. WILLIAMS: Thank you, Dr. Brown.

Good morning. May I have the first slide, please? As mentioned by Dr. Jacobs, those of us in the blood collection community left the December meeting of this group with a mandate to conduct additional survey research to try to fine-tune the data with respect to donors who have traveled to the United Kingdom and make use of those data to estimate both the impact on supply as well as the potential impact on a theoretical variant CJD risk to support the deliberations of the Committee at this meeting. So we have, in fact, done that.

The survey that I'm going to describe to you today was supported by numerous organizations. In fact, most of the data collection activities were supported by the American Red Cross research and surveillance program known as ARCNET. And the analysis activities were supported by the REDS Coordinating Center under the sponsorship of the National Heart, Lung, and Blood Institute. And this took place after the data was in hand at the Red Cross to meet OMB requirements.

In addition, in the planning phase and throughout, we worked in association with the American Association of Blood Banks and membership of American's blood centers to try to provide a coordinate effort. And I think you'll see that evidence throughout the course of the discussion.

So, first of all, the objectives of the survey, as stated, are to estimate U.S. donor travel and residence in the United Kingdom for defined time periods relevant to the BSE epidemic; secondly, to correlate travel and residence in the U.K. with other donation variables to estimate the impact of deferral on blood safety and availability.

Next, please. In conducting this survey, we enlisted the help of numerous blood centers. And a survey was conducted in whole blood community donors in 12 sites. And, in addition, we also had data collection from the military and a one-center collection specifically from apheresis donors.

To summarize the geographic areas where the study took place, I'll mention these 12 sites by their general metropolitan area. First is American Red Cross in Baltimore-Washington area; Detroit area; Los Angeles; Boston; Connecticut; Atlanta; San Francisco; Oklahoma City; New York Blood Center; Blood Bank of San Bernadino, California; Memphis; and Miami.

Next slide, please. Because time was limited, as were resources, we had to conduct the survey on a fairly simple basis and in discussions of our initial planning committee reviewed several different techniques for potentially collecting the data and after this discussion came to the conclusion that clearly the best way for us to collect the data was through the anonymous mail survey mechanism that had been in use in the REDS study for several years now.

I won't go into the reason for this decision unless someone wants to discuss them, but we did end up concluding that a mail survey would be both the fastest and most representative and economical for us.

We chose random samples representing one month, about ten percent of the collections, for one month at each of the participating blood centers. This in most cases came from the January '99 donations at the blood centers, in one or two cases came from the December '98 donations because the blood center was in the midst of changing their computer system and couldn't get the '99 sample.

We designed a one-page front-and-back anonymous mail survey to be read by optical scanning. This was sent out in a single mailing with a compelling cover letter explaining without graphic detail the purposes of the study and asking if donors would please respond.

And this was sent out just about five weeks ago. It was the last week in April that this was sent out. As of yesterday, our responses were 9,346 out of 19,000 mailed, for about 49 percent. And I suspect by the end of the week -- we still have surveys coming in -- we will probably hit the 50 percent range.

For a single mailing of a mail survey, that isn't a bad response rate at all. That's really pretty good. And we know that donors typically are pretty good responders to this type of data collection.

The presented data, we had to cut it off at some point to do the analysis we wanted to do. The analysis covers 8,666 donors as of May 24th. And we actually did three different runs of analysis: from early, midpoint, and end of the available data. The results were quite consistent. We really didn't see changes in the data over the course of time receipt of the surveys.

Next slide, please. The question categories included demographics of the donors. These were quite simple: age, gender, first time versus repeat donor, and educational level. We gathered a donation history for the donor, how frequently, how many times they donated in the past ten years.

We asked the primary question about travel or residence in the United Kingdom. And we added into this the Republic of Ireland. For a couple of reasons, that decision was made. One is because most people, blood donors, as an example, really do not understand the details of the split between Northern Ireland and the Republic of Ireland, and we didn't want to confuse the issue. Secondly, there is certainly geographic proximity to the U.K. And, thirdly, after the U.K., it is one of the highest countries with reported BSE.

It's arguable whether we could have done that, whether we should have made that addition or not, but I think the change in the overall travel figures are probably quite minor.

We split the travel into two different periods. This was at FDA request. We separated into intervals between 1980 and 1989 and separately between 1990 and 1996.

We also asked questions about beef ingestion during the period of travel in the U.K. And because historical questions about food ingestion are typically suspect, we asked about beef ingestion in the past year just to get a prevalence value for beef eaters.

In addition, we included in this analysis a further measurement of deferrable risk estimates from United Kingdom travelers. This didn't come from the traveler survey, which is going to form most of the talk, but this is by subsequent analysis or further analysis of the 1998 REDS survey, which was described at the second meeting. And I'll get into the deferrable risk values that we used near the end of the talk.

It really wasn't practical to try to remeasure these deferrable risk values. It would have made a much longer, much more extensive survey. and we chose not to do that.

Next slide. The question asked is: Did you travel to or live in the United Kingdom (England, Scotland, Wales, Northern Ireland, Isle of Man, Channel Islands, or the Republic of Ireland) between 1980 and between 1989; and separately, as a separate question, between 1990 and 1996?

Next slide, please. The summary results for this travel question, between the period of 1980 and 1989, 15.5 percent of the donor population reported travel; between 1990 and 1996, 13.4 percent; for the total period of 1980 to 1996, 22.6 percent.

Now, keep in mind that these cover different year intervals. So that probably is the major explanation for the difference in percentages.

The range for this 22.6 percent value, as you remember from December, there was quite a bit of geographic variation in the travel prevalence. For this measurement, it ranged from 11.2 percent all the way up to 30.5 percent for that 17-year period.

Now, just for compatibility with the '98 survey, we did an unadjusted figure for U.K. travel per year given that these are different yearly time periods. For the '80 to '89 time period, it is about 1.6 percent; 1.9 percent for the later period; and 1.3 percent overall given that there was some travel by donors in both time periods.

We compared that with a similar figure from the 1984 to '90 measurement made in the 1990 REDS survey, which is 1.7 percent, really right between those two figures.

So I think to the extent that we can validate the responses that we're getting, there is compatibility between the measurement in the '98 survey, which only asks U.K. travel as an ancillary question, and this survey, which asks it as a primary question.

Next slide. I want to mention briefly we do have breakdowns for the intervals for the two separate time periods, the '80 to '89 and '90 to '96, are included in the handout. And I do have a slide. I wasn't planning to go into it unless the discussion comes up, but it is available if you want to discuss those time periods separately.

Some of the demographic correlations we analyzed by logistic regression analysis just to consider their influence independent of other variables. And you can see that in terms of the age breakdown, setting the 17 to 29 age as a reference category, clearly travel increases with increasing age. And it's really the seniors that have the highest rate of travel, almost three times the likelihood of travel.

I think you will see an interesting -- next slide, please -- correlation there as well when we look at the gender analysis because, in fact, setting females as a reference category, females travel a little more than males. And this might be to the senior phenomenon again, where females are known to have longer survival and may, in fact, do traveling and produce a higher representation there.

In terms of first-time donors, similar to what we presented in December, those individuals who are first-time donors tend to have less travel, both because they're younger and probably have less financial means to do so.

Next slide, please. In terms of education, again, setting the low value as the reference variable, you can see that college-educated and college graduates have four to five times the likelihood of international travel or travel to the U.K.

Next slide, please. Now, looking at the individual intervals between those time frames, these are pooled data between the two time frames. We used a rather simple rule supported by midpoints of the intervals. And that is if people traveled during both of the time frames, we added the two values and put them to the next category if the intervals were the same or if one of the intervals was longer, we took that as representative of the total period of travel. And that is supported by looking at the midpoints of the intervals.

So for travel exceeding one day, that matches the overall travel to the U.K. during that time period, 22.6 percent. I mainly wanted to show this slide to show the tightness of the confidence intervals around these estimates, generally within a half to one percentage point all the way down the line.

This difference can be shown better on the next slide, which is a bar graph, same numbers, just shown differently. For the one to three-day period, 22.6 percent of the respondents traveled to the U.K.; four to ten days, 19.7 percent; eleven to thirty days, 11.8 percent; one to four months, 4.9 percent; five to eight months, 2.0 percent; nine to eleven months, 1.3 percent; one to three years, 1.2 percent; three to five years, 0.7 percent; and five years or more, 0.4 percent. Obviously these are cumulative looking at the longest time period first.

The next slide, please. Again, these are the same data but here fitted to a line graph. And what we did is run an equation to match this line. And you see this is a power equation. The r2 of the formula explains the data well, about 97 percent.

This is the formula that derives from it, and we can use this on the next slide to actually plot percentage of donors who would be affected by specific time periods that might be of interest. I think the obvious ones we chose here would be intervals that might serve as a source of discussion for the Committee for potential deferrals.

These include looking, for instance, at the right-hand side, for two years, that would affect 1.1 percent of the donors. For one year, 1.6 percent of the donors; nine months, 1.9 percent; six months, 2.4 percent; three months, 3.7 percent; one month, 7.0 percent; and one week, 16.3 percent of the donors would be affected. Now, this is not the blood supply. This is individual donors. I am going to have some blood supply calculations a little bit later.

Next slide, please. Now, one of the things we were asked to do as well was to -- let me make one point before going on to this slide. I do want to make it quite clear that the numbers that I am assigning quantitative values to are a one-year calculation. That's similar to introduction of a laboratory screening test, albeit a very nonspecific laboratory screening test.

These types of deferrals have multi-year effects. It is a very difficult model to build, but this is certainly more than a one-year effect to lose percentages of donors of this type. So please keep that in mind. It's a very complex formula to model, however.

Looking at the prevalence of beef ingestion by donors during the U.K. travel and currently, we asked the question whether they recalled eating beef during their U.K. travel for the two separate time periods.

For the '80 to '89 time period, 74.2 percent reported eating beef in the U.K.; 7.0 percent, no. And, as you might expect, 18 or close to 19.0 percent reported that they didn't know or didn't recall that value.

For the '90 to '96 time frame, the difference is kind of interesting. Seventy-two percent reported they ate beef. Fifteen percent said clearly they did not eat beef.

So maybe recognition of some of the early phases of the BSE epidemic kept some people away and they knew that, in fact, they had not eaten beef in the U.K.

It could be that or it could be the more recency of the travel and they had better recall. That's difficult to distinguish. And 13 percent didn't know for that time period.

Now, comparing that with -- you remember I mentioned that we wanted to get an overall prevalence for beef eating by asking those who had eaten beef in the last year. We are very surprised to see the figure that came out of there. Ninety-six point six percent of respondents indicate that they had eaten beef in the past year.

So I think it's useful to compare the validity of this type of answer. I think it bears out that it's tough to get a historical dietary question answered.

Next slide. Now, we didn't really set out to measure the impact of a potential deferral on different types of donors, but it did become evident that there would be some interest specific to apheresis donors and specific to the military.

We know in general and supported by REDS data that apheresis donors are significantly older and more educated than whole blood donors, and I'm not going to show the data. And higher travel rates would be expected.

Now, in collaboration with Ron Gilcher and Jim Smith in Oklahoma, we did run a small survey of 200 apheresis donors in Oklahoma and compared those donors' travel histories to the overall blood donor values.

Two hundred were surveyed. And apheresis donors had 20 percent higher 1980 to 1986 U.K. travel rates, at 13.3 percent, than whole blood donors, 11.1 percent.

So it's only one center, but I think it gives a rough estimation that apheresis donors are going to be hit a little bit harder than whole blood donors for some of the demographic reasons.

Next slide, please. We also included military donors in this survey with the collaboration of Lianne Groshel. These actually were all Air Force donors because Lianne felt that it would be the Air Force that would be most likely to have been stationed in the U.K. because of the base locations.

Military donors are more mobile on average. And, therefore, U.K. travel would be expected to be higher. And certainly if there was a base there, it would impact things.

Unfortunately, we had a fairly low response rate in the military. They sent out I think 300 questionnaires. And we got 25 back. So it was only a 12 percent response rate.

Given that, 8 of 25 indicated that they had lived or traveled in the U.K. or Ireland during the 1980 to '96 time frame, so again a little bit higher but some real broad confidence intervals around that one.

Next slide, please. This is one of the same slides I showed in December, and it serves as a basis for some of the blood supply impact discussions, I think.

The generally accepted figures for the U.S. blood supply, which AABB provides, is that there are 13 million allogeneic units collected, made into 22 million components annually. These derive from eight million donors and are given to four million recipients.

From this total number of donors, we know from the large Red Cross ARCNET database that 32 percent of these donors are first-time donors. Now, most of you -- well, some of you will recall that the proportion of blood from first-time donors is 20 percent. That's donations versus donors.

If you look at donors, it's actually a little higher. It's 32 percent. So, using that ratio, we can break the donor base down into 2.6 million first-time donors and 5.4 million repeat donors.

Next slide. Extending those calculations a little further, annual loss of units donated by first-time donors can be calculated as percent first-time donor travel loss times 1.3 units per year times 2.6 million first-time donors.

Annual loss of units donated by repeat donors is percent repeat donor travel loss times 1.8 units donated per year -- these two figures also derive from the ARCNET database -- times 5.4 million repeat donors.

Next slide. If you take that math and simply compact it, you can determine or convert a deferral prevalence into an impact on the blood supply and lost units from the blood supply by multiplying deferral prevalence times 11.9 million. And that gives estimated annual lost units.

If you then divide that by 13 million, the annual supply, it gives the impact on the percent of the U.S. supply. So for five years at a 0.3 percent deferral, 35,700 lost units, 0.3 percent of supply. Those are the basis of the calculations that go into some of the figures that I am going to show you coming up.

May I have the overhead, please, and shut the slides down for a moment?

DR. PRUSINER: Can you tell us why the number, the year number, is 1984 to 1990?

DR. WILLIAMS: That was from the 1998 REDS donor survey. And that figure was taken from the Lancet review paper, the two-part review, that mentioned '84 to '90 as the likely period of highest theoretical dietary risk. So that's how we referenced it.

You probably understand the happenings in Britain better than I do to correlate with those dates, but that was related to that Lancet review paper.

DR. PRUSINER: I see. Okay. All right.

DR. WILLIAMS: Now, one of the things we did -- and I have to thank Peter Lurie for getting us started on this -- is to not only look at loss of the donor base but the impact on a theoretical variant CJD risk coming into the U.S. from donors who have traveled in Britain.

What we have done here is calculated the theoretical risk associated with U.S. blood donor travel to the U.K. or Republic of Ireland during 1989 to 1996 as measured by the survey, taking the intervals, computing the midpoint of that interval and the number of persons who traveled as reported from the survey, and using that to calculate person-days, this as a representation of potential dietary exposure to BSE on the assumption that duration of travel can be related to magnitude of theoretical risk. That's a basic assumption in doing that.

So if you run these calculations, for the one to three day period, we have 494 person-days; four to ten day period, 4,600 person-days; and so forth. You can start to notice a larger number here as the time period gets longer. And I think that's going to provide some meaningful discussion.

The last time period here I'll mention, the question that we asked was greater than five years. I calculated the interval as 5 to 17 years because the overall interval that we were measuring was 17 years. So if someone asks more than 5 years, in fact, it could have been, you know, 5, 10, 16, 17 years. So I think it's valid to use the midpoint for that interval as well.

Calculation of the theoretical risk of donors traveling for those intervals was then figured by adding up the person-days and dividing those into the person-days for each of the intervals. So for 252,804 total person-days, we divided that into the interval person-days and got a percent contribution to the total. You can see again that a lot of this is clustered into the higher interval time frames.

Then we just added these cumulative in descending order to support today's discussions. For the greater than five year time period, 49.2 percent of the risk would be related; adding to that the three to five year interval, 67.1 percent; one to two years, 77.8 percent; and so forth. And you will see these graphically in a moment.

One of the graphs actually used residual theoretical risk, as opposed to remaining theoretical risk. And the figures for that are shown here.

Maybe I'll ask: Are there any specific questions to this calculation -- because I think this is fairly basic -- from the Committee?

DR. HUESTON: Did I understand correctly that you used the mid-range of your five to 17 years?

DR. WILLIAMS: That's correct.

DR. HUESTON: Is that pretty surely a skewed segment of your curve?

DR. WILLIAMS: I wouldn't say that inherently. For someone who has been there at least five years, chances are good they equally likely have been there ten years.

I am not sure, you know, I could address whether there is a bias there or not. I think it is a topic for discussion, but I wouldn't inherently assume that there is.


DR. EPSTEIN: Alan, is that supposed to be 1980 to '96? It says "'89."

DR. WILLIAMS: I'm sorry. Yes, you're right, Jay. That's 1980 to 1996. It's a computer error.

Next slide, please. Okay. This is the first graphic that I'm going to show utilizing these data. I'm not going to keep this up long because there's a better one to follow.

What this is, this shows residual theoretical risk, shown in the red line, for the full time period of consideration. In other words, for here this is the midpoint of the five to 17 year interval and then going down plotted against the percent of blood supply lost. You can see that as a figure which goes down slowly until you get near these lower travel intervals, where you start to lose more and more donors. I wanted to show this mainly to give the total picture and sort of the area under the curve out here that really explains a lot of the data here.

So next slide, please. This I think probably would constitute a good working slide for some of the discussions. It's the same graph, but it's really zoomed in on probably the more likely deferral periods that the Committee might want to consider.

For instance, looking at percent of blood supply lost, the figures are labeled here. For a one-year deferral, it would be an impact of loss of 1.5 percent of the blood supply; for six months, 2.2 percent; three months, 3.4 percent; one month, 6.4 percent; and one week, 14.9 percent. And then that can be compared with the values for theoretical remaining variant CJD risk.

We tried to mathematically assign a function to this line, and it just didn't work well enough. So I think you're probably just as well off trying to do a visual comparison where needed.

For the one year time period, that equates to about 22 percent residual. For example, a six month time period, that equates to about 13.0 percent residual; for three months, almost right on the same point, 6.7 percent residual; one month, about three percent; and so forth.

Obviously, as you can see, most of the theoretical risk is accounted for by the time you get to about one year. And then the efficiency declines, and you start to lose more and more donors as you get to a later time period. So I think that is going to be an important consideration.

CHAIRMAN BROWN: Alan, why are the time points different top and bottom?

DR. WILLIAMS: Because the bottom one is based on what we thought would be likely discussion points for deferrals. The top one, in the absence of being able to assign a function to that line, we didn't try to exactly plot those points. And the fact that we didn't, in fact, put them up there and label them simply was an omission. But you can extrapolate to those time points.

Next slide. Now, the request was also made to consider impact of deferral on traditional risk. This is a difficult issue to get a handle on. We have been working with a quantity known as deferrable risk for several years within the REDS donor surveys. And, really, given the rarity of post-transfusion HIV and hepatitis nowadays, it's almost getting impossible to measure. The studies in which you would actually measure this empirically are getting so expensive that you can't really conduct them anymore.

So we have defined this factor, known as deferrable risk. It was described at the December meeting. There was a copy of the JAMA paper there, which described it in detail.

Deferrable risk by the 1993 measure, as reported in the JAMA report, dealt primarily with the parenteral and sexual behavior risks, most important related to HIV and hepatitis transmission. The figure at that time overall was 1.86 percent prevalence in the accepted donor blood supply.

In the 1998 survey, we added another variable. We wanted to maintain continuity by being able to look at this one over time, but we introduced a deferrable risk '98. This includes an additional ten questions that would serve as a deferrable basis for donors but are perhaps less important in terms of magnitude in relation to transmissible disease than some of the others but still, nonetheless, are deferral questions and questions that some donors may not answer correctly, things like body piercing, tattoo, whether or not a donor has spent more than 72 hours incarcerated, birth in an HIV Group O endemic country, et cetera. There are about ten questions.

Next slide, please. If you look at donors who traveled to U.K. during the time frame and donors who did not remember, this is from the '98 survey, not the latest travel survey, so it's '84 to '90 period -- deferrable risk by the '93 measure is 2.1 percent, dead even in both groups.

Don't infer from this that deferrable risk is rising in the blood supply. There are other factors involved. For instance, we had different blood centers participating in the survey. So until that analysis is done completely, don't draw any conclusions to the '93 report.

The deferrable risk by the '98 criteria is 7.2 percent in the travelers, 7.7 percent in the non-travelers. And that comparison is not significant at all.

However, if you compare these values to first-time donors, who would need to fill in the gap were you to defer long-term repeat donors, deferrable risk for '93 in the '98 survey is 4.3 percent. That's highly significant. Deferrable risk for the '98 value, 13.3 percent. In both cases, the odds ratio in repeat donors is about half that of first-time donors. And it's highly significant.

I think it's important to mention that in some of the work done by Mike Busch and others looking at the lower-sensitivity HIV assay, to apply that to incidence of HIV, they, in fact, found a similar ratio, that first-time donors had a twofold higher likelihood of HIV incidence. So I think these data are very compatible with the lab-based findings between first-time and repeat donors.

Next slide. Trying to convert these risk estimates into something meaningful is a difficult job because there are estimates provided for HIV and hepatitis C, hepatitis B transmission. They're now all very rare. We have just started moving into a period of nucleic acid testing for hepatitis C and HIV. So it gets very theoretical to try to measure an impact.

To try to apply these deferrable risk values in that equation, you're figuring if you defer donors and have to replace two of them with first-time donors, you're doubling the risk in 2.0 percent of the blood supply, which is a 0.4 percent overall increase in risk, which is well within the confidence intervals of the current estimates for HIV and hepatitis C risk factors. So I think trying to quantitate that precisely just really becomes an exercise in numbers.

I'd like to end with mentioning the limitations of survey data collection. These estimates are reproducible and have been remarkably reproducible since the 1990s, but everything is based upon self-report. It's subject to potential differential response rates in the survey and to differential reporting. The accuracy has not been validated by other independent measures, but we know that between surveys, things tend to be very consistent.

Next slide. I want to make some specific acknowledgements here. First of all, the participating blood centers. In many instances, the blood centers cost-shared on this project and did not reflect their costs back to the Red Cross. So we thank them for that the PIs and the staff.

Ron Gilcher and Jim Smith for suggesting and conducting the apheresis survey at the Holland Laboratory. Melinda Tibbals coordinated the survey. Ed Notari and Roger Dodd helped with the analysis. Ed Westat, Dannie Ameti, and Kevin Watanabe were instrumental in helping with the survey.

We got some specific help from Committee members. I'd like to mention Paul Brown, Peter Lurie, Larry Schonberger, Jay Epstein, and Mary Beth Jacobs and the Planning Committee, made up of AABB and ABC and Red Cross representatives Celso Bianco, Richard Davey, Kay Gregory, and Steve Kleinman. I'll end there and be happy to take any questions.

Thank you.


CHAIRMAN BROWN: We now have theoretically a half-hour or so to ask questions. Bob?

DR. ROHWER: I just want to make sure that I understood you correctly. Your summary in terms of the replacement of donors lost is that it would be an insignificant increase in risk. Is that what you concluded?

DR. WILLIAMS: On a statistical basis, yes.

CHAIRMAN BROWN: Alan, I had a question. I may have missed a beat. On the slide which is the zoom-in slide, --


CHAIRMAN BROWN: -- the same one I asked a previous question about, what precisely do the figures on the top half of the slide represent? That is to say, the legend says, "Theoretical residual risk."

DR. WILLIAMS: Right. If you look at the single sheet that is part of the handout, the calculation that is there, over in the far column, it's the risk associated with each of the periods. And assuming that there is a deferral and that portion of risk removed, that last column represents theoretical risk remaining. And those are those figures.

CHAIRMAN BROWN: Okay. So an alternative legend would be cumulative person-days?



DR. SCHONBERGER: If I had traveled to the U.K. between 1989 and 1992; that is, one year in '89 and three years in the period 1990 to 1996, for a total of four years, how do I appear on the graph?

I would have had checked off one to two years for the earlier period and three to five years in the second period. How would I appear on this table of calculation of theoretical variant CJD risk?

DR. WILLIAMS: You would be in the three to five-year period.

DR. SCHONBERGER: I would be in the three to five-year period because you just take the longer --


DR. SCHONBERGER: -- period when there's a --

DR. WILLIAMS: If the periods were the same, we moved it to the interval. If one was shorter, one was longer, we took the longer period. There is a little error in doing that, but, again, --

DR. SCHONBERGER: As you point out, that --

DR. WILLIAMS: -- the way the data was set up, that's really the only way we could do it.

DR. SCHONBERGER: Right. And you pointed out that the number that overlapped was relatively small, as I recall. Is that right?


CHAIRMAN BROWN: I would suggest that the Committee not get too exercised about the distinction between these two time periods. I was in London last week in front of the Transmissible Spongiform Encephalopathy Committee. There really is no basis that can be defended for dividing this period into two. The consensus is that the earliest years of the 1980s and the latest years covered in this survey, '94, '95, '96, are less risky than, say, something from about 1983 through 1993.

I think we can spin wheels all afternoon or morning if we really worried about these two time periods because of exactly the kind of question you raise.

Suppose you visit for six months in 1984 and revisit three months in 1989. How do you stack it up? The fact is it's probably not important to make this distinction.

DR. SCHONBERGER: No. In response to that, I agree with you. I was really responding, in part, to your recommendation initially to the group to ask the question for the one period and just ask them how long you stayed.

I think that he's given me enough information to satisfy me that that error that has been introduced because of the breakdown of the two periods is not going to be that significant.

I would be more worried, however, about that five to 17 year group given that it seems to account for about half the risk if I'm reading this correctly. I would think it would be extremely unlikely for somebody to be there the entire 17-year period. And, yet, those 31 individuals are accounting for, as I say, 49 percent of the person-days of risk.

Is that right? Is that the right interpretation?

DR. WILLIAMS: That's right.


DR. ROOS: First, I wanted just to congratulate Dr. Williams and his colleagues who carried this out, because we had given you this mandate some months ago. And we do have the data that was requested. So I think we appreciate that information.

DR. WILLIAMS: Thank you.

DR. ROOS: Second, I had some questions about the military donors here. And I don't know whether we're going to pick up later with any speaker about that.

CHAIRMAN BROWN: Yes. There will be a presentation during the public hearing. It's really a separate issue.

DR. ROOS: Well, then maybe I just want to ask you a couple of questions: first, whether there is any information about the breakdown with respect to the time periods that those respondents in the military spent in U.K., as you did with the nonmilitary; and also whether you could tell us a little bit about the military donors. I mean, is that a separate group? I was a little bit confused here about where those donors go and how they're handled.

DR. WILLIAMS: I'll answer what I can. I think there are probably people in the audience who can answer some of the military-specific questions much better than I can.

Looking at the intervals, I don't have the data with me, but I think what you're getting at is: Of those military donors, were they all up in the one, three, five-year time frames?

And clearly not even a majority were, but I think overall the time periods tend to be somewhat longer than the whole blood donors. And of the 12 who reported travel, I think there were 2 or 3 up in that longer time period. So disproportionately they were up in the longer intervals, but that's an important consideration, yes.

In terms of the characteristics of military donors, I know these were all Air Force donors. The military maintains its own blood supply and has in comparison a relatively small pool of donors that it uses. And I think perhaps Captain Rutherford or anyone else who would like to add more should do that.


DR. SAYERS: Thanks.

Alan, I was interested in that slide you showed on the apheresis donors from OBI. You know, that's certainly a group of individuals who are becoming increasingly important as far as transfusion support for patients is concerned. And they certainly do donate at a frequency much greater than the 1.3 units a year or 1.8 units a year that the other donors that you referred to donate at.

Did you have any separate calculations for what the loss of pheresis platelets might be?

DR. WILLIAMS: We did not take the calculations that far. I'm sorry. I think to produce the data to support that type of analysis, we probably would need to do more than one blood center and get a reasonable geographic distribution. I think what we got is just a window into the likely comparison, but we probably would need more blood centers.

DR. GILCHER: Alan, with reference to the same point on apheresis donors, if you looked at the loss of donors in Oklahoma specifically because the data which you showed was specifically apheresis donors in Oklahoma, I believe it's somewhere around 4.6 to 6 percent on the whole blood side, which shows, then, that from the apheresis standpoint, it's much, much higher within our center.

Now, whether that would be true in other blood centers, I don't know, but that was what I noted about your presentation. That piece of information hit me in that this would be probably four to five times higher among our apheresis donors in our particular area than among our whole blood donors.

DR. WILLIAMS: So you are saying the impact on lost donations would be four to five times higher?

DR. GILCHER: I am saying that the impact on donors would be very high. And then the impact on donations would even be astronomically higher because this particular group of donors averages 12 to 18 donations per year as an apheresis donor. So I'm saying the impact in the apheresis donor base in terms of donations I think will probably be very, very high.


DR. McCULLOUGH: Alan, back to this table, with the 50 percent of the risk essentially being allocated against those who were in the U.K. between five and 17 years is based on -- you arbitrarily chose the midpoint of that range to do the calculation. Is that correct?

DR. WILLIAMS: Yes. It's not entirely arbitrary. That's standard procedure when you're working with an interval like that, yes.

DR. McCULLOUGH: Did you choose some shorter periods within that interval and rerun these? If you had used only six or seven years, instead of the 11 years, for that interval, it would reduce the contribution of that group to the total risk and, therefore, would increase some of the other categories. Did you look at the effect on the contribution of risk from some of these shorter stays if you reduced that?

DR. WILLIAMS: We did not do that. It would have the effect you referred to, but I guess it got back to the earlier question: How representative is the midpoint? I think if there was a strong argument that most of the five to 17 year group were closer to five than the 17, then that would be justified, sure. But it would have an impact if you changed that analysis point.

CHAIRMAN BROWN: Dr. Rutherford or Captain Rutherford, would you like to say something here?

CAPTAIN RUTHERFORD: Well, I guess I'm the only DOD contingent here. Speaking on the DOD for the Air Force as well as for the Army and the Navy, we chose the Air Force. I think we sent out 167 surveys, and 25 came back. Out of those 25, 8 had responded that they had been in the U.K. Three responded that they had been in one month or longer. So that's about a 12 percent.

The DOD collects around 85 percent of its blood usage from active duty personnel. So that would greatly impact us. The thing there, too, is we did not take into account the time periods in '83 through that period of time when we had a large contingent of 300 and some thousand Army individuals in Europe who probably went to the U.K. for some period of extended time. So that wasn't even considered.

The DOD opens all of its bases to the Red Cross and the American Association of Blood Banks and ABC members. So as they come back to the States, the large contingent of Air Force personnel at Langley Air Force Base, Keesler Air Force Base, Lackland Air Force Base in San Antonio would probably greatly impact the donations collected in those areas by civilians. The civilians do rely upon us a lot for blood donations.

CHAIRMAN BROWN: Captain Rutherford, in the collection of blood by the military, so long as the donors are active military personnel, are those donations used exclusively for the military or is there any mixing with the civilian blood supply while they're still in active duty?

CAPTAIN RUTHERFORD: There is a lot of mixing of blood within the DOD with civilians. They rely on us for blood at times as excesses are in the system. And then we rely on them also when we need emergency units.

All of our donor centers or OCONUS overseas are FDA-licensed. So the blood that's used OCONUS is collected from military active duty or civilian DODs or dependents who are on base and are used only on base. Only in emergencies do we use non-DOD blood of OCONUS.

CHAIRMAN BROWN: Alan, any of the areas which were surveyed, did they include areas in which there was a substantial military component?

DR. WILLIAMS: The one I'm aware of, again, is Oklahoma City. I know we do have a fairly large military contingent there. And it generally lowers their survey response rate because they don't like to return surveys.

How many of their donor base are comprised of military base individuals and what bases are I think Ron could probably answer.

CHAIRMAN BROWN: I think it might be important in view of the possible bias to this survey with respect to military contributions to get as much information about this as we could.

In other words, what I'm hearing is at least a possibility that the proportion of military donors in the actual real life nationwide donation program for civilians would be substantially affected if the proportion of military were not reflected as a true proportion in view of the extensive military presence in the United Kingdom.

Can anybody illuminate that problem? Bob?

DR. ROHWER: The other thing is: If the military doesn't like to return surveys, are we biasing the survey because we're not getting answers from people who have done a lot of travel?

CHAIRMAN BROWN: What do you think, Alan? Are these legitimate questions or --

DR. WILLIAMS: Yes. I think Bob's point is a valid one. I think that when the data become available, you will find that percentage of military donors as a proportion of the total U.S. blood supply is going to be really quite small, but I don't have a figure for that.

CHAIRMAN BROWN: Just a second, Bob.

Yes, sir?

DR. TABOR: Ed Tabor, FDA. I would just like to second the last comment and emphasize the importance of not basing decisions too firmly on portions of studies that have either very small returns of surveys.

I mean, the military one is not only small, but you don't know, at least we don't know here, the demographics. I mean, were the ones who returned them officers and the others enlisted and so forth?

Also, the data on apheresis is based is very, very small numbers in one location. And I think those are two areas where we really seem to have very little data at present. We should be very careful about drawing conclusions from them.


DR. HOLLINGER: Alan, you did a really wonderful job with this, and I know how difficult it was to get all of this data in such a short time. Nevertheless, 50 percent non-response rate is still pretty high. And a lot of the data is being made upon that.

Do you have any idea at all about anything about the demographics of the people that responded versus the demographics of the particular areas from which they were collected -- they're going to be different in different areas of the country -- to give some confidence that these are similar to what one might expect from donors in general?

DR. WILLIAMS: I don't have the specific demographics of our return rate. I wasn't able to get them yesterday based on your question. Typically in all of the surveys we have done, we have gotten about a ten percent lower than mean response from under 25 age donors and first-time donors and generally about 10 percent above the mean by older donors and repeat donors. And sometimes survey return rates go up as high as 80 and 90 percent when you hit older repeat donors.

So without having the numbers, I would say probably this survey follows the same pattern. And, if anything, there is probably a little over-representation of the older, higher socioeconomic repeat donors.


DR. LEITMAN: I'd like to return to the apheresis issue for a moment and to remind the Committee that greater than 50 percent of all platelet components in the U.S. are collected from apheresis donors, who, as Dr. Gilcher stated, are generally regarded as the most safe type of donor because they donate so frequently -- 12 to 18 times per year is our institute's estimate as well for our center -- and to state that an increasing proportion of non-platelet components, both red cells and plasma, are being increasingly collected by apheresis technology.

So the impact on the U.S. blood supply of deferring, of adding additional deferral criteria to apheresis donors, is much larger than that on whole blood donors. And I think I would like to see that data because the impact will be so huge, and that I think you need a larger number of apheresis donors surveyed to get that, of course.


DR. ROHWER: On this issue of robustness, another way -- this word you don't like, I know -- to look at that is to do the same calculation on the maximum and minimum values in each one of those year bins.

On the preliminary data that you provided a week or so ago, I did do that. And it doesn't vary that much. It just shifts the two tables by one interval one way or the other. But by the time you get up to around six months, you're still talking around 70 to 80 percent, 70 to 85 percent effect in terms of removing exposure.

The other thing I would like to note is that I think it was mentioned earlier that we have the Canadian experience to refer to. What strikes me is this distribution of exposure is almost exactly the same as the distribution that was obtained in the Héma Québéc study by Dr. Germain, who I think is here, and which again adds some credibility to the idea that this type of distribution of travel exposure among blood donors is fairly consistent across North America.


DR. BURKE: I want to return to the question of the qualitative difference between the travelers and the non-travelers. Your conclusion was that there would not be any change in the risk of the donor pool, that the donors who had been in the U.K. versus those that had not had no change in their other risks, their other deferrable risks.

But it seems that you would have to replace the repeat donors with a number of first-time donors so that there would not be a negligible impact on the donating pool but that there would be at least a temporary burst of a window there where you had to have more first-time donors who would have higher potential risk.

So my conclusion would be that it isn't a total wash, not a total even risk, but there would be at least a window of a period. Is that a reasonable conclusion?

DR. WILLIAMS: Yes. I think I was careful to say, yes, on a theoretical basis, there is a doubling of risk and if you use that particular cutoff, that two percent would have to be replaced by first-time donors.

I think the message I would like to make is that if you did the analysis, the difference would not be statistically significant, but on a theoretical basis, yes, you're bringing more risk in by bringing in more first-time donors. But it's not measurable given the current resources.


DR. EPSTEIN: Alan, can you speculate at all how these data might be extrapolated to source plasma donation? It's a big missing piece. If there is any thought later in the day that we should consider policies differently for whole blood and transfusable components versus plasma and, therefore, plasma derivatives, the impact on the source material, the availability of plasma for fractionation would need also to be understood. I know it was simply a limitation of what could be done quickly that you went to the whole blood centers, but we still would be faced with a question about source plasma.

DR. WILLIAMS: I think to the extent that the source plasma collectors can supply demographics of their donor population, we could probably do some rough calculations.

Just without having seen any data, my guess is that they tend to be a younger population, probably less financial resources to do international travel and on that basis may well be impacted less, but I think we're going to get a more accurate answer here.

CHAIRMAN BROWN: Just a second. Susan, did you have anything to add about source plasma in terms of demographics? Maybe not.

DR. LEITMAN: No. I was referring to repeat volunteer, non-paid donors. As you know, source plasma donors are paid. It's a completely different group of individuals and risks. And deferrable risk in those donors is markedly higher in paid plasma donors.


MR. REILLY: Hello. Jim Reilly, American Blood Resources. I'm not sure that I'll clarify any more than you have, Alan, but it's worth commenting at least.

We did make an attempt to collect some data rather quickly. Unfortunately, we didn't have a well-organized structure to do it in, such as Alan. So we didn't really end up with anything that we thought was particularly meaningful.

There are some differences in the population. There are some similarities. The deferral risks are not as different as you might think, but there are differences in the demographics. They tend to be a younger population. The socioeconomic status is admittedly different.

So we have some different travel patterns. And I think on the surface, we would suggest that the percentage that are traveling outside the United States is probably lower. But, similar to platelet pheresis, the frequency with which they donate is substantially higher.

So I think the overall impact of the supply is probably not meaningfully different, clearly would be, but I'm not sure that's really the big area.

I think if we were concerned about anything, it's the developing of a series of additional criteria would invoke all kinds of extra logistical problems with regard to look-back criteria, which probably have just as big a supply impact, if not maybe larger than the actual deferral criteria or the first question that you asked.

The other question that we would begin to raise is: Looking at the total list of questions, are we slowly but surely eroding the quality or the efficacy or the entire screening process? And what is the ultimate risk impact here?

The other difference in plasmapheresis is that the ultimate end product goes through a further manufacturing process, which has additional viral clearance steps. So there are questions about whether the ultimate product risk is the same or different.

I don't think that probably addresses the question that you had, but there are some criteria or questions that we have with regard to how this would be best implemented and what the ultimate value would be.


DR. McCULLOUGH: I'd like to go back to the assumption that the donors lost would be replaced by first-time donors. As I recall, Alan, the survey was done directing the question to people who had donated within the last year.

DR. WILLIAMS: Directed to folks who had donated within the past month.

DR. McCULLOUGH: Month. So it's well-known I think that many -- it seems to me the figure of 50 percent or more runs in my mind -- people who donate the first time do not donate again or there would be a huge cadre of previous donors out there who would not have donated within the past year. They would be the most susceptible to being retrieved and reentered into the donor pool, rather than trying to find brand new first-time donors.

So I think we shouldn't necessarily jump to the conclusion that donors that are eliminated if new criteria are adopted would have to be replaced by people who had never donated previously.

CHAIRMAN BROWN: Are records kept by the blood donor centers about such patients so that they could, in fact, be contacted? Dr. McCullough?

DR. McCULLOUGH: This will vary by different blood centers. Most blood centers would have a list of donors that would date back three to five years approximately. Some might be more. So the names of those previous donors who haven't donated within the past year would be available.

CHAIRMAN BROWN: Yes. As a practical matter, one has the option I guess of putting an ad in the paper, you know, "You donated blood. Won't you please donate again? We need it" or sending a postcard to the individuals.


DR. NELSON: One of your earlier tables showed the heterogeneity in the various blood collection centers visiting to the U.K. in various times. From that, I remember there was about a twofold variation from one to another, saying what you present --

DR. WILLIAMS: That's right, from the low to the high.

DR. NELSON: Can we, then, assume that this being an average curve, an individual blood bank might have -- the curve might be twofold higher if we did a cutoff of one month, six months, or something like that? What's the degree of variation between individual blood banks in that overall curve that you --

DR. WILLIAMS: Well, the range where the mean is 22.6, the overall is 22.6 percent, the range is 11.2 to 30.5 percent for that total time period. And, if I recall from the December presentation, that tended to cluster more around the urban areas, particularly New York City and San Francisco. So yes, the numbers would be markedly higher in some areas.

CHAIRMAN BROWN: I would like to introduce a qualification to being completely smitten by the notion of person-days, as opposed to time spent, just to point out that person-days depend on the notion that 100 hamburgers, for example, distributed in any way will have the same risk.

That is to say, if one person stays 6 months and eats 100 hamburgers, it is the same overall risk as if the 100 hamburgers were eaten by 100 different people who visited United Kingdom for one day.

That assumes that the risk of a single exposure is the same as the risk to multiple exposures in a single person. And that's an assumption. We do not have any evidence bearing on that question.

So that, for example, if a person is twice exposed within a week, he may be more susceptible to an infection than two different people exposed once during the same time period. So cumulative person-days may not be as attractive a way to analyze risk as it may be appearing.


DR. SCHONBERGER: I was wondering if there was any laboratory evidence to support what you just said. I noticed in the human growth hormone situation, there is actually epi data that would support the cumulative.

I mean, the one risk factor was lengths of treatment. But people often interpret that as meaning that with the longer period of treatment, you're more likely to get the one hit that you need.

Is there any laboratory data pertinent to this issue that you're aware of?

CHAIRMAN BROWN: I think the recent PNS paper with lemurs if I'm -- this is embarrassing because I'm an author.


CHAIRMAN BROWN: That would be one little point. One lemur was sacrificed. And I think, but I'm not absolutely sure, that the lemur that was sacrificed and was positive had two doses. Stan may have some additional information.

I'm not aware of any systematic study, although it has been talked about for some time, of analyzing this particular question of cumulative risk; rather, in the nature of radioactivity. People talk about it. It is expensive to do.

Do you have any information, Stan?

DR. PRUSINER: No. I just want to point out that I don't know of any data where you would take, for instance -- if we add more than one infectious unit, of course, the incubation time goes down. But that doesn't really help answer your question.

The question is: If we took a fraction of an infectious unit, gave that to an animal, and then later gave another fraction of an infectious unit, would we ultimately get one infectious unit? And would the animal get sick? I don't know of a study like that.

CHAIRMAN BROWN: I don't think so. I agree. I don't know. And, of course, that is a very relevant consideration since an infectious unit defined by intracerebral inoculation is well-known to be less. You need more than one intracerebral infectious unit when it is given by a peripheral route, which includes the oral.

So it really boils down to: Do these things hang around and somehow get together? You know, a fifth of an infectious unit once a day for five days is like medication. At the end of the five days, if you've got one infectious unit, that's enough to do it. There just is no information that I know of.


DR. PRUSINER: The one thing we do know, which is unpublished, is that there is clearly clearance from the brain. So it complicates all of these kinds of measurements. And we have no understanding of this from a peripheral route.

That's not helpful. I'm sorry.


CHAIRMAN BROWN: We'll, then, go on.


DR. CLIVER: This is a continuing concern of mine that we haven't really looked very much at the ingestion route. But, having said that, there is a lore of foodborne disease that differentiates between infectious agents and intoxicants, which is highly dependent on Avogadro's number.

With infectious agents, if all of the infectious material required to produce an infection is present in one ingested unit, then if only one in a million of those succeeds in inducing infection, still you can either feed a million units and get an infection in one person at high probably or probably you can feed one of these units to each of a million people. And probably one out of the million will eventually get infected.

With intoxicants where you require high redundancies of whatever your disease agent is, the dynamics are very different. And there cumulative exposure becomes much more significant.

CHAIRMAN BROWN: Did you want to add anything, Bob?

DR. ROHWER: That was the point I just wanted to make. And, from my point of view anyway, I don't see how virus or an infectious pathogen can take into account the fact that there are other infectious pathogens in its neighborhood.

They don't gang up like that. I mean, my guess is that's the exact same question as the pooling question, which we have debated endlessly. And I don't think it ever will be resolved without an experiment, and the experiment is an expensive one.

CHAIRMAN BROWN: "Endlessly" is a little too strong a term.

Other questions for Dr. Williams? We are sort of moving into the other term of the equation now and shifting off what Dr. Williams' major subject was. That is my fault.

If there are any further questions about the impact on blood supply? Yes, Dr. Sayers?

DR. SAYERS: Alan, from what you have calculated about number of donors that have visited Britain, the number of whole blood volunteers there are nationally, the number of transfusion recipients there are annually, I wonder if you could pitch this the other way around and have a graph which would show the likelihood that a transfusion recipient received blood from somebody who had traveled to one of these areas, taking into account the number of transfusions that individual had received.

DR. WILLIAMS: That would be interesting to do. I'm not sure I can do it in my head because there are numerous factors involved that -- to answer your question, no, we haven't done that, but it's probably something we could do.

CHAIRMAN BROWN: Other questions for Dr. Williams?

(No response.)

CHAIRMAN BROWN: It is a little ahead of schedule. It is ten past 10:00. I think we can have the break now and return to our schedule in 15 minutes, please.

(Whereupon, the foregoing matter went off the record at 10:11 a.m. and went back on the record at 10:32 a.m.)

CHAIRMAN BROWN: We now have three further communications before lunch. The first two are by invited guests from the United Kingdom. On the schedule, we have a talk about demographics of bovine spongiform encephalopathy, U.K. regulatory decisions, and the time course of new variant CJD.

I rather like the name Christl. You'd rather be called Christl, would you?



CHAIRMAN BROWN: Okay. Dr. Christl Donnelly?