalopathies Advisory Committee since December of 2006,
but we hurried to convene one on the 12th of June. Two of
the members of today's committee were present. What I am
going to present to you is largely my own impression of what
went on at that committee, and if they or the FDA staff have
something to add, I encourage them to do that.
The overall trend for variant Creutzfeldt-Jakob
disease has been favorable for the last few years. The
bovine spongiform encephalopathy accepted as the source of
food-borne variant Creutzfeldt-Jakob disease has been in
decline in Europe, although some uncertainties about bovine
spongiform encephalopathies remain, particularly some
countries that are believed to have been at risk for it have
not reported any cases, so we still consider it an infection
of worldwide significance.
The trends for variant Creutzfeldt-Jakob disease
have also been encouraging. Declining cases in the United
Kingdom, at least among persons with the prion protein
encoding gene, codon 129, which has been associated with all
clinical cases until recently, there has been a marked
decline. I'll show you some of that information in a minute.
However, last year a typical clinical case of
variant Creutzfeldt-Jakob disease was observed in a person
who was not homozygous for methionine at that locus.
Although the family declined to provide tissue either during
life or at autopsy, clinicians in the United Kingdom appeared
to consider that to be a bona fide variant Creutzfeldt-Jakob
disease, which creates increasing concern that there may be a
second wave of the disease among people of genotypes
previously spared. There is some other information that
Another concern has been the marked discrepancy in
estimates of the prevalence in the United Kingdom. I will
show you some of the information that was presented at the
meeting and why that is important.
There have been four transfusion transmitted
infections of variant Creutzfeldt-Jakob disease associated
with non-leuko-reduced red blood cell concentrates in the
United Kingdom, but there have been no new report of
transfusion transmitted vCJD since the winter of 2007, and
now this, the presumptive transfusion by a plasma derivative.
That is the reason for the meeting that we had on the 12th
The morning was devoted to a decisional issue,
which was to critique the review of a modified FDA risk
assessment looking at the risk for plasma derived factor 8 in
the United States. There were also a number of informational
issues. We had reports of the BSC situation in the United
States, Canada and the European Union, plus three research
topics related to TSEs and blood safety, one a promising
monkey model infecting monkeys with BSE and vCJD agent and
finding infectivity in blood, important because it provides a
possible blood reference material that it appears we are not
going to get from human infection. Efforts to develop
antemortem tests to identify infected people during the
incubation period of TSEs, the prospects at the moment seem
guarded, and problems in correlating abnormal prion proteins
with infectivity in blood and tissues, which has implications
for the future of antemortem testing. I won't go over any
of those this morning. They are available in the transcript,
and there is some supporting material that is available
through the Web.
Robert Will, founder of the United Kingdom CJD
Surveillance Unit, reviewed the vCJD situation in the United
Kingdom. In the Transfusion Medicine Epidemiological Review
Alan Williams reviewed FDA's geographic deferral policy to
reduce the risk of transfusion transmitted variant and plasma
derivative transmitted variant Creutzfeldt-Jakob disease.
Dot Scott, the properties of TSE agents in blood and
approaches to estimating their clearance by plasma
processing. Albrecht Groener, representing the Plasma
Protein Therapeutics Association, presented actual spiking
studies clearing TSE agents from various intermediates of
plasma derivatives. Then the risk assessment itself and
modifications to it for plasma derivatives were presented by
Steve Anderson from our CBER Office of Biostatistics and
In your package I have listed the agenda for
informational presentations, but time doesn't permit me to go
through those. But I do want you to see the questions that
were posed to the TSE Advisory Committee. The committee was
asked whether it agreed with updated and new inputs to the
FDA risk assessment model for U.S. licensed plasma derived
factor 8. The updated inputs were a change in the estimated
prevalence of vCJD in the United Kingdom for one set of
scenarios; a change in the estimated time in which infections
had been acquired. That was essentially a bookkeeping change
correcting an error from the original assessment from 2006.
Then the time during an incubation period when infectivity
might be present in blood. We felt it was prudent to become
more conservative, and we changed the estimated time from 50
percent of the total incubation period to 75 percent, that
is, onset of agent in blood starting 25 percent of the way
into the infection. Of course, we really don't know when
infectivity appears in blood.
Then three new inputs. The effect of the genotype
at codon 129 of the prion protein encoding gene on genetic
susceptibility to infection, now assumed to be universal, and
the proportions of persons in the United States with various
genotypes, because at least the incubation period and
possibly the attack rates will almost certainly be different.
Our assumption is that the genotype proportions in the U.S.
can be assumed to be more or less similar to those in the
United Kingdom, the distribution of incubation periods for
persons of different genotypes and the age of persons with
asymptomatic variant CJD infections.
The original second question was modified, because
after the information packet had been circulated, a risk
assessment appeared from the U.K. Health Protection Agency
suggesting that although the infection in their patient was
very unlikely to have resulted from food exposure, red cell
transfusions or endoscopy, but considering the probable high
prevalence or possible high prevalence of preclinical variant
CJD in U.K. plasma donors, the infection was at least as
likely to have resulted from treatments with non-vCJD
implicated factor 8, as from one batch of factor 8 to which a
known patient had contributed, which had originally been
thought to be the most likely source of the patient's
So the question was reworded to read, despite the
finding of minimal additional risk in FDA's modified risk
assessment, should the recent report from the U.K. Health
Protection Agency attributing a case to treatment with U.K.
sourced factor 8 alter our interpretation of the risk for
U.S. licensed preparations of factor 8, and based on
available information should we recommend additional risk
reducing steps for plasma derivatives, recommend revised
warning labels, or recommend modifications to our public
Bob Will summarized the 211 cases of variant CJD
observed so far worldwide. That does not include
asymptomatic infections, 168 of them in the U.K., 43 outside
the U.K., and eight of them outside the U.K. in persons who
might have been infected in the U.K. That would include
three of the four cases in North America, two cases in the
United States who were longtime U.K. residents. We have had
one case in a recent arrival from Saudi Arabia, thought by
the CDC probably to have been infected through meat imported
into Saudi Arabia from the United Kingdom.
BSE peaked in the United Kingdom in about 1991 and
variant CJD deaths in about 2000, the first transfusion
transmitted case December 2003, and the transfusion
transmitted case reported in February of this year.
There has been an interesting comparison possible
of variant CJD with a look-back study conducted by the
American Red Cross over many years for other forms of CJD.
For variant CJD only 26 recipients of labile components have
survived for more than five years without coming down with
variant CJD, whereas the report of Cary Dorsey and Shim Yen
Zu, Roger Dodd and others suggest that similar recipients of
CJD implicated red blood cells and labile components in this
country, that of 144 persons to have lived five years, none
of them has come down with CJD. So the risk is clearly an
order of magnitude at least less for sporadic CJD, non-
Incubation periods for food-borne CJD, minimum
possibly as short as four and a half years, usual 12 years.
Transfusion transmitted cases, incubation periods which can
be determined precisely, have ranged from six years and four
months to eight and a half years. The one plasma derivative
associated case, if the implicated plasma derivative was
responsible, the incubation period was 11 years, and if it
was one of the other plasma derivatives, it could have been
But we can conclude that during the last three and
a half years of incubation period of variant CJD, some
clinically healthy donors who later developed variant CJD
were already infectious. We don't know how much longer their
blood might be infectious, how much infectivity is present in
their blood. We don't know if donors with other genotypes in
the prion protein gene are also infectious during the
clinical illness. They have to be assumed to be, we believe,
how long they might be infectious and how much infectivity is
In 2006, Steve Anderson and colleagues developed a
risk model for looking at the theoretical risk of
transmissions attributable to plasma derivatives. One of the
important things they did was to conduct a sensitivity
analysis in which all of the elements of risk are held
constant except for one. They are varied stepwise one at a
time, and they look at the output to see which elements of
risk are the most important.
By far, the most important effect on risk came from
reductions in infectivity offered by the manufacturing
process. More about that in a minute. Also important was
the amount of factor 8 used by a person each year. In third
place, the prevalence of variant CJD in the donor population.
Other features that can't be controlled are the efficiency
of infection by the intravenous route, because most of what
we know comes from animal studies using the intracerebral
route. We know in most models, infection by the intravenous
route is less effective, and the quantity of infectivity in
blood, which we only know for animals.
Let's look at prevalence. A survey was conducted
looking at appendix and tonsil tissue in the United Kingdom
and published in 2002. This is based on the observation that
at autopsy, abnormal PRP is found consistently in tonsils,
lymph nodes, appendix and spleen of persons dying with
variant Creutzfeldt-Jakob disease. In the appendix,
fortuitously three patients have had appendectomies years
before they came down with variant CJD, and abnormal protein
was detectable at eight months and at two years, but not at
ten years before onset of illness, which was the basis for
When Hilton and colleagues looked at over 12 and a
half thousand appendices from persons without overt vCJD in
the United Kingdom, they found three positives. Two of them
could have their genes sequenced, and they were both
homozygous for valine at that locus.
A subsequent survey has used tonsillar tissue but a
less sensitive and all negative, with over 60,000 tonsils,
but considering the first survey, that could have been
expected. The prion protein detection test used was less
sensitive. So we are left with a possible prevalence of one
case out of 4,000-some in the United Kingdom, which
translates into over 160 cases per million.
Another survey was based on modeling using the
number of cases of variant CJD and a number of assumptions to
generate a mathematical prediction of the number of
infections. That was much less troubling. The original
prediction was 1.8 cases per million population per year, but
it ignored the possibility that the two genotypes that had
not been observed in patients might also be infected. When
they are included, the predicted prevalence comes up to
something more than four, four and a half cases per million
per year, much less than that predicted by the tissue survey.
Now, reduction of infectivity by the production
process. Albrecht Groener presented a very helpful summary
of spiking studies that had been assayed either for residual
infectivity, and by injection of animals, so-called
bioassays, which showed anywhere from a six to an eight log,
ten, decrease in infectivity, total decrease, or four to over
nine using immunochemical assays for abnormal prion protein.
The conclusion of PPTA was that manufacturing
processes do remove prions, TSE agents, reducing factors of
at least four logs demonstrated for all U.S. licensed
products, but in addition PPTA members do not use either U.K.
or other European plasma to manufacture U.S. licensed plasma
derivatives. Our U.S. recommended donor deferral policies
are in place.
They concluded that the recent report regarding the
patient with hemophilia has no bearing on the safety profile
of products manufactured by their members. The implicated
product 8Y that was put into the record by Dr. Groener was
sourced from U.K. plasma, and an analysis of that process by
the respected risk assessment and management consultant firm
Det Norske Veritas published in 2003 concluded that that
product was produced by a method with a clearance capacity
that was low, probably less than two logs.
The decisional portion of the meeting concluded
with a risk assessment and its modifications presented by
Steve Anderson, who leads the risk assessment section of our
Office of Biostatistics and Epidemiology, with three of his
colleagues. This was a modification of an assessment first
presented in 2006. I have taken 16 slides from Steve's
presentation. I don't have to show them all. If Bill will
wave his hands whenever you have had enough of those, I can
fast forward to the end of my own presentation.
For those of you who are not familiar with the
basic principles of these risk models, what they attempt to
do is to figure out all the things that have to happen for an
adverse event associated with a product to take place.
So for example, for a TSE infection from a plasma
derivative, first the donor would have had to be exposed to
the BSE agent, gotten infected by the exposure. Infectivity
would have to appear in the blood, and as well as being
present in blood it would have to be present in plasma.
Would have to enter the manufacturing process and then
survive the various steps, and these vary from company to
company, the various steps used to purify the plasma
Finally, from that the chance that a given dose of
the product was infected would have to be estimated, and then
based on how much of the product a patient used, you could
figure out what the possible risk was to that patient.
In the case of variant Creutzfeldt-Jakob disease,
it is very important to estimate what the chances are that a
given plasma donor is incubating infection. If you stop and
think about it, although we have had small numbers of BSE
cases detected in this country, so we are not absolutely free
of BSE, but it is a couple of orders of magnitude less than
what has been seen in the United Kingdom. So we reckon that
most of our infected donors in all the patients with variant
Creutzfeldt-Jakob disease in this country recognized to date
will have been infected in another country, the United
Kingdom being the greatest source of exposure and the
greatest source of donors in this country who have been
So a risk model was set up with four modules.
Module two represents the possible prevalence of infection in
our donors, and that largely results from the prevalence
demonstrated in the United Kingdom, which is why those
discrepant estimates of the prevalence turn out to be so
important for doing the final risk assessment.
The next module would be the reduction in infectivity by
processing, and finally the utilization by a typical patient.
To improve the information in each of those
modules, the risk assessing group proposed the three updated
inputs and the three new inputs that we have already listed.
You can't estimate the risk for every single
patient, so what they attempted to do was to estimate the
risk for two groups of patients who are at the highest
exposure, having the highest potential exposure to factor 8.
Those would be those with severe hemophilia A and those with
severe von Willebrand disease. What they attempted to
generate was the potential risk for one year of treatment,
that year being 2002, but they think that the results can be
extrapolated to other years.
They used an analytical approach that set up actual
probabilities for each of the steps needed to produce an
infected product, usually presented as statistical
distributions. I'll show you that in just a second. They
used a probablistic computer based model in which the
likelihood of each step is multiplied, and then a Monte Carlo
method in which the possible exposure values are selected at
random by the computer from the various risk elements.
So for example, the example of a distribution.
This shows schematically how Monte Carlo analysis was used to
estimate the number of IV infectious doses that might be
present in a milliliter of donor plasma.
We know that for hamsters with scrapie that most of
them have at least at the end of infection anywhere from
seven to ten lethal doses of scrapie agent per milliliter. A
few of them have as little as two, and an occasional one has
as many as 27. So what one does is to set up a triangular
distribution in which most of the numbers in the box from
which you are going to be pulling values are going to be near
ten. A few of them will be as low as two, and a few of them
will be as high as 30.
Now, obviously we don't know what the infectivity
is in human blood, but this is the only actual data available
to us at the moment. I might have been more comfortable if
we added a safety factor and said that in vCJD it could go as
high as 1,000, but there is no actual data to suggest that
for any TSE.
As for the percentage of infectious doses that
would enter the plasma, there is only one piece of
information generated by Bob Rohr, Lisa Gregory and
colleagues. That showed approximately 60 percent of the
infectivity was in the plasma. So instead of doing a
distribution there, they took a point estimate. So all the
samples are going to show 58 percent of the infectivity going
into the plasma. It is clear that in most models, it takes a
lot more infectivity, perhaps ten times more, to infect by
the intravenous route as by the intracerebral route. But
some studies show that IV infectivity is almost equivalent to
intracerebral infectivity, and we don't know how to sort the
data so they took what is called a flat distribution. So
those possibilities are presented as equally common in the
box containing those numbers.
When you multiply those together you get not a
point estimate, but another distribution of the possible
intravenous infectivity in plasma, and that goes into the
model to generate the next step.
I'm sorry I spent so much time with that, but it is
I think important in looking at these models to understand
what it is that the risk modelers, among whom I don't count
myself, are actually doing it.
At any rate, they increased the estimated
prevalence of vCJD for the lower prevalence model. They
decided to look at two different prevalence’s. Since you
can't just stick two such discrepant estimated prevalence’s
into the same distribution, they decided to do two. In one
of them they increased the possible prevalence from 1.8 to
four and a half million per million population, the other one
they left the same, adjusting just slightly for age and time
during incubation period when they might be infectious. The
goal of that is to estimate the U.S. prevalence, which would
mean the prevalence among our donors.
This summarizes what I just said. In order to
estimate prevalence, we have to consider the effectiveness of
our U.S. donor deferral policy, summarized at the bottom of
this slide. We assume from a study of donors regarding
history of IV risk that we eliminated about 92 percent of the
vCJD risk, the remaining risk being due both to people who
might have gotten infected during a short stay in the United
Kingdom or people who had made a mistake in giving histories.
Here are the age distributions for the probable
incubation periods for the two kinds of genotypes. This got
plugged into the model. The goal is to estimate from these
numbers the percentage of vials that would contain a dose of
vCJD agent and the quantity of agent in each vial.
There are problems with those, in that each plasm
product has a different purification step with different
potential clearance. A lot of the data we have is not
specific to a product. It is generic. It is not available
for all purification steps in the studies, the spikes and the
agents used all vary. But we believe that, as asserted by
PPTA, that probably all our U.S. licensed factor 8 is
manufactured by processes that provide at least a four log
clearance of spiked model infectivity and probably a good
deal more than that.
So the risk assessors stratified the clearance to
two levels, one a lower level of four to six logs and then a
higher level of seven to nine logs, and from that they
calculated a number of the infectious doses per vial, the
quantity of agent in a vial, and the annual use of patients
to generate an annual potential dose of infectivity per
patient, and hence the risk. They generated final estimates
as we have mentioned for severe hemophilia A and von
Willebrand, taking into account that people on prophylaxis
use more product than people who are treated just for bleeds
in surgery, and that people with factor 8 inhibitors or
immune tolerance also use more, in some of those data they
generated from the CDC survey.
To make a long story short, there was -- only
looking at the worst-case scenario, which would be assuming
the higher prevalence of over 100 cases per million in the
U.K., a four to six log reduction. In somebody with an
inhibitor and immune tolerance, the risk went up from one in
15,000 seen in the lower right to one in 12,000, which
considering all the uncertainties in this model is really a
The risk, assuming a lower prevalence of vCJD, but
going from 1.8 to 4.5, did increase to a modest but
significant extent. Looking at the various scenarios based
on the higher risk, the risk went up anywhere from five times
to 18 times. But as I will conclude in just a second, that
does not affect our final estimate of the real risk to
Similar result were obtained looking at von
Willebrand's disease. There is what I think is a typo in
your hand out that we will attempt to check with the
assessors and correct on the Web.
At any rate, to conclude, the updates to the FDA
2009 model, accounting for susceptibility to the entire
population have not caused important changes between the risk
estimates for December 2006 and June of 2009. The results
for the model indicate that estimates of vCJD risk had
increased using the lower prevalence by approximately five
fold to 18 fold. However, using the higher prevalence, the
risk estimates were similar to those in 2006, which means
that we have not substantially changed our interpretation of
the risk, because our 2006 estimate was based on the higher
prevalence, not the lower prevalence.
Accordingly, now as in 2006, we are assuming that
the current risk from the use of U.S. licensed plasma derived
factor 8 may not be zero, it is not zero, but it is most
likely extremely small.
I would add that this is a work in progress. The
risk model is being shopped out to various professional risk
assessors. Suggestions have been made concerning
improvements to the number of iterations, particularly when
risk is very small. These models tend to be unstable. The
way that that is addressed is by increasing greatly the
number of iterations that are examined.
Finally, what did the committee tell us? Question
one, did they agree with the update. There was no
disagreement. They made some minor suggestions, the risk
assessors did, to improve the model.
Despite finding minimal additional risk, should the
recent report attributing case to treatment with U.K. sourced
plasma derived factor 8 alter our interpretation of the risk,
and the vote was unanimous, 15-no. That is, they did not
recommend any changes to current policies.
Should we recommend additional risk reducing steps
for the manufacture of plasma derivatives? Discussion
suggested that there didn't seem to be significant risk.
However, we should encourage those processes that clear more
spiked TSE agents as opposed to those that are closer to four
Should we revise our warning labels for plasma
derivatives? Yes, because variant CJD can no longer be
considered just a theoretical risk, and that should be
disclosed on the label.
Should we modify our public communications about
the risk? Of course, we can no longer say that no cases
attributable to a plasma derivative have been observed. So
we should include a reference to the case reported in the
With that, I have finished what I wanted to tell
you about our meeting on the 12th of June. If there is time,
I'll be happy to answer questions. Otherwise, during the
DR. BALLOW: With plasma derivatives, was there a
difference in risk with the nature of the plasma derivative?
Like, one patient had factor 8, so is the risk higher
potentially in the model with higher factor 8 than say IVIg,
because it is derived differently?
DR. ASHER: That wasn't addressed in this risk
assessment. However, a pilot study suggests that factor 8
has the highest risk among the commonly used plasma
derivatives. The risk is much less for -- and this again was
not generated by this assessment, but by spiking studies --
the risk is considerably less for IG IV and less than that
DR. BALLOW: So are you going to relabel these
plasma products? Is that what I understand?
DR. ASHER: Am I going to?
DR. BALLOW: Well, is the FDA --
DR. ASHER: You know the way the FDA works. These
things are going to be entered into discussions in the
DR. BALLOW: I think that would be an important
point, because I can just see my patients now going
absolutely crazy trying to make a decision.
DR. ASHER: Well, there hasn't been a case. It is
true that the United Kingdom has warned recipients of at
least large amounts of IVIg that they may be at increased
risk for variant CJD. But taking into account the source of
the plasma used to produce IG IV in this country and the more
robust methods used to produce it compared with factor 8,
remember, cryoprecipitate has almost no reduction in spike
TSE infectivity using a variety of models, has almost no
reduction at all, whereas that is not true for IG IV. And
there has been no case of CJD of any kind attributed to
exposure with IG IV.
But is the risk zero? You know if the vCJD agent
is in the donor, it will get into the blood, it will get into
the plasma used in the finished product, used to prepare the
finished product. So the risk can't be zero if it is in the
raw material. On the other hand, the demonstrated risk, and
I suspect the risk that would be predicted by a probablistic
risk assessment would be considerably smaller than that for
DR. BALLOW: I agree, you can never say never in
medicine. The risk is never zero. But I hope when there is
consideration for relabeling these plasma products, that
there is additional information that is added to the labeling
as you just described, to give some assurance to patients
that the risk is very, very low.
DR. ASHER: Well, it is a risk communication issue,
I agree completely. I cannot tell you what the agency will
do, except that it will be given a lot of consideration, both
the real risk and the communication issue.
DR. HOLLINGER: David, how did they determine that
-- this is a 70-year-old hemophiliac, so presumably he was
not a severe hemophiliac. Maybe he was, but I doubt it. How
did they determine that this was not just an associated --
that it had no relationship to the plasma product that he
received, and that it just wasn't a confounding factor here.
DR. ASHER: There is a possibility, since everybody
in the United Kingdom, at least everybody who had beef had a
potential exposure, the number of cows at the peak of the
epidemic, infected cows, was very great in the United
However, there is a very striking age distribution
of patients. This would only be the second person or third
person over the age of 70, and maybe only one clinical case,
almost all the cases are considerably younger, their late
teens, 20s and 30s, which is one of the differences in the
behavior of this infection.
So looking at that, and looking at the overall rate
of exposure in the United Kingdom, of course the attack rate
is still relatively low for an agent with such wide exposure.
The probability of an infection coming from a food exposure
or from his transfusions with a number of units of red blood
cells from donors who did not come down with variant CJD or
previously endoscopy, that he had seemed to be much less than
the likelihood that he got it from a plasma derivative.
One of the derivatives that he was treated with,
one of those derivatives came from a pool to which a known
infected donor came down later with variant CJD, had donated.
So taking that all together, doing an analysis of the
probabilities, the likelihood that he got it from food
exposure is very small.
This question will be answered unfortunately just
as it was for transfusion transmitted variant CJD, where the
same kind of argument was posed after the first patient came
down. That is, if this bona fide, I think we can
unfortunately expect to see other patients.
DR. HOLLINGER: But as you know, when you have
somebody who has had a transfusion, everybody looks for
transmission. The question is, are they equally looking for
the same kinds of things in people who have not received
transfusion, who are 70, or other things as well with the
same degree of enthusiasm. As you said, they had to make 20
evaluations or at least 20 cuts looking at the tissue
extensively to find it. So I just wonder if they are doing
the same thing.
DR. ASHER: At least at the time this case was
observed, the answer is no, for several reasons. One, it may
not be as low as ours, but the general autopsy rate in the
United Kingdom for people dying of other things, the autopsy
rate is nowhere near as high as it is for people who are
under surveillance. This person had received treatment with
a known implicated lot, which means he was under surveillance
by the U.K. There are some 4,000 other people similarly
situated, not to mention the other patients with hemophilia
who did not receive such an implicated lot.
But the general population is not being followed
with such enthusiasm. And certainly 23 sections of spleen
are not being stained by immunohistochemistry from the
average autopsy, which is one of the reasons why we don't
know with any certainty what the prevalence is in the United
DR. EPSTEIN: I just wanted to come back to Dr.
Ballow's question about product labeling, to clarify the
situation and what we might do.
The current labeling for blood products identifies
a theoretical risk for a transfusion of CJD. That would be
classic CJD. That label was put in place a decade ago when
we were worried about CJD.
What we have learned since is that if there is a
risk at all, it is more likely from vCJD, and we can no
longer say that it is theoretical. So there is a need for a
technical correction to the label.
However, our assessment of the magnitude of the
risk has not changed. That is what makes any communication
about this tricky, because if we change labeling and draw
attention to risk, people will think there must be more risk.
The assessment of risk as you just heard has really not
changed, but we are uncomfortable leaving the labeling as it
is, because it is not telling the whole story, as it were.
DR. KULKARNI: I have two questions. The first one
is, the vCJD was found out on postmortem. Years ago at the
CDC when I was there, we tried to get all the hemophilia
treatment center segment, or get an autopsy on patients with
hemophilia who had died.
Do you think the time has come to resurrect that
between CDC and FDA to make sure that all these patients --
the reason it never succeeded was because of the cost of
autopsies at various institutions, even though I think there
was a central place for these to be done. That is question
Question number two. Recently I have been made
aware that there are patients on Coumadin for sustained
trauma who also receive PCCs, prothrombin complex
concentrate. Even though the exposure is for a short time,
nonetheless it is an exposure. So is there any plan to put
that in your model as to minimal exposure versus long time
DR. ASHER: The CDC has an ongoing study of
hemophilia patients which now comprises 30,000 patients who
have died to miscellaneous causes. I know that none of them
has had clinical evidence of Creutzfeldt-Jakob disease.
As you know, one of the problems one has with
definitive diagnosis of some of these conditions in the
United States is the generally very poor autopsy rate for a
variety of social and economic reasons, that I'm sure you are
all aware of.
The CDC does support to a certain extent through
the National Prion Disease Pathology Reference Center
performing autopsies on patients with suspected Creutzfeldt-
Jakob disease. I can't tell you what their success rate is
in getting those autopsies done.
As for doing ongoing surveillance of people exposed
to various products, particularly people who don't belong to
organizations that would cooperate with the CDC, not that
they are uncooperative, just that the organizations don't
exist, given the current staffing at the CDC, which amounts
to three and a half people for the whole national
surveillance system, I think it is relatively unlikely that
they be able to get that set up.
What they do now, in addition to investigating
clusters and cases under the age of 55, feeling that they
would greatly enriched for variant Creutzfeldt-Jakob disease,
is to do a survey of annual death certificates, looking for
codes that might represent Creutzfeldt-Jakob disease, which
has been pretty good when they did more intensive targeted
surveys of neurologists, neurosurgeons, neuropathologists, et
cetera. The death certificate survey is pretty good at
picking up diagnosed cases of CJD, but there is nothing that
would pick up a suspected infection in a person who had not
shown clinical evidence of CJD. It is hard enough in people
who have clinical illnesses that might be CJD.
DR. TRUNKEY: I think Dr. Hollinger's question was
very pertinent. There was a recent article, I believe it was
in Nature, where some neuropathologists were intrigued by the
amyloid in kuru and the amyloid in dementia. So they did
some studies and they showed indeed that the amyloid in
dementia was attached to a prion. But I don't think all of
those were blood transmissions.
DR. ASHER: This is a current and very important
issue. There are a number of amyloid forming diseases, but
only the spongiform encephalopathies have been convincingly
demonstrated to be experimentally transmitted. That includes
Alzheimer's disease. The amyloid proteins in general are
sticky. They stick to themselves, they aggregate. They
stick to other proteins, they stick to organelles.
So although the phenomenon that you describe may be
important in the pathogenesis of this disease and in other
diseases, I don't want people coming away to think that there
is a great likelihood that some of these other amyloid
forming diseases represent transmissible infections that
endanger the blood supply, in the way that Creutzfeldt-Jakob
disease clearly does.
DR. COLVIN: Considering the spleen studies and the
tonsil studies, is there any convincing data about what the
incidence of disease might be after exposure to the prions,
the vCJD prion that we are talking about?
DR. ASHER: Well, experimental animals, if you want
long enough, if the animal is infected, the animal becomes
ill. Some of the incubation periods are so long that the
animals die of old age related illnesses.
DR. COLVIN: But that may be with much higher
exposure than what we might see in a case of --
DR. ASHER: No, I am talking about animals --
DR. COLVIN: I understand, but what I am saying is
that they may not be the proper model. In that case you may
be overloading the system and they all may get it. But at
lower doses it may not be the same.
DR. ASHER: How many infected people reach the end
of their lives without having clinical disease? We don't
know. As you point out, it would probably depend on the
infecting dose and the age at which someone is infected. We
know that the incubation periods exceed 40 years and perhaps
50 years for kuru. There has been a well documented case of
iatrogenic Creutzfeldt-Jakob disease from a single very short
five-day exposure in childhood to growth hormone with an
incubation period of 38 and a half years. So we know these
incubation periods can run longer. But if you have an
infection that has not become symptomatic and you are hit by
a bus, you will never have the disease diagnosed.
DR. COLVIN: But what I am thinking is, looking at
the epidemiology, that doesn't seem to fit for vCJD. The
peak in the cow population with eight years before the peak
in the human population would suggest that the incubation
period is not -- it may be longer, but it is around that.
So it almost seems like there should be a way to
get an incidence of disease following exposure, based on how
many people we could find prions in in the U.K. versus how
many people actually get the disease. But it seems like it
is hard to figure out what that number is, what the N is.
My other question is, do we know the genotype of
the man with hemophilia in the U.K. who died?
DR. ASHER: They haven't published it. They know
it, we don't.
DR. COLVIN: They haven't published that? Finally,
in the U.S., moving back to the U.S. for a second, was there
any discussion about looking at any kind of surveillance of
the bovine herds in the U.S. at all, even if it is just
random? It seems like at the moment, we are not testing at
DR. ASHER: Oh, no, that is not so. There are
limitations to the Department of Agriculture testing system.
This is not the time or the place to go into that, but there
was a period of enhanced surveillance that picked up three
cases in the United States. Ongoing surveillance is
Is there enough? The Department of Agriculture
makes the point that we exceed the requirements of the World
Organization for Animal Health. I would be more comfortable
if we were doing more, because I can't believe that
considering that we had open borders, free trade with Canada
in live animals, meat products and rendered proteins until
May of 2003 and they have had 16 cases in a smaller cattle
herd, I can't believe that we have had only three cases. But
I have no data to substantiate my concern.
But there is ongoing surveillance in this country.
They screen by an ELISA test, and any repeat reactive by
ELISA they examined a section of the brain stem by
immunohistochemistry. So there is a surveillance program.
Is it perfect? No.
Agenda Item: 2009 A/H1N1 Influenza Update
DR. SIEGAL: Excuse me, but I think we have to stop
discussion on this very interesting topic and proceed,
because we are now 20 minutes late. So let's go on. Thank
you, Dr. Asher.
We now have an update on influenza A/H1N1 by Joe
Bresee, Chief of the Epidemiology and Prevention Branch,
Influenza Division, CDC, and Louis Katz of the Mississippi
Valley Regional Blood Center.
DR. BRESEE: Thank you so much. My name is Joe
Bresee. I am from the Influenza Division at CDC. I am
mindful of the time. What I am going to try to do is take my
15-minute talk and make it into a five-minute talk, to give
Dr. Louis Katz enough time. We are split into two. I meant
to give a talk on the update on H1, and then Dr. Katz is
meant to do the heavy lifting and give a talk on the meaning
of that for blood supply and blood safety. So I am mindful
that that is probably the meat of this talk, and I will cut
mine and hope to make it a coherent story while I cut down.
I'll make the slides available to you. You don't
have a handout, but I'll make the slides available to you,
whoever wants it, by however you want it. So don't fret if I
pass over slides very quickly.
I am going to start here. This is a slide I used
to show in a lot of my talks. It goes to the fact that a lot
of our efforts have been towards avian flu and H5
preparedness and less so on swine preparedness.
But just to show that we are not complete boneheads
at CDC, over the last five or six years we have actually paid
quite a bit of attention to swine influenza as well. Two
things drove that. First is the development of something
called the novel flu A case definition. The CSTE, Council of
State and Territorial Epidemiologists, made it a nationally
notifiable disease. So if a lab in the United States
identified an influenza A virus but couldn't type it as a
seasonal virus, they were obligated to notify CDC and send
the sample to CDC, so that we could characterize it. It was
developed as an early warning or increased certainty that if
a pandemic developed in the United States or was introduced
in the United States from somewhere else, we would pick it up
The second thing we did was the second bullet
there. We equipped state labs, all state labs and public
health labs, with the capability of doing real time RT PCR
for influenza and influenza A, insuring again that if there
were influenza in the community, especially clusters, that
state health departments would pick it up fairly quickly, and
we wouldn't rely on the existing surveillance to pick it up.
So those two efforts that started in 2005-2006 bore
fruit initially. We picked up 12 cases among humans of swine
influenza virus infections between 2005 and 2009, about three
per year for the past three years, compared to about one per
three years for the ten years prior to that.
What this bore fruit in as well clearly is that in
late March 2009, two kids in Southern California both
developed febrile respiratory illnesses. They didn't have
any common exposures, didn't have any exposures to each
other, had no pig contact, had a fairly mild illness. One
had a fever of 103, 104, but both recovered uneventfully.
It was interesting, because both went to the
doctor, both got tested, which is very strange in the first
place, and both happened to be living in sites that CDC was
doing enhanced surveillance for flu. That is the only reason
they got tested. That sample was flu positive, was sent to a
reference lab, was A positive but negative for seasonal types
and negative for H5. It was sent to CDC as planned. We
looked at it and found it was a swine H1 virus. That is not
too crazy, since we had seen significant viruses, but swine
viruses from Southern California was odd, because the pig
populations are so minor. When we found out they didn't have
any pig exposures, it became more interesting. Clearly
when we looked at the viruses more closely, the genes of the
viruses, they had a unique combination of genes from a
variety of sources, such that we hadn't seen either in humans
or in pigs in the U.S. or elsewhere.
To make a long story very short, after the first
couple of cases, there were a couple of cases identified in
teenagers in Texas, then a couple in Ohio, then other states.
So from mid-April or so through now, there has been a rapid
increase in the numbers of states reporting positives, and
the number of positive tests reported.
Last Friday, CDC reported 40,000 confirmed cases,
of which about 11 percent were hospitalized and 262 deaths.
All states and territories have reported at least one
confirmed case at this point.
After we identified the first two cases in Southern
California in mid-April, it wasn't a few more days before
Mexico identified their first cases. They had outbreaks of
respiratory disease prior to that, but the first testing was
positive on April 23 in Canada and the U.S.
Canada two days later had their first case. The
first case outside of America was noted in Spain a couple of
days later, and finally with additional cases and spread in
both North America and Europe. WHO raised the pandemic alert
to phase five, and went to phase six, which is a full
pandemic, on June 11. So in three months from the first
identification of a couple of kids in Southern California,
almost every country in the world has been affected by the
pandemic so far.
This is a map of WHO. You will notice, dark colors
are where more cases occur, light colors or white is where no
cases have occurred, or light yellow in your screen. It is
130, 135 countries that have reported cases so far. The
places that haven't, Sub Saharan Africa, Central Asia that
you see there, are places for the most part that have poor
detection methods and likely won't detected cases very easily
What do we know? We know this, first. This is
hospitalized patients. You see on the left side of the
screen, most of the patients have respiratory symptoms, which
is not terribly surprising, consistent with flu. On the
right side of the screen though, you see that a quarter of
the patients hospitalized presented with either vomiting or
diarrhea or vomiting and diarrhea. This is interesting. We
see this sometimes with kids and flu, not so much with
adults. Gastrointestinal symptoms have been present in 25
percent in both kids and adults with this virus.
The next three slides show the age distribution,
what we know now, and has some interesting features. This is
an age distribution by rate in odd age groups, I admit, but
they are our surveillance age groups, so we are stuck with
them. If you look at the graph, you see that there is a left
shift. The highest rates of confirmed cases have occurred in
the young children and young adults, less so in older adults
and even less so in elderly, atypical with flu. Typically we
have higher rates of illness in school age kids, but not to
this magnitude. This is the age distribution that has been
seen almost anywhere or everywhere that they have looked.
This is the same sort of age distribution, same age
groups, but now looking at hospitalized cases. You see again
the shift to the left. You see a little higher incidence of
hospitalizations among the elderly, but nothing like what we
see in seasonal flu, which I will show you in two seconds.
This finally is deaths, again shifted a little bit
to the right. Most of the deaths occur in young adults and
relatively few deaths in people over 65.
If you look at the deaths, a couple of interesting
things. The median age is higher in all cases, with mostly
adults, but 67 percent of the deaths and about the same
proportion of hospitalizations have a known underlying
disease that confers a known higher risk for severe flu. So
these deaths are occurring less so in young previously
healthy adults, more so in adults with known risk factors for
severe influenza complications.
I just for comparison show the age group of normal
seasonal flu from a study that Bill Thompson in our group did
a few years back. It shows more the classic J-shaped curve,
where high rates in young kids, especially kids under two
years old, but the highest rates in people over 65. These
bars are for hospitalizations. The red line I just put on
there is for deaths, which shows even a more dramatic effect
of high rates in the elderly population, with 94 percent of
all deaths occur generally among people over 65. So the age
distribution for this bug is very different than we would
normally see for flu.
I mentioned that people with severe disease have
underlying risk factors. These are what they are. You won't
be able to see it from the back. Just note that the light
blue is the prevalence of risk factors in H1 confirmed
hospitalizations. The dark blue is the risk factors in the
population background. It just shows what you would expect,
that people with chronic lung diseases like asthma or COPD,
people with diabetes, immunocompromising conditions,
pregnancy. Known risk factors for seasonal influenza
complications also appear to be risk factors for complicated
courses of H1 disease.
The last epidemiologic feature I want to highlight
is also a classic flu epidemiologic feature. That is the
locality of infections geographically. What we see with
seasonal flu is, you get spotty disease here and there
temporally in different places. So far in the truncated
spring wave we have seen just that, with some states like
Texas and Wisconsin and Illinois with relatively intense
transmission of the virus, with some states less affected.
If you look at this, this is the same data
presented slightly differently now for hospitalizations. On
the far left you see the age distribution of hospitalizations
I just showed in the bigger graph. But notice that some
states -- Illinois, Minnesota, New Jersey, New York, Utah,
New York City -- have exceedingly higher rates of
hospitalizations compared to the U.S. average, which of
course means that some states have much lower than the U.S.
average. So in the spring wave we see what we do in an early
influenza season in the United States, some places more
affected than others.
What is happening now? This is a curve of sentinel
providers, which are a group of doctors that report to CDC
each week during the flu season and all year round now, the
total number of patients they see that week and the
proportion of those patients that have influenza-like
You see the initial winter flu season. If you look
in the middle of the graph, the red curve represents the
seasonal flu and green represents the previous two seasons'
curves. You see the nice peak in seasonal influenza in March
and February, followed by our usual decline, and then a spike
with the emergence of novel or pandemic H1 disease, and now a
slightly -- if you look at the far right of the graph -- a
diminishing proportion in these clinics, hopefully meaning
that we are in for a slight lull before the fall increase.
You can look at that by region, and it is going down in
almost every region.
These are our lab data. I just want you for
instant pattern recognition to see two things. First, the
emergence of H1 caused a big bump in lab detections that you
see on the right in the orange, but on the far right you see
the black line, which is the proportion that test positive
for flu going down in the last three weeks. It looks like in
the last few weeks we are starting to see a decline in
overall circulation of this virus that we expected to see
some time ago, but may be finally seeing. But still pockets
of disease that are fairly intense.
Finally, my last virus slide is to remind you of
the fact that this virus continues to be sensitive to the
neuraminidase inhibitors, Oseltamivir and Zanamivir,
continues to be resistant to Amantadine, and those patterns
haven't changed appreciably.
Summary. Once this virus emerged in the human
population in North America, it spread very quickly, not to
all five states, but to all 50 states, globally it spread
just as quickly, and now is worldwide. Some areas clearly
more affected than others. We don't know what that means for
their risk in the fall.
We expect continued summertime circulation of this
virus, especially in areas where people cluster, like camps.
We have seen that. Elderly, seemingly relatively spared so
far, and there are probably immunologic reasons for that.
The virus, while we have seen a relatively mild pandemic so
far like seasonal flu, clearly capable of causing deaths and
hospitalizations, mostly in high risk populations, and the
sensitivity remains the same.
What is next? We expect it to persist. We expect
a fall increase and a fall wave. We don't know if this virus
will be the predominant virus or a seasonal virus will be
predominant this year. Nancy Cox, my boss, always tells me,
if you have seen one flu year you have seen one flu year. So
I don't think we can predict that yet. We know we will have
flu this year; we don't know what it will be. We are looking
at the Southern Hemisphere to give us guidance on what will
happen, and we have learned some lessons from that, which I
am happy to share. Vaccine development is on the way, and
surveillance is continuing.
I'll turn it over.
DR. SIEGAL: Thank you very much. Now Louis Katz,
someone familiar to us.
DR. KATZ: Thank you for the invitation. Some
might say a bad penny, Fred.
I am the Chair of the AABB's Interorganizational
Task Force on Pandemic Influenza and the Blood Supply. We
were tasked three years ago to produce some documents for the
blood community that would allow them to get started on
pandemic planning and to update them as necessary as new
We have produced a background paper, an issues
paper and a checklist for blood organizations to use in
planning. We are getting ready this winter to revise them
based on what we saw happening predominantly with H1N1, and
encouraging blood centers to do tabletop exercises and
whatnot, when we got the ultimate tabletop exercise starting
in late April.
What I want to do today is primarily describe to
you the state of planning in the blood community. I think it
is reasonably reassuring.
First of all, would you stop calling it swine flu?
I am from Iowa, substantially more pigs than people, and
when the price of hogs does this, we can't get potholes
repaired on the lane to my farm. So I would like people to
refer to it by the proper name, pandemic H1N1/2009.
This is probably the important slide, the status of
planning. This looks at 47 ABC centers that responded to a
survey in the 36 American Red Cross blood regions. What you
can see is that 36 of the ABC centers and all of the American
Red Cross regions in fact had plans in place before H1N1. So
that left 11 of the independent centers without plans before
H1N1, nothing like an event to focus the mind, ten now are
planning and one is not, and I am not going to tell you
which one, so don't ask.
The question on our survey was, did you activate
your plan. The numbers are a little odd here, but the
American Red Cross activated in their field units, divisions,
regions and labs. It is more of a unitary organization
obviously than the 74 or 75 independent blood centers that
make up America's blood centers network. So they did
activate their plan. Fifty-five percent of the ABC centers
did, 32 did not. The nothing to activate is that group of
centers, some of which had no plan, some of which started to
develop a plan on the fly and whatnot. So most of the blood
supply was operating under their pandemic plan by the middle
of May, in fact.
The extent of activation I have shown you here. We
asked about communication plans, supply chain assessment,
hand hygiene programs, command and control, assessment of
inventories and use of personal protective equivalent, staff
antiviral provision, the implementation of PPE plans, staff
screening at entry points and antiviral use. These are
percents that show you the extent of activation.
I don't think there is anything terribly surprising
here. The use of PPE was not robust. I think that since we
asked people to be well on the day of donation, we asked our
staff to stay home when they are sick, that should not be
surprising. Very little strong consideration of antiviral
use, although some places have in fact investigated
purchasing stockpiles of Oseltamivir for this indication.
Pretty straightforward stuff. The ARC as you can see
activated command and control, their communication plan, did
their supply chain assessment and hygiene. They actually
issued PPE not for their donor room staffs, but for some
clinical individuals that go into hospitals to do things like
Impact on collections was in fact not much. Of the
centers that responded, 75 percent said there was none. Then
you can see a smattering of blood drives that were lost
across the country.
My center represents one percent of the blood
supply. We run on an average day between ten and 15 mobiles.
So if you multiply it out, you can see that the number of
lost drives is quite small. A substantial number of these
were very early on in the activity, in Texas primarily, where
there was a fair amount of interest early in May. Those lost
collections were recovered very quickly.
This is the percent of respondents to the survey
with unchanged inventories. What we are seeing here is,
everybody's inventories remained pretty stable. In fact,
those with changed inventories, some had more inventory and
some had less. It has until the last three weeks been an
extraordinarily robust blood supply for a number of months,
and no real impact of our dry run during May and June on any
of the important components that we supply.
I have emphasized repeatedly to the membership of
AABB and the people that are using our materials that I think
one of the most important things to do very early on is to be
in contact with public health agencies in order to be sure
that we are on the radar of public health when the discussion
of social distancing, cancelling events and that sort of
thing starts. We want to be able to get messaging from
public health that says, except for blood drives. If you are
well, you can go to the blood center. They are going to
design their drives and design their donation sites to
protect you maximally, and a well-oiled health care system
requires blood on the shelves at all times.
Maybe not as much as I would like, but half or more
of the centers had talked to somebody at the state, local or
county level. When we drilled down on what the contacts were
about, it was basically about being sure that the public
understands when people start talking about social
distancing, that there is an exception built in for continued
Communication initiatives. I think you will see, I
have a couple of slides of lessons learned after action,
everybody has identified consistently that appropriate
communication was a critical issue.
These are the sorts of things that were done.
Overwhelmingly, contact with staff to let them know what we
think the impact of H1N1 is and should be, reminding them to
wash their hands, reminding them not to come to work when
they are sick, those sorts of things. Much less
communication to other stakeholders. I think over on the
right, not a whole lot of communication with hospitals, which
may be worrisome and may be a lesson learned as we start to
see things take off once again in the fall.
PPE. Not a lot of issues. A lot of us have been
very concerned that donors pushing on staff or staff pushing
on donors to use N95 masks or this or that or the other thing
would distract us from more important tasks, like making sure
sick people don't come to drives, sick employees don't come
to drives, and people do hand hygiene. Really very minimal.
I know that in some of the independent centers in Texas, the
staff wanted masks to protect themselves from well donors.
As I said, I'm sure that provision of respirators or surgical
masks to clinical personnel going to hospitals to do
therapeutic procedures is probably more widespread than it
looks here, but not really a lot. I think this shows a
We asked about, would you give donor flu shots if
you had the vaccine. My center has been running a pilot
project for two years now, providing influenza immunization
to committed donors to see how many we could get to take them
and what the effect would be on donation behavior during the
winter. It has been a very successful program. The donors
More interestingly, if we can get pandemic vaccine
from HHS and CDC through the state health departments, would
blood centers be willing to immunize their critical donors.
So that would be platelet phoresis donors, maybe O negative
and O positive red blood cell donors, not as robust a
positive response as I would have hoped. Most centers
already provide to their staff, 90 percent, as you can see at
the bottom. But if we could get pandemic vaccine, 43 percent
that would be willing to administer it to critical donors.
That depends on getting donors designated as a high priority
group. We are working through Jerry Homberg in HHS to try
and get them added to the tier one of potential vaccine
recipients. A lot of liability concern in provision of
vaccine, so I think that is really what is driving a somewhat
Do your hospitals have transfusion triage plans for
blood transfusions in a pandemic. This is my personal
nightmare, that I am going to be sitting in my regional blood
center, situated in the middle of a region that goes 350
miles north to south and 300 miles east to west, telling
hospitals who gets the red cell and who doesn't get the red
cell, when I am not at the bedside assessing the need of the
patient. I really don't want to be in that position. I want
my hospitals to have triage plans that will drive blood usage
down commensurate with whatever impact on supply there is
from lack of donors or lack of staff in the blood center.
At this point, I am not so happy to say that it
looks like there is a very substantial plurality of hospitals
that have not done any planning. The ARC believes most of
their hospitals have plans. I would like to see those plans
in writing. When I asked my 55 hospitals for their plans, I
got none. I got none.
So this is a major concern that I think we are
going to be pushing on. In fact, I am now working on
adapting a shortage plan that was written by Canadian Blood
Services in Quebec to the blood system to the United States
to use as a template to provide through AABB to the hospitals
that might get them moving along on how they are going to
ration blood in the event of a severe pandemic.
H1N1 doesn't look so bad so far, and I don't think
it is going to be a risk this fall. But we are getting a
nice dry run here, and I think it is important to think ahead
on this one.
No supply chain or transportation disruptions
recognized by anybody with the events of this spring. Very,
very few donors calling either the independent centers or the
American Red Cross with post donation information, which has
implications for our ability to study some things that FDA is
interested in like viremia in asymptomatic donors who
subsequently develop confirmed infection. We are not
developing a robust inventory of those plasmas.
I think at last check, Sue Stramer, who is going to
hold these for study, had about ten units. These aren't all
necessarily from individuals with confirmed infection with
the pandemic strain. Very little staff illness.
Lessons learned. The task force is a pretty
diverse group of people from public health and the blood
community and the regulators and whatnot. I think we all
agree that emphasizing command and control very early on is
critical. So it is a little disturbing that more of the
centers in the independent system did not activate their
command and control system. You can always turn it off, and
it is only a meeting a week or a meeting every morning or a
meeting every other day to be sure everybody is touching the
Planning is more important than the plan. Most of
the plans that were written were written for something that
looked more like 1918. This wasn't, and so things that were
written in the plan and triggers that were written in the
plan prove not to be necessary. But the point is that if you
had done the planning it was very easy to back off. If you
had done nothing, you wind up like at least one center that
we are aware of in the Southwest United States that was
caught short of platelets because they had not made contact
with public health early on and didn't have a plan for
bringing in their donors early. They spent about a weekend
short before public health messaging by their county health
department beefed up the supply quite nicely.
The messaging needs to be simple. Keep it simple,
stupid. This is very, very important. I think everybody
that is talking about this blood community and public health
from top to bottom has realized that extraordinarily simple
messaging is very important to allay angst.
There was a fair amount of confusing messaging
coming from a variety of public health agencies, particularly
leading up to the pandemic declaration by WHO. I think that
was an issue. WHO phases were totally geographical incidence
sorts of triggers. CDC has a pandemic severity rating, so at
one point WHO giving one set of messages, and the CDC says,
that is all fine, WHO says that, but it is not changing what
we are doing. The reason it wasn't changing what they are
doing is because they were in part triggering activities
based on perceived severity where WHO wasn't. It was
confusing in some venues.
There were varying donor deferrals around the
world, mainly having to do with what donors could you defer
without actually deferring any material number of donors. So
when Australia deferred donors who had travelled to Mexico,
it didn't cost them much, whereas if you are in South Texas
or California and you wanted to do that at the end of April
and May, it would have had a material severe effect on the
This is a key point. State plans for deployment in
SNS are individual. My main center is on the Mississippi
River in Iowa with access to the Strategic National
Stockpile, through the state of Iowa, and our county health
department, whereas half of our region is in Illinois and
part of it is in Missouri and part of it is in Minnesota.
All the plans for distribution are state level, so they are
all different. The centers that cross state lines will have
to pay attention to varying availability of items from the
Strategic National Stockpile under that circumstance.
We need to clarify PPE provisions, need to decide
what to do. To that end, the Society of Hospital
Epidemiology of America has recently released a position
paper advocating in most settings to use surgical masks, not
N95's, which should make our lives easier.
We need to coordinate supply chain information so
there aren't 140 blood regions all calling the same day
asking about bag supply or calling Abbott and asking about
test supply. We are working on that through an e-mail system
into the group purchasing functions of the major blood
organizations to allow the companies to report to essentially
That is all I have.
DR. SIEGAL: Thank you, Louis. We are now running
about 25 minutes late. I would like to know whether there is
anyone in the audience who is planning on speaking at the
open public hearing, because it will affect our planning. If
there is no one interested in speaking at the open public
hearing, then why don't we do our 15-minute break now and
reassemble at five of. That will give us enough time to get
through the remainder of the program, if people keep on time.
Thank you very much. We are in break.
Agenda Item: Topic III: Informational Session:
DR. SIEGAL: Topic III is another informational
session on hemovigilance. The first speaker will be William
Bower from CDC, talking about the pilot hemovigilance model
of the National Health Care Safety Network. Dr. Bower, are
Agenda Item: Pilot Hemovigilance Module of the
National Health Care Safety Network
CAPT. BOWER: Yes, right here, thank you. Thank
you, everyone. As was stated, I am Captain William Bower
from the Office of Blood, Organ and Other Tissue Safety in
the Division of Health Care Quality Promotions, CDC.
I am going to talk a little bit about our pilot for
hemovigilance, but first a little background on
hemovigilance. In 2005 the Patient Safety and Quality
Improvement Act was enacted. Its goal was to improve patient
safety by encouraging voluntary and confidential reporting of
events that adversely affect patients.
Following that, in 2006 Health and Human Services
Advisory Committee on Blood Safety and Availability
recommended to the Secretary that a national system for
surveillance of recipient outcomes of blood and blood
products was needed. Also in 2006, AABB formed an
interorganizational task force on biovigilance to develop a
comprehensive and integrated national patient safety program
to collect, analyze and report on outcomes of collection and
transfusion of blood combinations and derivatives.
CDC has also had a long term goal to improve
surveillance for adverse events associated with biological
products. Given that, the task force recommended CDC and
AABB form a public-private partnership to develop
hemovigilance surveillance using the National Health Care
Safety Network, or NHSN.
I just wanted to say that the hemovigilance is just
part of a larger planned U.S. biovigilance network
collaborative. Number one, the blood recipients which I am
going to talk about here is just one component, but there is
also a blood donor, which would be the safety of blood
donors, which is being worked on by HHS and AABB as well.
There is also a tissue and organ transplantation surveillance
system for diseases transmitted through tissue and organ.
That is being worked on by UNOS, or United Network for Organ
Sharing, and CDC. There has been a prototype devised for
that. Then for cellular therapies, the fourth component of
the biovigilance network, is to be developed.
As I stated, this hemovigilance is going to be
added on to the current National Health Care Safety Network,
which I will talk a little bit about. It is an Internet-
based surveillance system that monitors patient and health
care personnel safety. It integrates surveillance systems
previously managed separately in the Division of Health Care
Quality and Promotion, which were the national nosocomial
infections surveillance system, Dallas' Surveillance Network
and national surveillance system for health care workers,
which some of you all may be familiar with or have used in
The purpose of NHSN was to collect data from a
sample of U.S. health care facilities to permit valid
estimations of the magnitude of adverse events among patients
and health care personnel, including health care associated
infections and blood and body fluid exposures for health care
personnel, and to be able to analyze and report collected
data to permit recognition of trends.
Additional purposes of NHSN were to provide
facilities with data that can be used for inter-facility
comparison and local quality improvement activities, to
assist facilities in developing surveillance and analyze
methods that permit timely recognition of patient and health
care personnel safety problems, and prompt intervention with
appropriate measures, and to conduct collaborative research
studies with members.
NHSN would allow us to determine if there is a
problem. This would be done through NHSN by comparing of
local rates to benchmarks. Then if there appeared to be a
problem, then groups could get together at the local level or
at higher levels to determine what the problem was. Then
they could figure out some intervention that could be
determined to effect a change. Then NHSN could help to
determine if the changes worked by monitoring progression
towards improvement based on a change in the rates that we
might have seen here.
As I said before, the biovigilance component is
going to be placed within the NHSN structure. You see the
components. They are already in existence, the patient
safety component, health care personnel safety component, and
a research and development component. You can see that the
biovigilance component will sit here. Currently the
biovigilance module is the only portion of this component,
and we are thinking about trying to add some of those other
components into the biovigilance component.
The purpose of the hemovigilance component were
very similar to the purposes that I previously described for
NHSN, but they are specific to blood and blood products. It
can provide the number of adverse reactions and errors and
accidents associated with blood transfusions that can be used
in aggregate data for comparative purposes. Another
component would be that it could enhance at the individual
reporting level to encourage hospitals to report to FDA.
The methodology for the hemovigilance component.
It will have both active and passive surveillance. The
active component, that there be people in the transfusion
services that are going to be trained on the hemovigilance
module. They are going to know the case definitions, and
they will actively look for cases. But there would also be a
passive component, because there will be people in the
hospital that aren't trained in the surveillance system, but
they are going to be aware that if there is a reaction
potentially related to blood products, that they should
It is going to be patient based, in that when a
reaction occurs there will be a review of the patient's
chart, and the health care providers will be contacted about
the reaction. It is a perspective rather than retrospective
that is going to occur while the patients are in the
hospital, and their charts will be reviewed. Comprehensive,
in that it will include all transfusions that occur in the
hospital and not just a subset of a certain group of patients
or a certain unit in the hospital. Its intent is to
provide incident rates for adverse reactions or incident
errors. Initially these would be crude rates that will be
compared from one hospital to another, but as we get more
information, as more data comes in we can do more risk
analysis and incidence rates.
There is a protocol or user's manual for
hemovigilance. It provides rules for the hemovigilance
surveillance so that participating facilities collect
information the same way. It includes case definitions,
other definitions, code lists that will be used in the
hemovigilance module. This content was based on input from
AABB and CDC working group members, approved by the working
group, CDC and other agencies at HHS. When a facility agrees
to participate, they agree to follow the protocol when they
Again, the protocol will provide the rules for the
surveillance and the reference material. There are case
definition criteria for adverse reactions for blood
transfusions that are based on the International Society of
Blood Transfusion definitions, and they will provide
definitions for definitive, probable and possible cases, a
severity grade, and relation of the transfusion to the event.
It is reporting definitions also. The codes are derived
from MERS TM (?) and the test systems that were used at
Columbia University and in the Canadian system respectively.
The forms that are going to be used in the
hemovigilance module are listed here. There is an annual
surveillance survey which gives us information about the
facilities that are participating, a monthly reporting plan
for the hemovigilance component. The monthly reporting plan
is always the same. The reporting plan differs for other
components in the NHSN, but as I said, for hemovigilance it
will be the same.
There is a monthly reporting denominators
basically, the number of units and what type are transfused
each month, and then forms for adverse reactions and
incidents, which I will go over in a little more detail.
You can see there is the top part of the form for
adverse reactions. Each month each participating facility
will enter each transfusion associated adverse reaction,
which is defined as an undesirable response or effect in a
patient temporarily associated with the administration of
blood or blood products. This is based on the case
definition criteria that I described before that is part of
the protocol. It will also list the severity grade that goes
from grade one of no severity, where there is limited
clinical harm, all the way to grade four which would be
death, and the relationship of the transfusion to the event.
Also, information on components and unit details.
Here is a list of the adverse reactions that we
have definitions for that are provided in the protocol.
There are 12 that we have, from allergic reactions all the
way through infections. There is a category for others that
they can specify in the text field if it doesn't meet any of
these. Then there is a final category for a reaction where
there was unknown pathophysiology involved with the reaction.
Here is the incidence form. Each month the
facilities will report each incident associated with blood
products, which is defined as an accident or error that could
lead to an adverse outcome affecting the safety, efficiency
or quality of blood products, or the safety of the blood
All incidents will be included from product check-
in to product administration, and they will be collected on
the date in time and how and where the report was discovered,
the occurrence, the date and time when it occurred, the
incident results, the product action, and also it will be
linked to any adverse reaction that may have occurred based
on this incident by the patient ID.
Here is an example. As I said it is from product
check-in to delivery. There are several steps that an
incident can occur in. This is just an example where the
blood drawn for typing and cross match. Here is all the
potential incidents that can occur with the code that goes
with each one of those incidents.
Another important component of the data collection
is the reporting requirements. As part of this there should
be immediate reporting of complications that may be related
to blood donor or to the manufacture of the blood component
to the collection facility, and also reports of suspected
transfusions related to fatalities directly to the FDA.
Now I just wanted to go over a little bit of the
pilot so far. We have nine pilot sites. They range in size
anywhere from approximately 100 beds to greater than 1,000.
They are pediatric as well as adult facilities, and
university as well as community based hospitals.
Enrollment began May 7, planned for the pilot for
six months. Then after we have looked at the pilot data and
made any changes that may need to be, there will be
enrollment from all facilities that wish to participate
starting in January 2010. There are efforts underway to
develop methods for electronic submission of data. We have
contacted vendors to discuss this. As I said, this will allow
for transfer of data to FDA so that the participating
facilities won't have to double enter data.
Here, as of June 24, you can see that we have nine
sites enrolled. Seven have completed the enrollment process
and five have completed their annual survey, and five have
also completed their monthly reporting plans. Only two have
entered their monthly denominators for the number of blood
products that they have transfused each month. Four have
entered adverse reactions.
Three of them went back all the way to January
2009. After the pilot process is over and we start enrolling
facilities, the calendar year that they enroll in, they will
be allowed to go back to the beginning of the year to enter
data if they wish. You can see here that as part of the
pilot, three chose to go and enter data starting from January
1, and two decided to just enter data from the time that they
enrolled in May 2009. As of June 24, as I said, there has
been a total of 224 adverse events entered into the system,
and that is adverse reactions and incidents. I don't know
the exact number of adverse reactions and incidents, but it
is about a tenfold difference, with the incidents being
tenfold higher than the number of adverse reactions.
If you want more information on the hemovigilance
module, the website is there that you can go and get more
information. If you have any questions or comments, the e-
mail address there you can send a query to.
DR. SIEGAL: Thank you very much. We will now
proceed to the last discussion. This will be FDA's Sentinel
Initiative and CBER's Analytic Epidemiology Branch, Robert P.
Wise, M.D. of FDA.
Agenda Item: FDA's Sentinel Initiative and CBER's
Analytic Epidemiology Branch
DR. WISE: Good morning. I am happy to be here
this morning to discuss the sentinel initiative and a similar
set of activities in our new Analytic Epidemiology Branch at
CBER. What I would like to discuss is some background on
medical product safety surveillance in general, and then
specifically the Sentinel Initiative and the Analytic
Epidemiology Branch's similar work.
This entire group in the audience is probably fully
familiar, but just to set the stage, the evaluation of safety
throughout a medical product's life cycle, whether it is a
drug or a biological or medical device or something else,
really begins with preclinical work in animals, and then in
phase I studies for looking at safety, but of course only in
small numbers of patients. In phase II we have an
opportunity to discover somewhat less frequent adverse
events, but still only things that are relatively common.
Then in phase III trials, for some products we may have only
some dozen or a few dozen patients exposed to a new product.
In other kinds of phase III trials we could have hundreds or
thousands of patients exposed. The level of information that
we can learn about the safety of a new product in a phase III
trial depends a great deal on the size of that population.
Even for a vaccine product, however, with as many
as tens of thousands of exposed patients in a phase III
trial, we aren't usually going to have the power to find, to
reliably detect the least frequent and rare suspected or
actual side effects. In other words, diseases or symptoms or
signs or syndromes that may emerge in these treated or
exposed patients which are subsequent to the exposure and
potentially are side effects. At this stage, of course, if
there are very few observations of such events, it may remain
inconclusive as to whether they are actually due to the
exposure during that clinical trial.
Then approval occurs for many products. After
approval, orders of magnitude more patients typically become
exposed to these new products. It is therefore during the
spontaneous reporting, the passive safety surveillance from
the anecdotal case reports that patients and physicians and
sometimes parents or neighbors or pharmacists or others
submit to the FDA or to the AERS and VAERS program for
vaccine products. It is true that in passive safety
surveillance we often do learn again about tentative or
definite side effects that may be due to these products, but
occurring at much lower rates than were possible to detect
So that is the overall picture. Safety is our
theme. When we identify a concern, a definite or possible
suspected side effect, then we typically try to think about
how we can manage that risk, even though the risk may be as
yet an uncertain risk.
The most common consequence of recognition of a new
risk is probably addition of a few words or a term to the
package insert. In some situations, other kinds of risk
management interventions, particularly in the communication
domain, come into play.
The Food and Drug Administration's Amendment Act of
September 2007 brought new authorities and roles,
responsibilities and activities, to the FDA in the safety
surveillance theme. The FDAAA requires postmarketing studies
and clinical trials to assess a serious risk or a signal of a
serious risk. That is not quite accurately stated. It gives
the FDA the authority to require a postmarketing study and so
on. It gives us the authority, the power now to require
sponsors to make safety labeling changes, where in the past
this kind of revision of a package insert was a negotiation.
It gives the FDA the power, now the authority, to
require sponsors to develop and comply with risk evaluation
and mitigation strategies, REMs. We are now able to develop
validated methods to establish a postmarketing risk
identification and analysis system to link and analyze safety
data from multiple sources, with goals of including an
aggregate -- this is from the FDAAA language -- at least 25
million individuals by July a year from now, and at least
four times that many two years later.
The Sentinel Initiative and to a lesser extent the
Analytic Epidemiology Branch that I will be discussing today,
really illustrates the operationalization of this kind of
activity, to get access to information from large numbers of
patients in HMO or medical claims or other health information
systems, in order to allow us to systematically evaluate
hypothesized risks. We look first at the Sentinel
I think there is a graphic I may have taken out,
but in general the spontaneous reports often present the
hypothesis. We have got a small number of reports of a
particular type of possible side effect reported anecdotally
after licensure, or there may be inconclusive information
from before licensure, and using these large data sets we can
sometimes go out and systematically evaluate the hypothesis
of a relationship between that product and that disease,
syndrome or whatever.
An overview of the Sentinel project. Approximately
a year ago the Sentinel Initiative began to coalesce and take
real substance, so I am very happy to describe it now. It is
the development of an active electronic safety monitoring
system to strengthen our ability to monitor the safety after
licensure for these medical products. It is going to
augment, not replace, our passive safety surveillance and
other systems. It will enable the FDA to get access to data
from these HMO or other computerized health systems through
partnership relationships with insurance companies, health
maintenance organizations or other owners of electronic
The data will remain within the owner's control.
The data will not be delivered, the primary data about these
patients will not be delivered to the FDA. The data holders
will be able to run queries that we request or potentially
other parties may request, and the data holders can decline
to run a query.
What the data holders will deliver to the Sentinel
project or to the requester at FDA or elsewhere would only be
the results for our review, the result tables, the derivative
tabulations. All of this work would follow strict privacy
and security safeguards.
How does the Sentinel Initiative complement what we
are already doing? Well, safety issues can be identified and
evaluated in near real time. The H1N1 vaccine products that
will be rolled out this fall are going to be illustrating
that sort of approach, but it is not blood related. The
Sentinel is going to expand capacity for evaluating safety
issues, by giving improved access to data on subgroups of
population and other kinds of special populations. We are
going to have improved precision of risk estimates due to the
expanded number of populations available for study. In
general, active surveillance we believe is going to help us
to identify an increased risk of a common side effect such as
a heart attack or a fracture that health care providers might
not otherwise suspect to be due to a medical product.
If you think about it, the submission of an adverse
event report to VAERS, the MedWatch system or other
spontaneous passive safety systems, that submission currently
depends on the clinical suspicion or recognition of
unexpectedness on the part of the physician or of the patient
or parent, or whichever party it may be. Using systematic
data sets like HMOs, we can go into the proactive signal
recognition mode, as we use for example in acute safety
studies in phase IV immediately after licensure of a new
product. We can dredge the data intentionally to look for
diagnoses that are occurring more frequently, for example,
within one month after exposure to a new product, than during
the three months period previous to that exposure in that
There are a variety of designs available for this
kind of analysis -- case control, cohort mode, patient self
controlled -- there are a variety of techniques, but the
bottom line is that you are to look systematically and
potentially detect increases in risk that might not have
triggered clinical suspicion because the adverse event itself
is common and therefore not unexpected.
Work in progress within the Sentinel Initiative
includes an initial report that was issued in May a year ago.
There is a long term project concept, and we are just moving
in evolutionary stages. We are currently trying to work out
the nitty-gritty, the how and what of what the Sentinel is
going to be. Using a broad public forum and collaborations,
the FDA is delineating possible governance structures and
functions. We are developing mechanisms for sharing ongoing
related activities, scientific and methodologic projects, and
we are identifying the steps that are necessary or going to
be necessary to insure the strict privacy and security
The infrastructure is based in the Commissioner's
office, the Office of Critical Path Programs. A senior
management team has representatives from each center plus the
Office of the Commissioner for Informatics, Privacy and
Planning. The senior management team insures the involvement
of the appropriate representatives and subject matter experts
or SME's, an acronym we love to hate at FDA. Between the SMT
and the SME's we provide project briefings and updates to the
FDA centers and offices and occasionally an advisory
committee. We assist in prioritizing requirements and system
functions, and we insure that the Initiative is carried out
responsibly and in a timely fashion.
The Sentinel Methods Working Group has agency
epidemiologists from the Office of Critical Path Programs and
for the medical products centers. It provides expertise to
assist in developing scientific and technical aspects. It
contributes experience from activities in the centers to
inform the Sentinel Initiative and CBER's analytical
epidemiology material that I will be describing in a moment
as a strong example of that kind of activity from CBER, that
is helping the larger Sentinel Initiative folks understand
what is feasible and what we have learned from previous
experience. The members can function as agency
representatives to Sentinel related activities conducted by
Between March and October last year there were
multiple stakeholder meetings. Some of you may have
participated. In December there was a public workshop. In
May of this year we had a joint Sentinel rounds to provide an
overview of the Sentinel for the FDA community, and we had
presentations of contract deliverables by four of the initial
In June 2009 there was an internal FDA Sentinel
Summit to begin discussions on how FDA will identify and
prioritize the medical product adverse event pairs that could
potentially be studied in pilot projects and eventually more
systematically and efficiently within the Sentinel
Each medical product center has internal processes
already for evaluating postmarketing safety issues that are
going to have to interface with the Sentinel system for
inputs and outputs, for queries and for results.
There is a Federal Partners Working Group that
shares information and discusses issues related to
complementary efforts being carried out by the various
agencies within the federal government. I won't read all of
these participant acronyms, but ask me if you are not
familiar with any of them. You can see the broad range, from
the Consumer Product Safety Commission, completely non-
medical public health agency, I guess it is the Department of
Commerce, isn't it? And of course FDA, NIH and others.
Collaborations with CMS, the Centers for Medicare
and Medicaid Services, are in progress, the Department of
Defense and the Veterans Health Administration. Several
ongoing projects within the medical product centers are
evaluating potential mortality adverse event signals and
developing active surveillance and statistical methodologies.
The SafeRx program you may hear about in other
contexts. This is a collaboration with CMS to develop new or
real time active surveillance methods using Medicare data.
A variety of contracts have been let for looking at
possible database models, identifying methods to recognize
safety signals, examining the timeliness of medical uptake of
new products, figuring out governance and operational
structural alternatives, evaluating potential data sources
for the Sentinel. This is a project that CBER has
particularly led to find potential data sources for tissue
and blood -- I'm sorry, it is the second number five on
there; the pragmatic data contact is the one that we have
been primarily involved in. The Booz Allen is the more
general for drugs and devices in general. Number seven is
also device specific.
That second number five with the pragmatic is the
one where we are looking to learn more about what kinds of
data may be available in fewer than ten potential vendors of
or sources for data on blood and tissue safety questions.
Then there is a contract on the engagement of patients and
consumers to make sure that privacy is maintained.
There is an updated website that you can look at.
I'm sorry for the long name at the bottom, but if you just
think FDA's Sentinel Initiative and put that into Google, it
will probably pop right up. If you just put FDA Sentinel
into Google, it is probably sufficient.
We are increasing communication. There is a video
that explains the Sentinel Initiative. We have some
collaboration and interfacing with the Observational Medical
Outcomes Partnership, which is a somewhat analogous effort to
experiment and develop appraisals of the value of feasibility
and utility of observational data for identification and
evaluation of safety risks and benefits of prescription
drugs. The OMOP is primarily drug oriented, not
biologically, and yet a lot of the information and methods
that are developed there should have direct application for
The eHealth Initiative is another partner. It is
testing and analyzing the value and utility of using
electronic health information to detect and evaluate drug
safety signals with a variety of data, again including
clinical information as well as claims or claims plus
The eHealth Initiative collaboration is exploring
opportunities to use clinical information in these electronic
databases in two large health information exchanges to
identify and assess known drug adverse event associations
such as Warfarin and bleeding or statins and liver toxicity.
The next steps planned are going to be contracts
again. One will be called the Mini Sentinel. An RFI for the
Mini Sentinel has been posted. The synopsis was posted, and
the solicitation just went out about a month ago.
The Mini Sentinel will develop an actual
distributed data model, a pilot project to assess three
health care data environments with a variety of attributes
for conducting these analyses. It will involve the
development of a planning board, a safety science committee
for operational purposes, an infrastructure for communication
and queries, will identify and prioritize potential medical
product event pairs for evaluation, and will engage
appropriate legal counsel to identify and address data
privacy and security issues.
It will include evaluation of epidemiologic and
statistical techniques to find signals and to refine our
understanding of them and then finally to test hypotheses,
and will synthesize lessons learned, looking at what the
obstacles may be, how we can prioritize those obstacles and
overcome them, and what kinds of costs and time are needed to
remove each of these obstacles.
I would like to turn now to the CBER Analytic
Epidemiology Branch. This is an organization that does not
exist. Legally, creation of a new branch requires some steps
that are beyond my pay grade, but we are working as though it
did exist. It is sort of an acting Epi Analytic Branch at
this stage. Formally you might not see it on the Web, on a
diagram of the FDA's structure because it is not yet
The challenges that the AEB is tackling reflect the
diversity of biological products that require a variety of
surveillance and safety assurance strategies. AEB recognizes
like the Sentinel an open-ended and real time safety
surveillance need to detect unexpected or potential risks at
the earliest possible phase. And of course the FDAAA and
funding technologies offer a lot of promise, we believe, for
development of more robust capabilities to recognize signals
earlier and evaluate them systematically and more
The group that is now assembled as the Analytic
Epidemiology Branch has been working with CMS data,
particularly Medicare data, for some years. CMS data have
about 45 million people enrolled. The great majority are
elderly folks, but there are also about seven million other
people with end stage renal disease or disabilities that
qualify them for Medicare assistance.
Individual health utilization data are available
for 85 to 90 percent of the population covered by Medicare,
which is enrolled in fee for service Medicare. A
prescription drug benefit began in 2006 which in turn is
going to make a new body of data available to us that we are
hoping to work with very shortly.
I mentioned the SafeRx collaboration previously.
This graph illustrates the potential power of the
enormous numbers in the CMS data. What we are looking at
here are cellulitis and urinary tract infection diagnoses
among hospitalizations less than a month after or within the
first month after pneumococcal vaccinations.
The urinary tract infections are the control group.
You can see that that they are pretty constant across this
period with minor fluctuations, but cellulitis diagnoses in
hospitalized patients jumps abruptly after the pneumococcal
vaccination. Those might not be real patients with true
bacterial cellulitis. They might simply be exuberant local
hypersensitivity reactions, but the physician or somebody
thought, we need to rule out cellulitis or prescribe an
antibiotic in case it is cellulitis. Our point here is
simply that we have an opportunity with this very large data
set, although it is cumbersome to work with, to recognize a
potential anomaly, a potential pattern that may indicate a
safety risk for a new product.
Transfusion related acute lung injury in elderly
Medicare claims data, elderly patients, inpatient only, 2007
is a project that we have launched. Our reason was that
there are no previously published population-based TRALI
studies focusing on the elderly and our AEB staff have
previously used CMS data to describe blood product
utilization in this age group from the same data, and so we
felt that this study would be feasible.
We are simply trying to assess the occurrence, the
incidence rate, of this diagnosis in recent CMS data. The
methods are using ICD-9 diagnosis and procedure codes. It is
an exploratory analysis of TRALI rates among the inpatients
who have a recorded transfusion.
We anticipate the ability to break the data down by
type of product received and to compare the TRALI diagnosis
rates to others, in terms of variables like the demographic
characteristics of the transfusion recipients, what clinical
diagnoses and procedures these patients had, hopefully
identify predisposing factors related to the transfusion or
related to the recipient or the donor. We would like to
identify possible preventive modalities or options or
potential preventive strategies, and describe the extent of
severity and mortality from TRALI.
We think it will be necessary to get to the actual
medical records to be able to reliably distinguish between a
diagnosis that is rule-out in the mind of the person who put
it down, and a real occurrence of TRALI. Purposes for
looking at the medical records include the validation of the
diagnosis and further assessment of potential risk factors,
pulling out additional information about the transfusion and
about the donor, and about recipient diagnoses, and so on.
In 2008 AEB issued an RFP, a Request for Proposal,
whose primary goal was to get us access to data systems
analogous to the Sentinel effort, in order to be able to
respond to urgent public safety concerns about the use of
vaccines, blood or tissue products. We wanted to conduct
collaborative mission oriented pharmacoepidemiologic research
to test hypotheses on these kinds of products and their
safety and effectiveness, including those coming out of the
spontaneous reporting to AERS and VAERS. We want to develop
improved methods for rapid detection of adverse events from
biological products in large U.S. population-based data
One example, this first task order under this
contract, is looking at idiopathic thrombocytopenic purpura
after vaccination. We are working with the Harvard Pilgrim
Group and the HMO Research Network based on the Northeast,
where the weather is sometimes better in the summer, but this
year I think we have got better weather.
There is a known association between measles,
mumps, rubella vaccination and ITP. It is not clear yet
whether that association pertains to other vaccine products
as well. We do have spontaneous reports of ITP after other
vaccinations, including live virus varicella vaccine and
others, but also reports in some numbers after vaccinations
that do not include live virus products.
The objective in this study is to assess the
incidence and risk factors for ITP after routinely
administered vaccines and to enlarge and refine the research
network infrastructure for future studies to allow for more
efficient and timely monitoring of suspected side effects,
including near real time monitoring, that we will be
attempting for the H1N1 products.
Other kinds of AEB collaborations are looking at
herpes zoster vaccine safety in the Medicare population, MMR
and anaphylaxis in the Defense Department setting, rapid
cycle evaluations in CDC's Vaccine Safety Datalink, H1N1
vaccine safety evaluation in the military. The European
Medicines Evaluation Agency and Canadian regulatory
authorities and Health Canada are working with us to
coordinate pharmacovigilance activities and information
exchanges, and we expect along with CDC to collaborate with
WHO and others to foster the technical capacity for H1N1
vaccine adverse event investigations.
That is as far as I am going to go today, but I'll
be happy to take any questions before I disappear, if that is
DR. SIEGAL: I was just going to invite Mark Ballow
to ask his question. Thank you very much, Dr. Wise.
DR. BALLOW: That was very impressive. It looks
like this initiative dovetails very nicely with the
Administration's goal to incorporate electronic health
records into physician offices. Have you been working with
the EHR companies to develop modules that might help your
DR. WISE: I probably shouldn't answer that
question, because my 23-year-old son has just taken a job
with one of the important software manufacturers, Epic
Systems, outside of Madison, Wisconsin. But seriously, he
doesn't have anything to do with this work so far.
We are not working directly right now with the
electronic medical record vendors. We are working actively
with standards development organizations. Particularly -- I
was talking with Bill earlier -- the HL7 organization, one of
its standards is called ICSR, individual case safety report.
An ICSR can be a submission to AERs, a suspected side
effect. So we are trying at FDA as part of the eGov
initiative to use and work through and help develop standards
of various sorts that can help us do our mission more
effectively. The ICSR format is one of those.
The vendors can then recognize that there is a
stable pattern, a stable file format or message format or
whatever that has been formulated and identified, and then
they can build linkages from their internal data architecture
to that output file and facilitate more efficient
That sort of answers the question?
DR. BALLOW: Yes. I was just going to say, because
physicians of their time commitments, sometimes things fall
through the cracks as far as reporting these adverse events,
particularly in practice who you report it to. You have to
go through a government agency, I guess, the CDC and FDA, who
may watch other mechanisms.
But it just occurs that the electronic health
record would be a mechanism whereby, if physicians are
already incorporating that into their everyday practice, that
might be one mechanism to more easily -- for the physician to
have access to this kind of Sentinel program.
DR. BIANCO: Dr. Wise, we heard your program. We
heard Dr. Bower's before. How do both connect? We didn't
hear you mention hemovigilance, for instance.
DR. WISE: I had to pay him a lot of money to ask
that question. Thanks for a very good question. We haven't
discussed today the enhancement, the successor to the AERS
program. I have mentioned AERS, the Adverse Event Reporting
System. The successor program is going to be called MedWatch
Plus. It is going to be a unified portal or mechanism, web
based, for people to submit adverse event reports for any
kind of medical product regulated by FDA. Actually beyond
medical, also foods, for example, veterinary products.
Now, the MedWatch system according to current plans
is going to be able to receive an adverse event submission
from an individual person. So what we are envisioning, still
somewhat tentative, details are still being worked out, is
that the AEBB, NHSN, CDC hemovigilance adverse event report
or incident report, after submission to the NHSN, will then
present the reporter with an option: would you also like to
submit this report to the FDA. Or there will be some
equivalent like a check box that the reporter can choose to
click on if he or she wishes.
The key distinction is that the NHSN reporting is
anonymous. Once the data are received by CDC, there is
absolutely strict protection of the reporter's identity and
certainly the patient's identity. The internal CDC folks
will know the characteristics of that hospital and what
hospital it came from, so that they can extrapolate these
data to national patterns by hospital size and geographic
region and so on. But the identity of the reporter and of
the patient can't be transmitted outside of that internal
NHSN technical office.
Therefore, voluntary submission also through FDA
would allow the next screen to pop up, which is the MedWatch
Plus portal screen, pre-populated with all of the details
that have already been filled in. If this information is
coming from an electronic record, now the patient's identity
and age and birth date and place of residence can be pulled
in electronically. But any missing data can then be filled
in by the reporter and with another click the record comes
over to the FDA. This allows the FDA to follow up when and
where it is necessary to ask further questions or to
recognize that there is a cluster forming of some sort in a
certain area. So we think that this partnership is going to
be very productive.
Thanks again for asking that question. I would
have paid you a lot of money if I had thought of it in
DR. SIEGAL: Are there any more questions? In that
case, thank you very much.
DR. WISE: You're very welcome. Thank you.
DR. SIEGAL: Is there any further discussion or any
new issues that the members of the committee wish to raise
before we adjourn?
DR. FREAS: I would just like to remind the
committee members that we have a two-day meeting scheduled
for November 16-17, which will be our next BPAC meeting.
DR. SIEGAL: Thank you all for attending and
participating. This meeting is adjourned.
(Whereupon, the meeting was adjourned at 12:00