Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
II Guarding the Blood Supply
The Retrovirus Epidemiology Donor Study: Rationale and Methods Thomas F. Muck I am going to briefly outline the rationale and methods of the Retrovirus Epidemiology Donor Study (REDS). I am going to provide no data except to look at the enormity of the database that has been accumulated. A request for proposals (REP) was published in 1988 by the National Heart, Lung and Blood Institute (NHLBI), because of the unknown infectious risk of transfusion, concern about HIV variants, the need to understand human T-lymphotropic virus (HTLV) infection, a need to create repositories to examine emerging infectious agents, and a need to evaluate emerging technologies to detect these agents. The study will run from 1989 to 1998, and it is likely that it will be extended beyond then. The purpose is to monitor the safety of the nation's blood supply through studies of the epidemiology of known agents, essentially retroviruses, among volunteer blood donors. What I am going to share with you today has recently been published in Transfus~on.37 Blood centers were selected on the basis of the quality of their proposals. There are four high-risk centers and one low-risk center, risk being defined as areas having a high background prevalence of AIDS. The REDS high-risk participants are the Red Cross Greater Chesapeake and Potomac, the Red Cross Southern Michigan, the Red Cross Southern California, and Irwin Memorial Blood Center. Oklahoma was considered to be low risk. The coordinating data center was Westat, Inc. The scope of the study is $40 million over the course of 5 years, and it is considered the most complicated study that NHLBI has ever launched. The structure is governed by a steering committee that has two investigators from each of the participating centers and the committee is chaired by an independent center director, who happens to be me. Subcommittees of this steering committee developed the proposals and protocols that we have been following over the course of the study. We have a continuing and an ongoing close relationship with the Centers for Disease Control and Prevention (CDC). 37Zuck, TF, RA Thompson, GB Schreiber, RO Gilcher, et al. (1995). The retrovirus epidemiology donor study (REDS): Rationale and methods. Transfusion, 35: 944-951. 27
28 attitudinal issues. BLOOD AND BLOOD PRODUCTS: SAFETY AND RISK There are five major REDS components: monitoring donors, creating a general repository, creating special repositories, studying a cohort of people infected with HTLV, and surveying donors concerning the prevalence of certain behavioral and Three substudies are pursuing these five objectives: establishing and maintaining a serum and cell repository, following the cohort of HTLV-infected donors and patients, and conducting the mail surveys directly. There are three REDS repositories: the General Serum Repository, the General Leukocyte and Plasma Repository (GLPR), and a Special Repository. Serum taken from donors routinely is selected at random based on a complicated sampling methodology devised by the statisticians at Westat. Samples in the special repositories are: donations that are repeatedly reactive for HTLV, but for which confirmatory testing is unclear; donations from sex partners of HTLV-positive subjects and their controls; donations that were repeatably reactive for HIV- 1 but which produced indeterminate Western blots; donations that were repeatedly reactive for HIV-2 but HIV-1 negative; and donations that were repeatably reactive for HIV but of unclear etiology. The HTLV cohort study uses a case-control methodology to investigate HTLV risk and transmission factors and to define the natural history of HTLV infection. Other than some studies in Kyushu, Japan, we know little about the natural history of HTLV infection. One of its outcomes, T-cell leukemia, is so infrequent that it is difficult to estimate how often it occurs once an infection has been identified. There were also few data on symptoms related to HTLV-associated myelopathy and tropical spastic paraparesis short of those two diseases themselves. They are actually identical conditions but defined in different parts of the world by different names. We were essentially searching for unexpected clinical outcomes, because in 1988, when this study was designed, the literature was unclear on the clinical outcomes of this infection.
GUARDING THE BLOOD SUPPLY 29 The donor mail surveys were done in waves. The survey subjects were selected by sophisticated statistical techniques that randomized the selection of survey recipients. The resulting sample was enriched with donors from certain types of populations that we knew had high infection rates. The idea was to ascertain risk behaviors of the donors. The survey was stratified by several variables: center, race, age, sex, zip code, birth year, etc., trying to focus again on those people among the donor population who were more likely to engage in high-risk behaviors. This is randomized in a biased way in terms of enriching it for younger people. We mailed out about 64,000 questionnaires; the response rate was about 70 percent. The most important outcome is the extensive databases that are available for use today. We have demographic information Mom the donors of 4.9 million donations. The General Serum Repository contains 500,000 samples stored in multiple vials which are now the property of NHLBI. The GLPR has both leukocytes and plasma so that we can look at any kind of virus that may be only intracellular, such as HTLV, in which the genome is not free in the plasma. We have 546 HTLV-infected donors and patients who have been enrolled and who are being followed longitudinally for the presence or emergence of symptoms. The information in this database is immense and enormously valuable. To date, the surveys have found that more risky behavior is being encountered than predicted: up to 1 percent of people report prior drug use, sexual activity with a previous drug abuser, etc. The extensive data also permit incidence calculations. One of the difficulties we have had over the years is dealing with prevalence. We know what our current rate of infection in the donor base is, but in the past we could not look at incident infections, that is, in how many people per year does it occur? Those calculations are now being made. They contribute to the decision making regarding HIV antigen testing in a negative way. They show that antigen testing is probably not of great public health value. We have established an orderly process to access the database. An application is sent to the REDS Publication Committee, which looks at the request to see whether we want to provide a series of samples or process a request for data or data analysis. The committee then decides whether to recommend to NHLBI that the database or the samples be accessed for the requested purpose. NHLBI can veto the use of data or samples. This is particularly important for samples because once you have thawed them and then refrozen them, the test results gained from those thawed and refrozen samples have less credibility with certain kinds of testing technology. Thus NHLBI must carefully guard the repositories. The contributions of REDS to date have been important, and I have outlined just a few here that are by no means all inclusive. We were also involved with a team dealing with the idiopathic CD4 lymphocytopenia (JCL) crisis. Within days of the report of JCL, REDS formed a task force with CDC.
30 BLOOD AND BLOOD PRODUCTS: SAFETY AND RISK CDC had a meeting that included two Nobel laureates within 3 or 4 weeks of the announcement in Amsterdam. We used the laboratory facilities of REDS to work with the technology for CD4 counting. There were some recommendations from the panel convened at CDC that perhaps we ought to screen for CD4 counts. We quickly mobilized several REDS labs. We found that the technology was not available to do it practically with reliable results. It turned out that the data from people with JCL, so-called people with "AIDS without virus," were merely outliers in the normal variation of T-cells counted in blood donors, although it took us awhile to discover that. REDS was at the center of that, with incredible cooperation from the CDC. REDS also developed consensus conference data that were presented both at an NIH Consensus Development Conference and at the Blood Products Advisory Committee. These involved incidence data and window period estimates for viral diseases and how much we would close the window by antigen testing and whether there would be a magnet effect. REDS will continue to make general contributions to the risk literature. We are still probably the largest database of donors that can be accessed quickly with a great deal of accuracy. The Institute of Medicine report on HI V and the Blood Supply recommended continuing monitoring for risks in the blood supply. REDS is a very comprehensive database, and, importantly, can yield incidence data, and the use of incidence data is really the most reliable way to make decisions on what is happening. We can continue to track elements specified in the REP which are the data elements that I have outlined. Most importantly, as with ICE and the lessons with JCL, we were able to respond within days because we had the five centers in place. We had Westat crunching the numbers. We were able to respond in a way that no one else can because of the magnitude of the database that we have to deal with. One of the things that we might want to consider is not maintaining the current level of funding, but using a reduced level of funding to keep the REDS mechanism in place for the future so that we can respond and answer queries when we need to. It took us 18 months to set up this system. It was extremely difficult and it is extremely complicated, but it is in place now. It is on autopilot. It would be a pity if we lost the opportunity to have continuous surveillance and answer queries about unknown agents, new testing technologies, and the like.
Demographic and Serologic Characteristics of Volunteer Blood Donors38 lames A. Koreli~z I would like to provide a brief overview of the demographic profile of blood donors in the Retrovirus Epidemiology Donor Study (REDS) as well as the prevalence of infectious markers that has been observed. I would then like to share with you the attempts to estimate the incidence of infectious disease markers, how incidence differs Tom prevalence, and how the incidence rate can be used in conjunction with estimates of periods of true positivity but seronegativity (window periods) to assess the risk of an infectious unit entering the blood supply. In addition to standard questions, such as those regarding gender and age, REDS collects information on additional donor demographics such as race, ethnicity, education, country of birth, and transfusion history. The standard battery of serologic tests is perfonned on all donations including tests for retroviruses and hepatitis viruses. A key feature of REDS worth emphasizing is that a unique donor identifier is created so that donations Tom the same donor can be linked for farther analyses such as the incidence analysis, as well as with other components of REDS. What are the demographic characteristics of blood donors? Based on approximately 2,000,000 donations (excluding autologous) collected during 1991 and 1992, we found that · slightly more donations are from males than Tom females, · more than 70 percent of donations are Dom the 20~9 age group, with about 10 percent from those under 20 and about 20 percent fiom those 50 and older, over 80 percent of donations come Dom white, non-hispanic donors, . and 38Talk presented at Forum on Blood Safety and Blood Availability, July 12-13, 1994. An updated analysis is given in Schreiber, GB, MP Busch, SH Kleinman, JJ Korelitz (1996). The risk of transfusion-transmitted viral infections. New England Journal of Medicine, 334: 1685-1690. 31
32 BLOOD AND BLOOD PRODUCTS: SAFETY AND RISK . donors are generally well-educated, 70 percent having some college experience and almost 90 percent being high school graduates. An important factor that will come up when we talk about incidence is the fact that 79 percent of our donations are from repeat donors and 21 percent are from first-time donors. It has been generally recognized that f~rst-time donors have a higher prevalence of infectious disease markers compared with repeat donors. With that demographic profile in mind, what kind of prevalence values are we observing? During 1991 and 1992, the prevalence of human immunodeficiency virus (HIV) and human T-lymphotropic virus (HTLV) were each about 1 per 10,000 donations. For hepatitis C virus (HCV), based on 19 months of data corresponding to when the supplemental HCV test was implemented at the blood centers, the prevalence was about 22 per 10,000. Prevalence gives us an idea of how many people are currently infected, or positive, for an infectious disease marker. It has obvious importance, especially from a broad public health perspective, for estimating the current magnitude of a health problem. However, prevalence does not provide us with any indication of when the infection occurred, and the timing of an infection is critical in the blood donor setting. After all, people who were infected long ago will show up as prevalent cases, but their donations will test positive and be excluded from the blood supply. A more important question relates to the rate at which negative donors are becoming positive, or seroconverting. People who have seroconverted will test positive, and their donations will be excluded, but people who are seroconverting may be in the window period where their donations are infectious but not detected by current tests. The definition of an incident case is fairly straightforward: it is when a donor who previously gave a negative donation shows up and gives a positive one. Remember that with REDS we have a linking donor identification number, so these donors can be identified. The incidence rate is then calculated as the number of incident cases divided by the total person-time observed. Let me give a brief example to explain person-time. Suppose person A gives two seronegative donations 18 months apart, and person B gives two seronegative donations 6 month apart. Neither one is an incident case, so we could say the incidence rate is O out of 2 donors. This method treats each donor equally. However, we would like to incorporate the fact that donor A was observed to remain seronegative for a longer time period than donor B. Likewise, suppose we observe two incident cases, or seroconverters, who initially give seronegative donations but subsequently give seropositive donations. The time between donor C's seronegative donation and seropositve
GUARDING THE BLOOD SUPPLY 33 donation may have been 20 months, whereas the time between donations for donor D is 7 months. A way to express the incidence rate when subjects are observed for varying amounts of time is to say there were 2 cases per 51 observed person-months (18 + 6 + 20 + 7), or 0.039 cases per person-month. Thus, for each REDS donor, we determined whether or not they were an incident case, and the person-time between their first and last donations. Table 3 details each marker, the number of donors, the number of observed person-years, the number of incident cases, and the resulting incidence rate, expressed as number of cases per 100,000 person-years. The HCV rate is based on data obtained after the second generation screening test was implemented, so the sample size at this point is much smaller than for other markers. The hepatitis B virus (HBV) incidence rate is based solely on the HBV surface antigen (HBsAg) test. It does not include donors who went from negative to positive on the antibody to hepatitis B core antigen test. TABLE 3 Preliminary Results of Incidence Analysis39 Number of Number of Person- Incident Incidence Marker Donors Years Cases Ratea Anti-HIV 426,149 421,777 11 2.61 Anti-HTLV-I 426,134 421,767 3 0.71 Anti-HCV 151,708 62,444 4 6.41 HBsAg 426,101 421,734 28 6.64 Per 100,000 person-years. Although the rate for HTLV is based on only 3 incident cases, it is interesting that although the prevalence of HTLV was a little higher than the prevalence of HIV, the incidence of HIV is considerably higher than the incidence of HTLV. The next question is exactly how do you use, or what is the relevance of, an incidence rate? Using HIV as an example, exactly what does 2.61 cases per 100,000 person-years mean? The incidence rate, when it is a small number such as 2 or 3 per 100,000 person-years, is essentially equivalent to a probability, or risk. So we can say that if the rate is 2.61 per 100,000 person-years, the risk of a person seroconverting within a 1-year period is 1 in 38,000. Normally, we think of 39See Schreiber et al. (op cat) for an updated analysis.
34 BLOOD AND BLOOD PRODUCTS: SAFER AND RISK using incidence rates or probabilities to predict the future. That is, we might say that 1 in 38,000 donors will become infected within the next year. But you could also use this figure to say that 1 in 38,000 donors were infected within the past year. The reason I make this distinction is that in the blood donor setting, the key question really is not "What is the probability that a donor will become infected after donating blood?" The critical question is, "What is the probability that a donor was infected before donating blood?" As I said before, if the donor was infected long enough ago, the serologic test will be positive and the donation will not enter the blood supply. For example, if everyone who was infected more than 6 months ago tests positive, then the question is, "What is the probability that a donor who shows up today to donate blood was infected within the past 6 months?" This is because only this infected donor will test negative and this donation might be used for transfusion. If we calculated a risk of being infected within the past year to be 1 in 38,000, then the risk of being infected within half of that time, or 6 months, should be half of the risk, or 1 in 76,000. Likewise, we can estimate the risk of being infected within any time frame to be proportional to the risk calculated on a "per year" basis. If we believe that the window period is 45 days, that is, only people who were infected within the past 45 days will have a negative serologic test today, then the risk of one of today's donors being such a person is 1 in 311,000. On the other hand, remember that we calculated our incidence rate from donors who gave at least 2 donations during our study period. What about first-time donors? It is generally assumed that first-time donors will have higher rates than repeat donors. To see what impact this can have, we can just assume a certain rate for first-time donors relative to that for repeat donors and weight that rate by the percentage of first-time donors in our study. For example, let us use the observed incidence rate of 2.61 per 100,000 person-years and say the window period is 45 days. If the rate in first-time donors is the same as that in repeat donors, then the risk is 1 in 31 1,000. But suppose the rate in f~rst-time donors is 50 percent greater than the incidence rate in repeat donors. This means that the rate in first-time donors is about 3.9 per 100,000 person-years. We observed, and other blood donor studies have also reported, that about 21 percent of donations come from first-time donors. So, if we count the 3.9 rate for 21 percent and the 2.6 rate for 79 percent, then the weighted average, when combined with a 45 day window period, adjusts the risk up to 1 in 281,000. Of course, the 50 percent increase was arbitrary. It could be 80 percent, 100 percent or some other value. One previous study estimated that first-time donors would have 1.8 times the incidence rate of
GUARDING THE BLOOD SUPPLY 35 repeat donors.40 Table 4 delineates the impact of varying ratios of incidence rates for first-time versus repeat donors. The point here is that while the risk goes up, it is not overly sensitive to the unknown rate among first-time donors. TABLE 4 Adjusted Risk of HIV Transmission for First-Time Donors a If the ratio of incidence rates for first-time vs. repeat donors Is: Then the risk of window-period donation is: 1.0 1.5 1.8 2.0 1:3 1 1,000 1:28 1,000 1 :266,000 1 :257,000 a Assumptions are that donors have (1) an incidence rate of 2.61 per 100,000 person- years and (2) a window-period of 45 days, and that (3) 79 percent of donations from repeat donors. It is interesting to view this risk estimate in light of previous studies. Table 5 presents a partial list of risk estimates in the literature. This is a mixed bag of studies that used different study designs and methodologies to estimate risk. It does show a range of estimates, and depending on what range of variation you are used to working with, you might conclude that they are all in the same ballpark, or you might feel that the estimates are widely disparate. One important factor in looking at risk estimates for HIV is the time frame. As you know, we have had a very dynamic situation with HIV in teens of donor screening and testing. The studies are listed according to the year that the study ended. The risk estimate from each study is multiplied by 18,000,000, which is the number of donations or units that are collected each year in the United States, to get the number of window-period donations that would be expected to enter the blood supply each year. There appears to be a trend of decreasing risk with time; that is, more recent risk estimates appear to be lower than older risk estimates. This downward trend over time is what you would expect if you were decreasing the incidence rate (by effective donor screening) and/or decreasing the window period (by improved HIV tests). 40Cumming, PD, EL Wallace, JB Schorr, RY Dodd (1989). Exposure of patients to human immunodeficiency virus Trough the transfusion of blood components that test antibody-negative. New England Journal of Medicine, 321(14): 941-946. Comment in New England Journal of Medicine, 322(12): 850-851.
36 BLOOD AND BLOOD PRODUCTS: SAFETY AND DISK TABLE 5 Risk of a Window-Period Donation Due to a Seroconverting HIV- Positive Donor: Estimates Tom Previous Studies Risk No. of Window-Period Source Study Dates Estimate Donations Expected Ward et al.4i 198~1987 1 :38,000 474 Kleinman and Second42 198~1987 1:6S,000 265 Cumming et al.43 1985-1987 1:153,000 118 Cohen et al.44 1985-1989 1:36,000 500 Busch et al.45 1987-1989 1:61,000 295 Kleinman46 1988-1989 1: 106,000 170 Nelson et al.47 1985-1991 1:60,000 300 Petersen et al.48 1988-1991 1:220,000 82 REDS49 1991-1992 1:257,000 70 Vyas et al.50 1987-1993 1:160,000 112 Finally' Table 6 points out how the incidence rate can be used to help assess the impact of shortening a window period' again using HIV as an example. It might be that there is uncertainty or controversy over the total length of the window period. Is it 45 days, or 60 days, or 30 days? However, . award, JW, SD Holmberg, JR Allen, DL Cohn, et al. (1988). Transmission of human immunodeficiency virus (HIV) by blood transfusions screened as negative for HIV antibody. New England Journal of Medicine, 318(8): 473-478. 42Kleinman, S and K Second (1988). Risk of human immunodeficiency Virus (HIV) transmission by anti-HIV negative blood: Estimates using the lookback methodology. Transfusion, 28: 499-501. 43Cumming et al. (1989), op cit. 44Cohen, ND, A Munoz, BA Reitz, PK Ness et al. (1989). transmission of retroviruses by transfusion of screened blood in patients undergoing cardiac surgery. New England Journal of Medicine, 320(18): 11 72-11 76. 45Busch, me, BE Eble, H Khayam-Bashi, D Heilbron et al. (1991). Evaluation of screened blood donations for human immunodeficiency virus type 1 infection by culture and DNA amplification of pooled cells. New England Journal of Medicine, 325(1): 1-5. 46Unpublished. 47Nelson, KE, JG Donahue, A Munoz, ND Cohen et al. (1992). transmission of retroviruses from seronegative donors by transfusion during cardiac surgery. A multieenter study of HIV-1 and HTLV-IIII infections. Annals of Internal Medicine, 117(7): 55~559. 48Petersen, L, ~ Buseh, G Satten, R Dodd et al. (1993). Narrowing the window period with a third generation anti-HIV-1-2 enzyme immunoassay: relevance to P24 antigen screening of blood donors in the United States [abstract]. International Conference on AIDS, 9(2): 717. 49See Sehreiber et al. (op cit) for an updated analysis. 50Vyas, GN, ED Rawal, G Babu, MP Buseh (1994). Diminishing risk of HIV infection from transfusion of seronegative blood [abstract]. Transmission, 34(5uppl.): 63S.
GUARDING THE BLOOD SUPPLY 37 regardless of the total window-period length, it might be possible to estimate how many days sooner a new test will detect HIV antibodies. For example, we might not know whether the window period is 30 or 60 days, but we might determine that a new test shortens the window period, whatever it is, by say 5 days. If we use the REDS incidence rate and make the other usual assumptions, we can estimate that shortening the window period by 5 days will detect between 8 and 13 HIV-infected donations that otherwise would have tested negative. The same approach could be applied to any other shortening of the window period. This information helps estimate the yield or benefit of a new test, which then might be incorporated into a cost-benefit analysis of a proposed new test. TABLE 6 Impact of Shortening the HIV Window Perioda If the window period is shortened by: Then the number of additional HIV-infected donations detected will be: 1 day 2 days 3 days 4 days 5 days 10 days 2-3 3-5 5-8 6-10 8-13 16-26 a Assumptions are (1) a donor incidence rate of 2.61 per 100,000 person-years, (2) 79 percent of donations are from repeat donors, (3) f~rst-time donors have 2.0 times the incidence rate, and (4) there are 18 million donations5i In conclusion, while the prevalence of seropositive donations is relevant, especially in a broad public health context, it is really at best an indirect measure of risk to the blood supply. The incidence rate, which we were able to calculate because we have information that links together a donor's multiple donations over time, is a better measure of risk (specifically transfusion- associated risk) because it estimates the risk of recent infection. We saw how the risk of donation during an infectious window period depends directly, proportionately, and equally on the incidence rate and the length of the window period and how the incidence rate can be used to assess the impact, or yield, or benefit of shortening the window period. I have used 5'See Schreiber et al. (op cit) for an updated analysis.
38 BLOOD AND BLOOD PRODUCTS: SAFER AND RISK HIV as an example, but the same points and same methods can be applied to other infectious diseases as well. We are currently in the process of estimating demographic-specific incidence rates. I think it will be interesting to see if the same demographic patterns that have been observed in the prevalence of infectious disease markers hold for the incidence rates. Finally, REDS will continue to monitor and report on changes with time, and will incorporate results from the other REDS components, such as the special donor surveys, as part of its mission to monitor the safety of the U.S. blood supply.
CDC Surveillance of Donors Eve M. Lackrifz Surveillance is the ongoing and systematic collection, analysis, and interpretation of health data. The Centers for Disease Control and Prevention (CDC) monitors the transmission of transfi~sion-transmitted diseases through these ongoing data collection systems. Different branches and divisions at CDC, including the Division of HIV/AIDS Prevention, have their own surveillance activities Surveillance provides information for action: decision making, policy development, and development of prevention strategies. Within the Division of HIV/AIDS Prevention, several branches have surveillance activities to monitor the transmission of HIV by blood transfusion. The Surveillance Branch manages the HIV/AIDS Reporting System, which receives reports of all patients with AIDS, including those who were infected from screened blood or blood products. In addition to managing the surveillance system, the Surveillance Branch conducts special studies to follow-up on AIDS cases that were reported to have been transmitted by transfusion of screened blood. The HIV Seroepidemiology Branch is responsible for two surveillance projects that I would like to discuss. One study collects information on all blood donations from 19 of the 45 American Red Cross blood services regions. The second study, the CDC HIV Blood Donor Study, involves interviewing HIV-positive blood donors from 20 different U.S. blood centers across the country and maintaining a repository of serum and cell samples from these donors. The blood centers in the interview study are not the same as those in the Red Cross study. The collaborative study with the American Red Cross collects information on approximately 2 million donations each year from 19 Red Cross regions. These regions were selected because they have compatible computerized data management systems, thus allowing data from different regions to be merged and analyzed. Information collected on each donation includes donor demographic information, whether the donor was a first-time or repeat donor, the dates of the most recent and previous donations, and whether the donor elected to confidentially exclude his or her donation from being transfused (the 39
40 BLOOD AND BLOOD PRODUCTS: SAFETY AND RISK confidential unit exclusion system, or CUE). All laboratory test results of these donations are also included in the database. The regional blood centers voluntarily forward their data to the American Red Cross, which then cleans the dataset and forwards it to CDC. This dataset gives us a great deal of information that we use in policy- making and decision-making. For example, the requirement to test all blood donations for syphilis has been maintained because the syphilis test is a surrogate marker for HIV risk behaviors. A surrogate test removes donations by HIV-infected volunteers donating during the window period of seronegativity that would otherwise not be detected by routine HIV testing. Analysis of the Red Cross data allowed us to evaluate the cost and effectiveness of syphilis testing as a surrogate marker for HIV window period donations. We have also used the Red Cross dataset to evaluate the effectiveness of the confidential unit exclusion system and more recently the risk of HIV transmission by screened blood in the United States. The second surveillance project conducted by the HIV Seroepidemiology Branch is the CDC HIV Blood Donor Study. Twenty U.S. blood centers participate in this project. These centers tend to have more HIV-positive donors than other areas. Participating blood centers contact all blood donors who test positive for HIV and request that they return to the center to be informed of their HIV test results and counseled. At the counseling visit, the centers ask the donor to participate in a standardized interview as part of the CDC study. The donor interview takes about an hour to complete. Donors are asked about their HIV risk behaviors and their motivations for donation. CDC also receives laboratory test results of these donations. Partner testing is offered, as is a follow-up interview for those who don't have arty identified risks for HIV. Donors often do not discover that their partners are HIV positive or identify other risk factors until after the donor notification and counseling session. A blood sample is collected at the time of the interview for the serum and cell repository, which has become a very valuable research tool. We have used information obtained from the CDC HIV Blood Donor Study for several different purposes. First, we have evaluated trends in risk behaviors among HIV-positive donors and have identified methods to improve predonation interviews and the donor deferral system. We have investigated such as issues as whether donors were given enough privacy during predonation interviews and why donors with known risk behaviors donated blood. Second, we have studied the effect of donor incentives by evaluating the motivations and risk characteristics of HIV-positive blood donors. Third, the study has served as a means of monitoring for HIV-2 in the blood supply. Fourth, we have used the serum and cell repository to monitor for the
GUARDING THE BLOOD SUPPLY 41 introduction of rare HIV subtypes into the blood supply and to ensure that current HIV tests are detecting rare HIV variants. Currently, we are examining whether HIV-seropositive individuals are donating blood in order to be tested for HIV. The study of test-seeking behavior is particularly important because of the recent licensure of the p24 antigen test. We estimate that, among the approximately 12 million donations made in the United States each year, 7 to 10 donations will be HIV antigen positive and antibody negative. If individuals at risk for HIV infection donate blood to receive this special HIV test, the benefits of antigen testing may be offset by an increase in donations from donors at high risk for HIV infection. Therefore, monitoring this "magnet effect" is critical. These ongoing surveillance projects allow us to monitor trends in transmission of HIV by transfusion and risk behaviors among blood donors. However, CDC also stays poised to deal with new and uncharacterized threats to the blood supply through special investigations and outbreak investigations. These allow CDC to respond rapidly to new or emerging diseases or adverse events. A recent example of one of these special investigations is the investigation of HIV Group O. an HIV variant potentially not detected by routine HIV antibody screening. CDC collaborated with the Retrovirus Epidemiology Donor Study (REDS) and other study groups to retest blood donations that had positive and indeterminate HIV Western blot results and negative EIA results with high optical density readings. Other special studies include collaborative look-back investigations to determine the length of the HIV infectious window period and the risk of HIV transmission by screened blood. Currently, we are developing an evaluation of the benefit, cost, and magnet effect of HIV p24 antigen testing. CDC is not a regulatory agency; we work through collaboration and consensus building to collect and analyze information to promote public health in the United States. DISCUSSION James Allen: Let me initiate the discussion by asking about costs versus benefits. We are in a cost-crunching time right now. Can we justify the continued relatively high cost of these programs in terms of what we are getting from them? Thomas Zuck: The current funding level of REDS should not be sustained beyond the defined period of 1998. We are in the process now of calculating how much it will cost to keep the system in place without the continuing acquisition of new data and without additional physical examinations of the HTLV-infected cohort. The cost has been about $4 million a year over the 9
42 BLOOD AND BLOOD PRODUCTS: SAFER AND RISK years of the study. For about 1/10 of that we can keep the surveillance part intact. If you consider the costs of AIDS and hepatitis C infection, then that is a fairly low price to pay in relation to better than $1 billion of health care. James Allen: Five years from now we will be well beyond the current data. We will have the laboratory repository but one would then have to question the utility of data that are 5 or more years old. Thomas Zuck: Another aspect of the database that may be equally as important is its ability to validate new technologies. There will be a day, probably within the next decade, when polymerase chain reaction (PCR) will be automated and the specificity problem will have been sorted out. Maybe it will be ligase chain reactions instead of PCR, but a genetic technology will be available within a decade. Those repositories are enonnously valuable to validate the quality, specificity, and sensitivity of those tests. Eve Lackritz: There is also a difference between actual risk and perceived risk. What often drives blood safety policy and programs is the public perception that the blood supply is not safe. In terms of HIV, the blood supply is very safe, but the public remains concerned. We need to continue to monitor the safety of the blood supply, but we must also identify ways to more effectively communicate our findings to the public. Paul McCurdy: Many people in the audience know that in the late 1970's the National Heart, Lung and Blood Institute sponsored a transfusion transmitted virus study, the so-called TTV study. As part of that study a repository was established that has been used to look at prevention of hepatitis C virus transmission by antibody testing. It is now being used to look at hepatitis G virus which has relatively recently been described and probably will be used to examine some of the antigen testing procedures. The REDS repository and data may turn out to be equally valuable for unknown reasons sometime in the future. Thomas Zuck: One of the major advantages of the TTV study is that the TTV data are linked to recipients, and that makes it uniquely valuable. One of the weaknesses of REDS is that we don't have recipient linkages. Now, we could go back and create them, because the data are all encrypted, but the great strength of TTV is that it has that recipient linkage. Paul McCurdy: The transfusion safety study also had a repository and a lot of data.
GUARDING THE BLOOD SUPPLY 43 Thomas Zuck: Again, however, they were unlinked to recipients. The uniqueness of TTV is the recipient linkage. Lew Barker: These sorts of studies are extraordinarily valuable, but I wonder how to maintain surveillance as cost-effectively as possible because it is going to continue to be important, and how to maintain the ability to look at new technologies relatively quickly and efficiently. Thomas Zuck: That is the great value of the repositories. If a strategy appears (and we don't know what kind of genomic analysis is going to come down the line), then the REDS repositories will be the fastest place to go to find the answer to new technologies or emerging viruses. Eve Lackritz: I think the repositories are going to become increasingly important, and we are putting more of our emphasis on that. There are ways to streamline the budget, but notifying donors and bringing them in for interviews and sample collection are labor-intensive activities. Harold Sox: I was part of the Institute of Medicine Committee that produced HIV and the Blood Supply, and back in 1983 there was information, like a ligand, but there was no receptor over at the Food and Drug Administration (FDA). You mentioned that you are a surveillance agency, not a regulatory agency. Could you tell us whether that receptor is in place now and whether the internal workings of the cell, just to extend the analogy, are such that the signal gets transduced to take some action? Eve Lackritz: We collect information for the purpose of taking action. We communicate regularly with FDA through a Public Health Service (PHS) conference call that FDA coordinates twice a month. When we realize available information is limited, other people are contacted to provide additional infonnation to that forum. Thus, there is a system in place. For example, we routinely do analyses for Blood Products Advisory Committee meetings. Whenever an issue comes up, we are poised to respond with the appropriate dataset, and vice versa. If we receive a report of an emerging disease or adverse event, we contact FDA. The ties are very strong, and it has been a good relationship. William Sherwood: Repositories have limitations. We had a repository of samples from hemophiliacs. It wasn't all in one place, but the libraries of sera went back to the l950s. Yet they couldn't be used. We learned in 1985 that at least 20 percent of hemophiliacs were infected with HIV before we even
44 BLOOD AND BLOOD PRODUCTS: SAFER AND RISK knew that there was a such a virus, and far more were infected before we had a test. You cannot really use these repositories until you have a test. Eve Lackritz: The other important thing is that we have interviews from these donors. AIDS reporting started before an etiologic agent had been identified. You can start doing epidemiology without a known test, and our existing systems are flexible enough that we can modify them and capture those events, using a new case definition if need be. Also, having repositories in place allows for a more rapid response to a new problem, such as when we used peptide assays to study Group O in the U.S. blood supply. Thomas Zuck: You really cannot get prevalence data without a test, and lack of a test was what was confounding things in the early 1980s. No one thought that the prevalence was as high as it was when we started asking the high-risk questions. You really need a test to get a handle on prevalence. That was the key element missing in 1983 and 1984.
Surveillance of Recipients James R. Allen Our session today is titled Guarding the Blood Supply. Since all the presentations focus on the role of data collection through surveillance to detect potential problems rather than data developed through special studies or research, I am going to begin my remarks with a short exegesis on surveillance. Surveillance is quite different Mom special studies or research. Special studies start with a specific hypothesis that is to be tested. The studies are focused, are conducted in a fairly defined population, and most often within a closed time frame. Special studies rely on defined data elements gathered to foam a fairly rigid data set, and the methodology is well established to insure that the data collected are accurate and reliable, and that they can be confirmed. Surveillance on the other hand is much more nebulous. The intent is to identify a problem, not to fully study it. Surveillance, therefore, has characteristics very different from those of special studies or research. Surveillance is a data-gathering system, whether active or passive. Most surveillance systems are passive; that is, they rely on observations and reports of people who may not recognize that they are part of a routine data collection system. The individual reports are collected in a central repository, whether it be at the local or state health department, the Centers for Disease Control and Prevention (CDC), a blood collection center, or any other organization that may be collecting the information. Surveillance systems may use defined methods or they may simply rely on random reports being generated and sent in. The population under surveillance is often open and may change over time for various reasons. The population is often not well-defined, even when there is a specific target population. Bruce Evatt will address surveillance in special populations, people with hemophilia in particular. This is a small well-defined population, but I think Bruce will readily acknowledge that there are many people with hemophilia who may be outside of the defined population system, and that there may be events going on with them that do not get reported back in routinely. 4s
46 BLOOD AND BLOOD PRODUCTS: SAFETY AND RISK The data gathered through a surveillance system most often are highly restricted, since it is not practical to collect a large amount of information, especially through a passive surveillance system. The primary data collected are demographic, the specific event of interest, and a limited set of data about the circumstances that surround the event or important laboratory findings. Also, surveillance systems are often open-ended, since there is no defined time frame within which the study is going to be completed. continuing process that forms a rolling database. Surveillance is a ~7 Although that brief discussion provides the essence of surveillance systems, I want to mention two other characteristics that are extremely important to successful surveillance. First, somebody has to be responsible for analyzing or looking at the data on a regular basis. Too often a surveillance system may have data entered into a database system on a continuing basis but never have that wealth of information analyzed or looked at critically by a human being who is asking questions, plotting trends, and trying to make sense out of the data. The data may get dumped into tables. They may be published or made available in various formats, but nobody really looks at the data, thinks about them, asks questions, and tries to make sense out of them. Second, the final component of surveillance is for that data or information to be disseminated to those who need to know and who can act on the data. If the data are reported to a surveillance system but remain in a repository and are not disseminated to those who can act on it, they do not do any Rood. ~ __ -11 _ ___ . . .1 r ~ . ~ ~ _ _ ~ ~ ~ ~ _ ~ waft now turn to the focus ot today7s meetin~recipients of blood and blood products. We will define recipients for this purpose as being a person who has received an infusion of any blood or blood component as part of his or her therapy. We are not talking about the special populations or people with a defined disease or a defined situation. We are talking about the general population here. The notion that if you don't know where you are going, any road will get you there epitomizes the problem that we have with defining surveillance for disease or for adverse events in a population. There are subcomponents that we can break out. The time course of the event or presumed event is extremely important. For example, doing surveillance for transfusion reaction is relatively easy. There is a short time course from the time of infusion to the time at which the event occurs, the patient is usually still on site, and it is a clinical event that is fairly well recognized. We then have the intermediate time course problems, such as transfusion-associated hepatitis, and there may be either active or passive surveillance systems for monitoring what is happening. We are talking about a time event of 6 to 8 months. There may be a clinical event or it may be subclinical, but we have good laboratory markers for many of the hepatitis types that are transfusion transmitted. The clinicians involved may report back
GUARDING THE BLOOD SUPPLY 47 to the blood banking system, or the blood banking system may have an active system in place, or we may set up special studies if we have particular concerns and want to look at the situation very intensively with a very sophisticated study for a relatively short period of time. What about surveillance for longer-term events, however, or events that are ill defined? This becomes more difficult. The paradigm that we can use is what we now know to be transfusion-transmitted human immunodeficiency virus (HIV) infection occurring in the early 1980s. The first cases of AIDS were recognized in mid-1981, and were reported to CDC. The first thing that had to be done was to put together the clinical parameters and then the immunologic parameters. Over the first 3 to 6 months, CDC, health departments, and clinicians around the country did a superb job of assembling the basic epidemiologic parameters. By mid-1982, just 12 months after the recognition of the epidemic, we understood the potential for AIDS to be transfusion transmitted. At CDC we recognized that we had this unusual clinical condition in people with hemophilia. But how did we know it was the same disease in all of them inasmuch as people with hemophilia have a lot of other events occurring to them? They are not your standard patient population, and the disease was ill defined. We had no lab markers. Yet, the system began to respond immediately. We became aware of the potential for transfusion- transmitted diseases, and when the first reports did reach CDC that fall, the investigations began to try to get all the pieces of information that were going to be used to confirm that event. It was late November 1982, when the first report came into CDC of an infant who became ill, when the information began to fall into place. At that time, even though we had a number of children from New York and New Jersey in particular who had a condition that seemed to be a pediatric AIDS-like condition, no one had yet formally accepted that children got AIDS. There had not been a case report of pediatric AIDS published in the Morbidity and Mortality Weekly Report or anywhere else at that point. We had all these unusual events that were coming together, and we had to try to sort them out and put them together. It was remarkable that the information was obtained as efficiently and rapidly as it was, and that the pieces of the puzzle were able to be put together extremely well, such that by early 1983 we had a fairly clear picture of the outline for the puzzle, even though we didn't have all of the center of the puzzle filled in. In retrospect, could we have done better on that? I don't think we could have done much better in terms of rapidly recognizing an unexpected or unpredictable event with a pathogen that was brand new and with a disease that was ill defined in the absence of any specific laboratory markers.
48 BLOOD AND BLOOD PRODUCTS: SAFETY AND RISK Could we have done better in terms of getting close cooperation and collaboration among all the participants, the CDC, the National Institutes of Health, the Food and Drug Administration, the blood collection centers, hospitals, and various other aspects of the health care system? Yes, in retrospect, that might have worked more smoothly. Nonetheless, given the diversity of our system, our concern about the rights of individual people and confidentiality of patient records, and the absence of any preexisting collaborative system, it still worked relatively smoothly, and we were able to get a lot of answers very quickly. What then might we learn in terms of lessons for today? Can we put a more formalized surveillance system into place for transfusion recipients, one that will get information back to us rapidly, help us to identify either known or unknown problems in the future? If you really want to do that, you have to have an active tracking system for following up transfusion recipients, to gather clinical information and blood samples for additional testing and placement in a repository. However, it is not feasible, nor is it necessary. It would be extremely expensive, and it would not be cost-effective. For the known problems, it is important that we continue to publish information about current events and trends and to be certain that clinicians as well as health officials, blood banking officials, and federal government officials are all aware of that information. We need to be very aggressive at instituting special studies as appropriate. We need a strong general database at all levels, particularly computer-based patient records. One of the issues related to transfusion-associated investigations in the early 1980s was trying to get back to the patient record to determine if that patient did in fact receive that unit of blood. The unit would have been signed out by the blood bank, but there would be no record in the medical record that the unit was ever returned back to the blood bank. We often couldn't confirm that the patient received that unit of blood. We need much stronger linkages between the blood collection center, the hospital transfusion service, and the patient record. We need to refine the linkage as one gets beyond the transfusion service to the patient record, the clinician, or the current physician of record if we are really going to do a good job of general surveillance. Over the next 5 to 10 years, as we move to managed care, some of these issues will begin to be addressed. We will need to have long-term maintenance of the record. How long is long enough? If you go back a decade, medical records will be boxed up and not available if the patient hasn't come into the system in the preceding 5 or 6 years. Records should be maintained for at least 10 years, and maybe 15 years, for us to be able to go back and access patient and transfusion records.
GUARDING THE BLOOD SUPPLY 49 Surveillance for unknown problems begins with having a high curiosity level and a high index of suspicion and being willing to get back to an appropriate reporting or communication source very early on. It obviously means that it must be incumbent on the public health system that is receiving that report to make an appropriate analysis and pass it up through the system. I am very concerned about what is happening to our state and local public health agencies, as well as at the federal level, given the budget and personnel cuts that are coming. Finally, it is possible to get useful information out of large surveillance systems that are in place for other reasons. For example, CDC has a hospital-based record system, the National Nosocomial Infection Study, intended to look at infections occurring in hospitals. This is a multiple- hospital reporting system that has been used in the past to detect and evaluate problems with contamination of large-volume parenteral fluids. In about 1970, a nationwide outbreak occulted via the fluids distributed by one manufacturer, and even with the small neophyte system at CDC then, we got a glimmer of the problem. The CDC system picked up the unusual pattern of infections reported, and made queries to other hospitals that then confirmed the pattern. When CDC later did much more intensive analyses using its surveillance system, the data were there. The problem could have been detected had CDC been looking at the data promptly and in exactly the right manner. It is intriguing what can be done with these systems if you have the resources to look at the information that is there and that is being reported. Might such a system be helpful in transfusion-transmitted diseases? It all depends. If you are getting sick from infusion of large-volume parenteral fluids, that is occurring very proximate to the hospitalization. However, with transfusion-transmitted infections we are looking at events that occur months to years afterward. There isn't any way of easily getting those records back into the hospital-based system. Nonetheless, I think models that might be appropriate in the future are available, but they will be costly. In summary, we are much better off going with a general awareness and a much less structured system, similar to what we had in the early 1980s. That system worked very well. DISCUSSION Harold Sox: You mentioned that an ongoing surveillance system was too expensive. Has there been a thorough cost-effectiveness analysis of a wide variety of different approaches to implementing a screening process? James Allen: When I said that it would be too expensive, I was talking about an active system that is specifically trying to get back information. That
50 BLOOD AND BLOOD PRODUCTS: SAFETY AND RISK would be extremely difficult and very costly, with very little useful information coming back. I am not aware of any cost analyses of that. Scott Wetterhall: If you have a database or a serum repository but you don't have a disease, there is nothing to survey at that point. There may be benefits in terms of maintaining serum repositories in terms of obtaining additional information once a disease has been recognized. Clearly, what is relevant for this group, given that the safety of the blood supply is the issue, is that you need a diagnostic test. You are maintaining a system, and yet you don't have anything that you are measuring. It is really only in a retrospective sense that it may provide you with information. Harold Sox: Suppose you are worried about a disease occurring that would have the same catastrophic effects that AIDS did. You could do a model which would basically assume the disease you were looking for. The only reason for doing the surveillance would be to detect something as bad as AIDS and then trying to get a handle on what different levels of surveillance might cost and what the benefits would be. Without that type of analysis we run the risk that we will repeat the AIDS history. William Sherwood: We know from the repositories that were tested in 1985 that the AIDS virus entered the hemophiliac population in about 1978. If we had all the money in the world, what surveillance program could we have put in place in 1978 to prevent what happened? Bruce Evatt: The key is that the surveillance program should be coupled to other programs. But surveillance programs of high-risk groups are extremely expensive. If surveillance is part of an ongoing program that deals with other issues and is integrated into those programs, then it is very feasible to do. But with a surveillance system, it is somebody's responsibility to examine for unknown events. Repositories do no good unless you have a test, but here we are not talking about repositories. We are talking about clinical events that occur out of the usual occurrence. Therefore, ongoing programs that deal with that issue would have picked this up. I can tell you with HIV in the early 1980s the surveillance program that helped us was the pentamidine investigational new drug (IND) application. That is where we found all the early cases of HIV infection among both hemophiliacs and transfusion recipients. Because the requests all came to us, we used that file to identify new patients for whom there were no known no risk factors. That is how we identified the those cases. If we hadn't had the 1ND for pentamidine and if pentamidine wasn't used to treat Pneumocystis carinii pneumonia, we wouldn't have identified
GUARDING THE BLOOD SUPPLY C 1 AIDS in hemophiliacs, because most of the times when we investigated AIDS among hemophilia patients or transfusion recipients, their physicians had no idea that these people had AIDS. We ended up informing the hemophilia treatment center doctors that these patients probably had AIDS, and that it was a new disease. What was useful for us was the presence of a system, a data collection system that existed for another purpose. You must identify such information collection systems that might be useful and that can be justified on an ongoing basis for some other purpose, plug into those systems, and make it somebody's responsibility to look for unusual clinical events. This is because the first occurrences of new diseases will not be discovered by a test that you can use to go to a repository. It is going to be unusual clinical events that point the direction to a new syndrome that is out of the ordinary, and that means you have to know the background levels of such events. You must know the incidence of those unusual events, and you must look for changes in trends. That is the form that surveillance of these kinds of problems is going to have to take. Setting up a surveillance program for the purpose of surveillance at $4 million a year is a big waste of money. It must be incorporated into the standard ongoing programs for which data are being collected for other reasons. You must have somebody looking at one or two questions that identify the unusual risk. James Allen: What you just described there is a key point that we have made over and over, and that is that somebody must be looking at the data, interpreting them, and getting them out where they will be useful. Data systems by themselves aren't very useful if nobody is looking at them.
CDC Surveillance of High-Risk Recipients Bruce Watt Serum repositories have been collected in the past, but few have been very helpful in retrospect. Ongoing information collection is the most useful source of information in the identification of emerging situations involving blood-borne infections. Congress has given the Centers for Disease Control and Prevention (CDC) the responsibility to develop a long-term program for the prevention of the complications of hemophilia. This program consists of several phases. First, we are to define the scope of the complications within the hemophilia population. The second task is to assess prevention opportunities for the major complications. Finally, CDC is to design prevention programs, develop resources to address those complications, and evaluate whether these programs are clinically effective and cost-effective. Hemophilia is an extremely expensive disease. It costs about $1 billion a year to care for approximately 17,000 patients. Thus, if a 10 percent improvement in the effectiveness and efficiency of the programs can be made, an improvement in the health care delivery system for the individual hemophilia patient will occur at a considerable savings of health care resources. In the early 1980s we thought that most of this population was cared for primarily by hemophilia treatment centers. We discovered that this wasn't the case when we began to examine mortality data collected from surveys of the hemophilia treatment centers. We found that when these data were compared with the death certificate data, only about two-th~rds of the hemophilia patients were within hemophilia treatment centers. Another one-third were outside those centers. Nothing was known of the source of their care or the nature of their disease. Thus, the first part of the system was to define the hemophilia population, where they received their care, and the nature of their complications before we could prioritize the complications. The present hemophilia surveillance system was designed to identify this population. This surveillance system is unique in that there are no existing surveillance models for chronic diseases of low incidence. There are 53
54 BLOOD AND BLOOD PRODUCTS: SAFETY AND RISK surveillance models for acute infectious diseases and chronic diseases of high incidence such as diabetes. This is not an inexpensive system, nor is it a system that has universal application. Likewise, it is not a system that needs to last indefinitely. This system was directed at six states, which contain about 25 percent of the supposed hemophilia population. They are Massachusetts, New York, Georgia, Louisiana, Colorado, and Oklahoma. This system was designed to characterize the distribution of the types of hemophilia, the severity, the resource utilization, the complications, and the level of joint disease as well as other aspects of hemophilia. Confidentiality is rigidly enforced. It was much more than the standard annual infectious disease surveillance system. It consists of an annual retrospective chart review of all the hemophilia patients within the six states. Most of the patients would be identified in hemophilia treatment centers. A very important part of the study, however, was to define individuals who did not receive their care from hemophilia treatment centers. It is easy to obtain information from organized hemophilia treatment centers, but very difficult to track down and identify a patient outside hemophilia treatment centers. Nonetheless, the goal of this system is to identify 100 percent of the patients in these states. All six state health departments have requested that hemophilia be a reportable disease in order to facilitate case finding. In addition, patients are also identified with the aid of the local hemophilia chapters, people who supply clotting factor information; pharmacies, hematologists, clinical record systems, hospitals and physicians' offices also supply patient information. Laboratories report any patients with Factor VIII or IX below the level of 30 percent. All are investigated, so that all patients will be identified. The data collection consists of demographics, source of care, hospital issues such as the hemophilia type and severity, the number of bleeds, the factor used, inhibitors, and the number on prophylaxis. More importantly, surveillance information on viral infections, joint disease, HIV status, and any new infections will be gathered to track emerging trends. We will also document all hospitalizations, including the dates and diagnoses, as well as other kinds of infections, such as sepsis, and other complications of transfusions as part of this database. Currently we have 2 years of data in the database. These data are used by the local community as well as being transported electronically to a centralized database at CDC. We are particularly interested in identifying viral infections with this system. We should have the ability to identify both acute and chronic hepatitis B virus (HBV) infections, acute and chronic hepatitis C virus (HCV) infections, acute hepatitis A virus (HAY) infections, and HIV seroconversions. Because this is designed to be a retrospective review of information on these
GUARDING THE BLOOD SUPPLY 55 patients, it will not detect infections quickly in any individual patient. We will be able to detect unusual syndromes in individual hemophilia patients, as well as trends in infections. If a number of unusual diagnoses occur clinically we will be able to detect these. Approximately 115 hemophilia treatment centers in the United States currently care for about 14,000 individuals with hemophilia A and B and another 3,000 or 4,000 individuals with von Willebrand's disease. We currently finance these hemophilia treatment centers for prevention activities. The Health Resources and Services Administration provides these centers with another $5 million a year for health care services. We want to broaden the scope of prevention activities in these centers to include studies of prevention interventions for the complications of hemophilia, including blood-borne infections. As part of that we will require a data collection system to help us determine whether our prevention programs make a difference in health outcome. The two highest priorities for us are reducing the level of joint disease in the community and monitoring blood safety. The data collection system needs to be simple and ongoing. We are designing a universal data collection system that will become part of the clinical activities of all of those 115 hemophilia treatment centers. Such a system would monitor two-thirds of the U.S. hemophilia population. This data collection system would be prospective, would assess the level of hemophilia care that is being provided, and would identify issues for specific studies. Time constraints limit the amount of data collection, and we recognize that. We need to address the amount of data that are available and the type of collection instrument that we will use. We must consider routine practice within the hemophilia treatment center and create a data collection system that is useful to those centers as a whole. The initial targets will be hepatitis viruses and HIV. Those are of major concern to the hemophilia community and must be addressed. The data to be collected will include the basic clinical profile, annual viral testing for the known viruses, and other issues of interest. In the first year we will only be able to measure prevalence rates. In subsequent years we will be able to follow incidence rates in this population. The occurrence of acute cases of HCV infection, lIBV infection, and HIV infection will trigger specific investigations to ascertain if these infections were acquired through blood and blood products. We will do this in collaboration with the Food and Drug Administration and the National Institutes of Health. This surveillance program is to be implemented within the coming year. Realistically it will take a year to get the program in place. If FDA and the National Heart, Lung and Blood Institute (NHLBI) feel that more frequent reporting is necessary, we can incorporate that into the system design. In summary this system will identify new infections to be investigated for the
56 BLOOD AND BLOOD PRODUCTS: SAFETY AND RISK source of infection. Also, the system will be used to monitor the regional trends and variations in blood-borne infections. Finally, I would like to talk very briefly about the potential risk of Creutzfeldt-Jakob disease (CJD) being transmitted in blood products. Currently there is no evidence that it is transmitted in blood products. However, the risk cannot be assessed with any degree of accuracy. It is felt by everyone to be extremely low, if at all, and remains a theoretical risk at best. The troublesome thing is that it can be transferred to experimental animals by a transmittable agent, and the agent is present within the blood of the patients with CJD. That makes it a theoretical risk. The epidemiologic question is how do you do surveillance on a disease that has an incubation period of 20 to 30 years? It is not impossible, but it doesn't lead to any kind of ideal solution where you ascertain the risk very quickly. The hemophilia community is quite concerned about CJD, and they are anxious to have answers as soon as possible. The only way we can reasonably approach this problem is with a special study. Currently, 400 deaths occur annually in the hemophilia community in the United States. About 300 of these are caused by AIDS; the other 100 are caused by non-AIDS-related events. Of the 300 who die from AIDS, about 30 to 50 have central nervous system (CNS) AIDS. The hemophilia population has been routinely receiving concentrates for 20 and 25 years now. We would thus expect that if CJD was caused by a transmittable agent in blood products, you might be seeing cases now in individuals with hemophilia. The only condition that could be misdiagnosed would be individuals who have been diagnosed as having CNS AIDS. I have received several letters from individual physicians who treated hemophilia patients who had died with CNS AIDS before the publicity on CJD. There were no autopsies and no post- mortem examinations of these patients. Now these physicians are wondering if these might have been possible cases of CJD. It is this kind of rumor that spreads quite quickly in the hemophilia population because it is a very small, closed community. Both the physicians and the hemophilia patients now agree that the only solution to this is to try to obtain as many postmortem examinations as possible in individuals who die with CNS AIDS over the next several years. We have established a collaborative project between ourselves and Dr. De Arman at the University of California, San Francisco, the various hemophilia treatment centers, FDA, and NIH, as well as the various peer groups from the hemophilia population. We hope to obtain 10 to 15 brains per year for this study. The basic design of this study is to obtain permission at the time of death to remove the brain for a CNS autopsy. One-centimeter cubed biopsy specimens will be removed from the frontal, midparietal, and cerebellar regions and frozen. The remainder of brain will be placed in formalin for 2 weeks.
G UP RDING THE BL OOD SUPPL Y 57 These specimens will then will receive a routine pathologic workup for dementia and an examination for the prion protein. Suspect cases will be reviewed by a panel of neuropathologists selected by ourselves, NIH, and FDA. A single case of CJD under the age of 40 would be very significant. A CJD case in an individual over the age of 40 could mean that we must continue this study, since a case of CJD in an older person won't carry the same weight as one in someone who is young. In summary, the collection systems, with the exception of CJD, which is a special study, are part of a national program designed to assist in lowering the complications of hemophilia. These data collection tools, however, can also be utilized very effectively to monitor blood-borne infections in this community and for observing whether there is any new incidence of unusual diseases. It may be that the only way that we will be able to obtain the information needed to monitor blood safety for relatively rare events or unknown events in a cost-effective manner. DISCUSSION Paul Russell: What evidence do you have that it takes 20 or 30 years for Creutzieldt-Jakob disease to be manifested? Bruce Evatt: It comes from several types of data. Some of it is Tom studies of cannibals in terms of kuru and other types of slow virus diseases. Certainly, the delay in onset was much shorter than 20 to 30 years in the cases of growth hormone-induced transmission. Thus, the delay is probably related to the dose of the agent, although it is probably related to other issues as well. David Rothman: When you made hemophilia a reportable disease, did you already know all about stigma? Do you have any built-in patient confidentially in any of these surveillance efforts? Bruce Evatt: Yes, we worried about a reaction from the hemophilia community. You have to understand that one of the issues that you deal with, confidentiality, is of utmost importance in this community, especially as it concerns HIV and AIDS. One of the problems we had during the 1980s was the fact that the hemophilia population didn't want to be identified. There was no registry of hemophilia patients. Even if you wanted to, you could not have notified them of risks associated with a new disease because they didn't want their names obtained by anybody. The lack of a registry in the hemophilia community has been a major issue. Obtaining a complete accounting of all individuals with hemophilia in the six states has been accomplished very successfully. We managed this in two ways.
58 BLOOD AND BLOOD PRODUCTS: SAFER AND RISK First, the state health departments are the only entities that have the names. They have the right to do that in all of these states. They all have normal procedures to keep names confidential. In many states people in state health departments can go to jail for breaking confidentiality. Thus, they have a track record of maintaining confidentiality over a long period of time; that reassures people. Furthermore, there was an encryption of the identifiers. We do not know who, where, or what they are. It is to maintain a unique identifier for the database so that they can be followed subsequently by the state health department. The second way this was managed was that personnel Tom the state health departments met with the hemophilia chapters to explain the study. They told the chapters how the study is to the benefit of their community because it helps maintain resources and safety for them. Without this database in the system, they are at increased risk for unknown events. The personnel from the health departments regularly attended hemophilia board meetings, so that if there were complaints, somebody was there Mom the state level who could answer the questions and solved problems very quickly. One or two states ran into a few problems, but that was because of some unwise choices and not because they did not inform the community early and often. Those problems were also rectified, but it was not as smooth as in the states where they worked with the community in a more direct way. The hemophilia registries are unusual in that hemophilia is a chronic condition, whereas most diseases reported to state health departments are infectious diseases. However, in the states where this surveillance was implemented, the health departments are addressing the issues of chronic diseases. They are more sympathetic to these approaches than health departments in some of the states where acute diseases are the only things that are reportable. David Rothman: It is still a moderately scary precedent to have state legislatures pass these laws. James Allen: It is obviously a very complex problem. It is not only a chronic condition, it is also a genetic condition, and that has many other implications.
CDC Surveillance for Unknown Pathogens Scott Wetterh all Surveillance has to be linked to public health action. The purpose of surveillance is to assess status, define priorities, evaluate programs, and stimulate research. The thesis that I am going to make, quite frankly, is that surveillance systems per se probably will not detect unknown pathogens or unknown etiologic agents. However, having ongoing surveillance systems provides a contact, an infrastructure, and a series of relationships such that detection can occur and a response can be mounted. Some fundamentals are needed to do surveillance, such as an organized health care system, a system for classifying disease and injury, and measurement techniques. Surveillance has many different activities. For the purposes of this group, I would focus on triggering investigations or detecting epidemics as the best uses for public health surveillance. You have heard about a number of different surveillance systems in the United States. Of these, the notifiable disease reporting system is the one that probably would serve as the basis for recognition of some sort of unknown pathogen or etiologic agent. This system exists in each of the 50 states and serves as a significant link between the Centers for Disease Control and Prevention (CDC) and the state health departments. There are other systems, and others are being developed-among them laboratory-based surveillance for antibiotic resistance patterns and hospital-based surveillance. What is a notifiable disease? Some people don't realize this, but CDC cannot designate which diseases are notifiable. This authority is within the purview of either the state health board or the legislature of any given state. When AIDS became a notifiable disease in the 1980s, it was because laws or regulations were passed in each of those states. CDC works with the Council of State and Territorial Epidemiologists to determine what the notifiable diseases are. We currently have 51 diseases on our list. Diseases are added and deleted from this list. For instance, Escherichia cold 0157 has been added, as has antibiotic-resistant Streptococcus pneumonias. Cases of these diseases are reported to CDC on a weekly basis. 59
60 BLOOD AND BLOOD PRODUCTS: SAFETY AND RISK You have heard a little bit about passive versus active surveillance. A passive system is one where physicians are required by regulation or law to report these diseases to local or state health departments. An active system, on the other hand, is one where reporting is initiated by a state health decal l~nent. The data are transmitted to CDC from state health departments electronically on a weekly basis and are published in the Morbidity and Mortality Weekly Report. For example, Figure 1 shows data tracking the decline in the number of cases or rate of malaria following World War II and subsequent rises primarily from returning veterans or foreign immigration. These data are available, and they are used to follow long-term trends. 1000 - ct too U) c' ~ to ce - Ct 0.1 a: l ~ Relaoses -- Overseas cases \ ~ Relapses from Korean veterans ~ Retuming Vietnam veterans /~\ JIBE / \/ ~ foreign immigradon ..... - . 0.01 1930 1935 1940 1945 1950 1955 1960 1965 1970 1975 1980 tam 1~ 1~5 FIGURE 1 Reported cases of malaria per 100,000 population in the United States, 19301992. How good are the data? Notifiable diseases carry with them regulations, fines, and various admonitions if a physician does not report a case of a particular notifiable disease. Yet we clearly know that there is a lot of underreporting. Table 7 presents data from Vermont comparing the number of patients hospitalized with notifiable diseases with the percentage of cases that are actually reported. Underreporting clearly is a problem, even though reporting is mandated by law.
GUARDING THE BLOOD SUPPLY TABLE 7 Evaluation of Notifiable Disease Reporting, Vennont, 1982-198352 61 Disease Hospital Cases Percent Reported Hepatitis 64 31 Aseptic meningitis 127 6 Bacterial meningitis 65 62 Gonorrhea 30 95 Pertussis 15 40 Salmonellosis 63 67 We have similar data from Washington, DC. They are about 10 years older, but the trends have basically been the same. There is tremendous underreporting of disease, except when new diseases occur, such as AIDS. The case reporting for AIDS has actually been fairly good for two reasons: it is a relatively new disease and it is an exotic disease. Also, a lot of money was put into surveillance for it. There are several reasons for underreporting, and these are basically relevant to any surveillance system, particularly passive ones. There is often a lack of knowledge of reporting requirements, and there are negative attitudes toward reporting. There are misconceptions and even suspicion of the government and what it may be doing with this particular data. What are the ways to improve the surveillance system? Make it simple. Systems are often far too complex for what they are intended to do. Provide frequent feedback, widen the net, get increasing sources of information, and conceivably, do active surveillance. There has been talk about active surveillance. The advantages are that you can identify all the cases, you get better-quality data, and some of the data may be useful in special circumstances. The disadvantages are that it is incredibly time-consuming and costly, and the additional data may not be worth the cost, except in special circumstances. When new diseases emerge, we must have sufficiently simple, flexible, and acceptable systems already in place such that surveillance for these new pathogens or new disease entities can be incorporated into these systems quite readily. Using as an example the current attention being directed toward emerging infections, CDC is undertaking efforts to improve its surveillance activities in 52 Vogue, RL, SW Clark, S Kappel (1986). Evaluation ofthe state surveillance system using hospital discharge diagnosis, 1982-1983. American Journal of Epidemiology, 123: 197-198.
62 BLOOD AND BLOOD PRODUCTS: SAFETY AND RISK this arena. A number of major factors are contributing to the emergence of infectious diseases: human demographics and behavior, technology and industry, economic development and land use, international travel and commerce, microbial adaptation and change, and breakdown of public health measures. Increased population density and population encroachment, along with increases in international travel, are the same factors that would likely result in the introduction of some particular pathogen into the blood supply. The top seven emerging infectious diseases in 1993 were cryptosporidiosis, coccidioidomycosis, E. coliO157:H7 disease, multidrug-resistant pneumococcal disease, vancomycin-resistant enterococcal infections, influenza A/Beijing/ 32/92 virus infections, and hantavirus infections. I don't think these are necessarily ones that could be found in the blood supply, but the potential is always there for some emerging pathogen to find its way into the blood supply. These are new diseases or emerging diseases for which we need to fonn responses. Figure 2 provides data on the incidence of one of these, hantavirus. Hantavirus is a good example of the detection of a disease that wasn't previously notifiable. Emerging diseases are reported because people at the front line of public health (the public health nurses, physicians, primary care doctors, and medical examiners) notice an unusual syndrome and call up health officials. That is how you detect unusual or new diseases. 14 12 10 8 4 2 ~J S N |J ~ J S N 1992 1 1993 J ~J 1 994 FIGURE 2 Reported cases of hantavirus pulmonary syndrome in the United States, January 1-August 31, 1994 (1 1 cases prior to 1992 not shown).
GUARDING THE BLOOD SUPPLY 63 There have been several notable instances of outbreak detection going back to the 1970s. Legionnaires' disease was first detected when a Veterans Administration pathologist came in to the morgue over the weekend and found three or four elderly men who had died from pneumonia. He called up a friend at CDC to report that something was going on. You know the story of AIDS in terms of a physician noting a cluster of illnesses in a certain group as well as the increased medication use that was alluded to earlier. Hantavirus was reported by a medical examiner who called up a medical examiner at CDC. E. cold 0157, which caused the famous multistate hamburger outbreak, was detected because a pediatric gastroenterol-ogist noticed hemolytic-uremic syndrome in a young child. The salmonella-contaminated ice cream outbreak that began in Minnesota was detected by the use of state lab serotyping data. In the state lab in Minnesota where the serotyping takes place, they found an increased incidence in one of the serotypes, which resulted in an investigation and detection of a multistate outbreak. Surveillance provides an infrastructure and relationships such that if something does happen, the person who notices something unusual can make the phone call. The calls can then be channeled to the appropriate persons. The CDC Epidemic Intelligence Service (EIS) is the active aim of CDC in terms of doing outbreak investigations. It was founded in 1951 because of concerns about biological and chemical warfare at the height of the Cold War and the Korean crisis. It is a training program that has graduated approximately 2,200 individuals from the program since its inception. EIS officers are not the only ones who may detect outbreaks, but the training program serves as a very useful informal network, linked by a yearly directory, for making contacts with our colleagues as we identify new and unusual things. We are ingrained Mom day one as EIS officers with the steps of an outbreak investigation: . . . . establish existence of outbreak, verify diagnosis, define and identify cases, and characterize by time, person and place. Then we must studies, and . develop hypotheses, evaluate hypotheses, refine hypotheses and conduct additional lab and environmental implement control measures.
64 BLOOD AND BLOOD PRODUCTS: SAFETY AND RISK You learn these in a relatively rote way, but you find that they are useful for almost any situation, including the unknowns. In the EIS network, 50 officers are currently have assignments in 26 state health departments. We have graduates in field epidemiology training programs in 19 countries. We also have a very active international visitor exchange program, such that often colleagues who are overseas detect things such as the Ebola virus outbreak in Zaire. They then give us a call and things begin to move along. It is an informal network, but it is a relatively effective one. It has worked well in the past in terms of identifying unknown pathogens. Finally, we are researching the history of the n otifiable disease surveillance system because it underscores a lot of interesting relationships between the federal government and states. Disease reporting is a state function, and Henry Baker, who was quite clairvoyant back in the beginning of this century, made this argument for collecting information on disease and health. "The only way to learn what diseases cause most sickness is to collect the statistics of sickness."53 That serves as our foundation for identifying new and unusual pathogens. We continue to do that. Hopefully, we can do a better job at it. DISCUSSION Bernard Horowitz: You described a system that was put in place several years ago that is an extension of an older one for monitoring hemophiliacs. One of the arguments is to assess blood safety. By the very nature of the treatment of hemophilia A or hemophilia B patients, who are largely treated with concentrates which are highly purified and virally inactivated with solvent/detergent or other other techniques, it is not as complete a reflection of safety from viruses as, for instance, the use of other patient groups such as those with thalassemia. Are you aware of comparable systems for other groups of patients, or in what way are you using rare hemophiliacs to understand the infectivity of donated blood? Bruce Evatt: I am not aware of other groups. I think that hemophilia patients are a unique group in that they are exposed to blood products from large numbers of donors. They are a sentinel group, but they now are also 53U.S. Public Health and Marine Hospital Service (1903). Transactions of the First Annual Conference of State and Territorial Health Officers with the United States Public Health and Marine Hospital Service. Public Health Bulletin, No. 11.
GUARDING THE BLOOD SUPPLY 65 receiving more and more purified kinds of products. It is expected that their transfusion exposure will decrease in future years as they go to all recombinant products, except for the blood transfusions that they require. James Allen: If money is to be made available for surveillance for transfusion-transmitted diseases or special issues of that sort, we need to be certain that we have a very strong local, state, and federal disease investigation system and general surveillance process rather than looking to build a superspecialized type of vehicle for something unknown. Instead, our general surveillance system must be very strong. I have very real concerns about the strength of that today.