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.
Interim Findings and Recommendations This chapter presents the interim findings of the committee, eight months into a 24-month study, regarding the Centers for Disease Control and Prevention (CDC) research program on the anthrax vaccine. Within the CDC, the research activities related to the anthrax vaccine have been under the purview of two parts of the agency, the National Center for Infectious Diseases (NCID) and the National Immunization Program (NIP). As was mentioned in the previous chapter, the committee was charged with advising the CDC on the completeness and appropriateness of the agency's plan to study the safety and efficacy of anthrax vaccine, addressing the following aspects of the Congressional mandate: risk factors for adverse reactions, including gender differences, the determination of immunologic correlates of protection and documentation of vaccine efficacy; ant! optimization of the vaccination schedule and routes of administration to assure efficacy while minimizing the number of doses required and the occurrence of adverse events. Also as described previously, the anthrax vaccine and the Department of Defense (DoD) policy decision of universal vaccination for troops have been associated with concern among some military personnel and their families. The dissatisfaction couict result from various combinations of factors' including potential for local or systemic adverse events, the number and timing of injections, the perceived risk of exposure to B. anthracis spores, and beliefs about the level of protection affor(lecl by immunization against anthrax. The CDC has developed a research agenda inclucling projects clesigned to study adverse events and correlates of immune protection and to increase the acceptability of the anthrax vaccine. At the same time, other projects will address the administration of the vaccine so as to assure that changing the route of administration and the number of closes (so as to reduce adverse events) at least maintains the level of safety and efficacy of the current vaccine. The safety component focuses on true adverse events associated with immunization, including both improved accuracy in detecting adverse events and modifications to the product and its use that wouIc! reduce the likelihood of adverse events. The CDC research plan also inclucles projects addressing the acceptability of the anthrax vaccine. The CDC has not yet described a comprehensive plan explaining how the projects fit its overall goals. The committee was not able to assess the plan's completeness and appropriateness with respect to its goals while the plan itself was still cleveloping. At the time of this interim 21
22 CDC ANTHRAX VACCINE SAFETY & EFFICACY RESEARCH PROGRAM report, it is not clear whether the CDC as a whole is approaching the research primarily with a view to changing the labeling of the current AVA product, or with a view to furthering scientific unclerstanding of the anthrax vaccine and its use and context more generally. These goals are not inconsistent, but if both exist then a comprehensive plan should describe their integration. In any case, some of the studies planned by the CDC are not relevant to changing the labeling, but wouic! be contributions to the wider context of anthrax vaccine research. It is therefore very important for the committee to understand as soon as possible how the CDC interprets that context, how they have identif~ec! missing elements which their plans will address, and what may yet remain to be done. In the absence of such a comprehensive plan, it is difficult for the committee to evaluate fairly the CDC's plans for completeness anct appropriateness. At the time of this interim report, the committee's general recommendations below reflect, first, the need for a comprehensive research plan from the CDC, that is, all the anthrax vaccine relater! research the CDC contemplates, and how that research wouIc! fit into the wider anthrax vaccine research context. In addition, the committee notes that the CDC often seems to be seeking scientific advice on specific points of protocol development of a highly technical nature that wouict be most useful to them if provider! directly and immediately rather than in a formal published report. Finally, the committee notes the scientific importance, and also the scientific opportunity, attendant upon potential collaboration between the CDC and both the DoD and the NIH and encourages the CDC explore fully areas of such potential. The committee highlightecl the example of analyzing potentially complementary databases, but also realizes other areas of collaboration may emerge in later discussion. The committee made general recommendations on these points (see Box 2-~. BOX 2-1 General Recommendations for CDC's Anthrax Vaccine Research Program General Recommenclations The CDC should produce a comprehensive description of its research program, in- cluding statements of the goals of the program and how the plans now undertaken will meet those goals. In addition, the CDC should continue and complete develop- ment of the individual projects in the research program. The CDC should consider engaging protocol design consultants representing broad scientific expertise who would provide immediate and direct consultation on specific technical matters of study design and execution. The CDC should continue and strengthen collaboration with the DoD wherever pos- sible, including for example much more extensive use of DoD databases such as the Defense Medical Surveillance System (DMSS).
INTERIM FINDINGS & RECOMMENDS TIONS 23 The NCID plans inclucle a human clinical trial of changing the route of inoculation with the vaccine from subcutaneous to intramuscular, and reducing the total number of closes. Other NCID projects support and extend this work using animal trials and laboratory assays on basic pathogenesis, immunology, and correlates of immunity. The clinical trial and some elements of the laboratory and animal studies are clirectly and forthrightly designed to satisfy regulatory requirements of scientific support for a change in the labeling of the licensed anthrax vaccine AVA. The plans for the clinical trial and related studies seem to be complete and appropriate with respect to the goal of changing the labeling directions regarding route of administration and dosage schedule. Other related studies appear not to be necessary for the labeling change, but perhaps are important in the wider context of anthrax vaccine research. If so, then an assessment of the completeness and appropriateness of those other studies should properly be made from a different perspective, that of the larger research context. The apparently clual goal of the set of studies, in part to support a labeling change and in part for a wicler research contribution, creates challenges for an overall review of the study plans. In particular, the committee notes! several omissions in the set of study plans where the omitted study might be unnecessary for the labeling change but quite important in the wider research context, including future studies of safety and efficacy of products yet to be approved. The primary elements of the CDC research plan uncler the purview of the NIP include studies of the knowledge, attitudes, and beliefs of populations regarding aspects of the anthrax vaccine immunization program (AV1P), and also include several approaches to making better and more intensive use of existing data relevant to the safety ant! the efficacy of the vaccine. These projects are clearly intencled to contribute to the larger anthrax vaccine research context. In general, the NTP-sponsored studies are not as far along in their clevelopment as are the NCID- sponsored studies related to the labeling change. This interim report reflects two results of the difference in the development time lines. First, the studies that appear to be designed specifically with a view to supporting a labeling chance net greater scrutiny at this time than do anv of the 1. 1 - ~ ~ ~ ~ ~ ~ ~ ~ . . . .. . . . . ~ .. . . . . . . studies pertaining to the wider context ot anthrax vaccine research, simply because the committee received more detailed information for review and discussion. Second, the distribution of studies within the CDC results in the NCID-sponsored studies (since they are concerned with the labeling change) getting more scrutiny than the NIP-sponsored studies. Again, this is an artifact of the relative stage of development, and hence of the relative amounts of information available, in each case. The projects in either part of the CDC that were describer! in full or in part to the committee by the time of writing the interim report are listed in Box 2-2. NCID RESEARCH ACTIVITIES The NCID issued a call for proposals to carry out one or more parts of the following three- part research plan. Part A: Anthrax Vaccine Adsorbed (AVA): A Human Reactogenicity and Immunogenicity Trial to Address Change in Route of Administration and Dose Reduction Part B: Non-human Primate AVA Dose-Ranging, Immunogenicity ant! Challenge Trial . Part C: Anthrax Pathogenesis, Immunology, and Correlates of Protection Against Inhalational Anthrax
24 CDC ANTHRAX VA CCINE SAFETY & EFFICA CY RESEARCH PROGRAM Part A: Clinical Trial The Part A clinical trial is designed to evaluate the effects of changing the administration ot the vaccine from the current subcutaneous (SQ) route to the more common intramuscular (IM) route, anct to evaluate whether this change affects the rate of adverse events in male and in female recipients. Part A also will evaluate the efficacy of the vaccine uncler the new conditions, based on correlates of immunity, that is, laboratory tests of the level of antibody in the blood that is correlated with immunized humans and with animals that have been shown to be protectect from infection. Then that information will be the baseline for evaluating the effects of reducing the total number of vaccine doses. The labeling currently specifies six closes given over IS months, followed by annual boosters, but the study as currently planned will assess reclucing that to as few as four doses over IS months with boosters every two years. The trial will be double blinded, randomizecI, and controlled using saline as the placebo. The active treatments will use the current supply for the first four doses (Iot FAV-048b), and a different lot (to be proclucect by BioPort, but used only if certified by FDA as releasable) for subsequent closes. Given the length of the data collection period, the trial must last at least three and a half to four years. Box 2-3 summarizes the study design. BOX 2-2 Projects Presentec! to Date in CDC Research Program NCID- in development Part A: Anthrax Vaccine Adsorbed (AVA): A human reactogenicity and immunogenicity trial to address change in route of administration and dose reduction Part B: Non-human primate AVA dose-ranging, immunogenicity and challenge trial Part C: Anthrax pathogenesis, immunology, and correlates of protection against inhalational an- thrax Human leukocyte antigen sub-study NIP - in development or planned for development Survey of knowledge, attitudes and beliefs (KAB) prevalent among military service per- sonnel and also of military health care provid- ers regarding the anthrax vaccine Survey of the KAB prevalent among military health care providers regarding practices of re- porting to Vaccine Adverse Event Reporting System (VAERS) Data-mining in the VAERS database SF-36 Survey of clinical trial participants Hormonal correlates of adverse events in fe- male clinical trial participants Long-term follow-up of any available previously immunized populations Meta-analysis of safety and efficacy studies on the anthrax vaccine
INTERIM FINDINGS & RECOMMENDATIONS BOX 2-3 Saucy Design for Part A Clinical Trial Purpose: To evaluate the effects of changing the route of administration from subcutaneous (SQ) to intramuscular (lM), and of reducing the number of doses. Group Route Vaccine N Primary (wks) Boost~mns) 1 6SQ AVA 260 0-2-4 6-12-18-30-42 2 51M AVAIS* 260 0-S-4 6-12-18-30-42 3 41M AVA/S 260 0-S-4 6-S-18-30-42 4 41M AVA/S 260 0-S-4 6-S-18-S-42 5 IM S 260 S-S-S S-S-S-S-S *S = saline placebo 25 The research sites that will be participating in the Part A trial include Baylor College of Medicine in Houston, Texas; Emory University Vaccine Center in Atlanta, Georgia; The Mayo Clinic and Foundation in Rochester, Minnesota; and the Walter Reed Army Institute of Research in Silver Spring, Maryland. As shown above, the plans call for a total enrollment of 1300 participants; a new call for proposals has been issued in order to include an additional site so as to increase enrollment and meet that goal. The committee found that the proposed research on changing the route of administration of the vaccine and reducing the number of doses required is appropriate. These results also are ~ 1 . , · · 1 , 1 .1 · ~ . ~ ~ ~ · .~ necessary trom a regulatory perspective In order to change the vaccine product labeling so that the military could change its practice and remain in accord with FDA-approved guidelines. The committee considers some potential mollifications, particularly to one group, of the human clinical trial as perhaps advantageous, while recognizing that it may be difficult to implement any further changes at this date. . The committee recommends that, in the human clinical trial, the CDC should consider including a study group immunized at the start of the series (time zero), and one and six months later, followed by placebo, in order to assess adequacy of a simplified three dose regimen in the development of immediate and long-term immunity to anthrax. The committee believes this couIc! be done by modifying the regimen of the existing stucly group two, since that regimen has airea(ly been investigated in a pilot study. In the table in Box 2-3 above, Group 2 wouict then appear as follows: 2 3IM AVA/S* 260 0-S-4 6-S-S-S-S The committee has not yet seen the methocis section of the human clinical trial protocol, notably the specific types of statistical analyses that will be employed in the analysis of clinical trial data. The committee is concerned about the methods to be used. . The committee recommends careful selection of statistical methodologies, as certain techniques including intent-to-treat analysis may be less appropriate in developing conclusions for what will eventually be a military application than they would be for general civilian vaccine development. The committee recognizes that the human clinical trials are intended to investigate gentler differences related to adverse reactions, as specifically mentioner! in the Congressional mandate.
26 CDC ANTHRAX VA CCINE SAFETY & EFFICA CY RESEARCH PROGRAM Previous work (Pittman et al., submitted for publication) has shown a significantly greater number anc! extent of local adverse reactions to subcutaneously administered anthrax vaccine in women. The CDC'S plans for recruitment of volunteers anticipates a sufficient number of female subjects, 20 percent of total enrollment, to detect a difference of similar degree, but not for a smaller difference. The CDC also plans a sub study of hormonal status that would further subdivide the female volunteers, so that study wouict require a larger number of participants. As planned, the clinical trial would be powered adequately to detect a substantial difference related to sex (a biological parameter) or gender (a psychological and/or social phenomenon) in the rate of adverse reactions to subcutaneous inoculation, given successful recruitment of female subjects. However, several factors may complicate such recruitment. Belief among potential recipients that women may suffer more reactions (also indicated in the Pittman study), and plans for recruiting study volunteers from, among other groups, first responders (traditionally professions employing more men than women), together cause the committee to cloubt the likelihood of enrolling female volunteers in sufficient numbers to provide adequate power for the planned studies. As planned, this study will not have adequate power to address aspects of sex or gender differences beyond route of administration. The military population and organization, however, may provide especially gooct opportunities for additional research inclucling pharmacoepidemiologic studies (Tilson, 1992; Strom 2000) or perhaps active surveillance among military personnel. . The committee recommends that the CDC consider, in addition to the proposed clinical trial, a prospectively (lesigne(l pharmacoepi(lemiologic stiffly of military vaccine recipients with both active surveillance and his- torical data from DMSS for moderate to severe adverse events in order to assess sex or gender and perhaps other risk factors for adverse events among military personnel receiving the anthrax vaccine. The committee deliberatecl long and carefully regarding the use of saline, as opposed to the adjuvant aluminum hydroxide or alhydrogel. The question of which substance is the appropriate placebo reflects the question of what is the most important variable to measure in this trial, ant! thus to which vaccine component the participants and investigators must be blinded. Since the AVA product is the vaccine adjuvantec! with aluminum hydroxicle, the question of adverse events due to the product (the complete adjuvanted product) would be best addressed by an aluminum- free placebo such as saline solution. Alternatively, since the active ingredient is PA, the question of adverse events due to the antigen alone would best be addressed by an aluminum hydroxide placebo. The committee also discussed whether it wouicl be ethical to use an aluminum hydroxide placebo. Some members argued that such a placebo would cause discomfort to the subject perhaps without any personal benefit, and others pointed out that the subjects would be civilian volunteers, not deployable and not likely to be exposed to biological weapons, and thus the groups receiving the vaccine would probably not benefit personally either. In the end, the committee die! not develop a consensus on the best choice of the placebo. CDC proposalfor use of the Short Form 36 The CDC proposed to acid to the human clinical trials a self-assessment using the Short Form 36 (SF-361. The SF-36 is a survey instrument developecl in 1992 by the RAND Corporation for self-assessment of health status. This survey has been very well studies! and extensively validated. The committee found that the rationale for including such a measure in the human clinical trial was appropriate. However, the SF-36 can be augmenter! by existing, validates! items
INTERIM FINDINGS & RECOMMENDS TIONS 27 to acIdress populations more specifically. Further, the time of administration of the survey is quite important so clinical personnel should have specific instructions about administering the survey, and also about the use of the results (as some may be relevant to the examination of the patient in the context of the immediate visit). The committee recommends that the CDC consider including additional items with the SF-36 specific to adverse events possibly associated with immunization, and clearly indicate how the use of the SF-36 will be included in the protocol. Part B: Non-human Primate Dose-Ranging and Challenge Stu(ly The animal studies would compare different doses of vaccine to establish the appropriate dose for the rhesus macaque, the animal species of interest, ant! the effect of reducing the number of doses on the animals' circulating antibody and ability to survive challenge with aerosolized anthrax. The committee found that supporting animal studies are of great importance in an anthrax vaccine safety ant! efficacy research program. It is not possible to do a human clinical trial of vaccine efficacy for inhalational anthrax: the disease has a high fatality rate, and improved working conditions in goat hair and woolen mills, as well as increased use of synthetic materials, have (fortunately) eliminated the major source of naturally occurring cases in the United States. In brief, the strategy for supporting the efficacy of a vaccine for which human clinical trials with challenges could not be clone is baser! on correlations of laboratory measurements in humans ant! animals. Some of these steps in anthrax vaccine may have been done or be underway but, in general, such a strategy might work like this: (~) Passive animal-animal immune transfer study: Immunize animals of a suitable species, measure levels of antibody, in serum, then give different concentrations of animal antibody-containing serum to naive animals (i.e., passively immunize the new animals). Then challenge the passively immunized animals along with some non-immunizect control animals with aerosolized B. anthracis spores. Note the concentrations of antibodies associates! with surviving the challenge. (2) Passive animal-human immune transfer stucly: Repeat using human antibody-containing serum from humans vaccinated with a filet course of anthrax vaccine, then challenge the immunized animals (anct controls). Note the concentrations of antibodies associated with surviving the challenge, and compare the levels of animal and human antibody needed to provide passive protection allowing animals to survive challenge. The results provide a surrogate for clinical effectiveness of the vaccine, since clinical effectiveness in humans cannot be tested. The surrogate would be the ability of a vaccine to stimulate, in humans, levels of antibody correlated with the level necessary to protect animals from inhalational challenge by passive immune transfer. Not all of this may be necessary for a labeling change in the current product, because as describer! in the previous chapter, the current product was developed using data from an actual human field trial (though of a slightly different vaccine and concentrating on cutaneous anthrax). But for development of future anthrax vaccine products, it will be absolutely necessary to establish a complete set of steps correlating animal immune response to circulating antibody in the fully immunized human. But because it was not entirely clear to the committee whether the
28 CDC ANTHRAX VA CCINE SAFETY & EFFICA CY RESEARCH PROGRAM goal of some of the studies planned by the CDC was the labeling change only, or also the larger research context, it was likewise not entirely clear whether the research program was simply limited in scope, or seriously incomplete, in the absence of some steps in establishing the correlation sketched above. As a model for effects in humans, animals that develop inhalational anthrax, rhesus macaques, will be immunized using different schedules. In earlier trials, however, the macaques received the same dose of vaccine as a human would, that is, much more vaccine per unit body weight than a human would receive. The CDC plans a dose-ranging study to establish the appropriate dose of vaccine for a rhesus macaque as a basis for future studies. The animals will be immunized with different dilutions of vaccine, and then their immune systems will be challenged by exposing them to a dose of aerosolized B. anthracis spores, to evaluate whether they will have developed an antibody level that protects them *om infection. This study is being overseen by the Batelle Memorial Institute. Often such a series of studies would be preceded by passive human-to-animal and animal-to- animal transfer studies as described above. The CDC's plans do not currently include such studies to demonstrate passive transfer of serum antibody to macaques and determine the level of antibody necessary for protection, and the committee believes this omission should be addressed either in the CDC plans, or by reference to work done or underway elsewhere, from the point of view of establishing correlation of immunity in the larger context of anthrax vaccine research. The committee discussed the dose ranging trial at some length, with particular concern for the dilution series specified. The plan calls for two-fold dilutions after the first step of the series, which would be a l:5 dilution, has large steps (l, I:5, 1:10, 1:20, etch. The committee suggested that the series might miss important information, and that the series should be two-fold all along (l, I:2, I:4, I:S, etch. The committee emphasized further that the basis of these (and many future) studies should be a passive antibody transfer study, as previously discussed. The committee concluded finally that the CDC research plan remained incomplete in that it lacks a study demonstrating passive transfer of human serum antibody to macaques and determining the level of antibody necessary for protection, unless such a study is known to be underway elsewhere. The committee recommends that the CDC consider both the addition of a passive antibody transfer steely, and that the animal friar dose-ranging study design include a more appropriate dilution series. The design of the challenge study is shown in Box 2-4 below. The committee did not make additional recommendations on the challenge study, but urges, as in the case of the human clinical trial, careful consideration at the outset of appropriate methods of statistical analysis. Part C: Assays of Immunologic Correlates of Protection Part C is related to both Parts A and B above. Part C is a series of assays, listed in Box 2-5 below. The overall goal of the series is to validate currently proposed immunologic correlates for protection, especially anti-PA antibody levels, and possibly to identify new immunologic or genetic correlates of protection. The Part C Assays are to be carried out by Emory University and overseen by the Batelle Memorial Institute, which has arranged subcontracts with Ohio State University in Columbus, Ohio, and with the Centre for Applied Microbiology and Research in Porton Down, United Kingdom.
INTERIM FINDINGS & RECOMMENDATIONS 29 BOX 2-4 Study Design for Part B Non-human Primate Trial Purpose: To evaluate the effects of dropping doses at 12 and 30 months on immunogenicity and survivability, to determine the amount of circulating antibody at challenge, and to assess the role of memory response. Group Route N* Schedule Aerosol Challenge 1 IM 10 0-4 wks, 6 mns 12 mns 2 IM 10 0-4 wks, 6 mns 18 mns 3 IM 10 0-4 wks, 6-18 mns 30 mns 4 IM 10 0-4 wks,6mns 30mns * + 2 controls for each group BOX 2-5 Part C Assays . . . . . . T Cell proliferation, activation, cytokine production Enumeration of PA, LF, and EF specific plasma cells Cytotoxic T cell responses to anthrax toxin Anti-PA monoclonal antibodies for epitope mapping of the PA toxin Anti-PA, EF, LF ELISA Anti-PA IgG subclasses 1-4 ELISA TNA titers as measured by PA-633-LF cytotoxicity, LF endopeptidase inhibition, EF adenylate cyclase inhibition assays Opsonophagocytosis via flow cytometry using HL-60 mRNA and protein levels of TH1 and TH2 cytokines cAMP assay High-avidity serum IgG ELISA The Part C assays use blood cirawn from the Part B non-human primate trials and also from the Part A human trials. The committee found that the scientific account of imn~une correlates has developec! rapidly and commendably in the case of the plans for most of the assays. One assay proposecl by the CDC, however, was still not fully explained. The CDC proposes to use microarray technology to assess whole-cell gene expression in response to vaccination or challenge. The CDC noted at the time of presentation that the technical proposal for use of this assay was still in development; indeed, at that time the rationale for the use of microarrays was not yet clear to the committee. The committee was also not convinced, however, that the opsonophagocytosis assay as described would be likely to be very informative. . The committee recommends that the use of microarrays receive further critical attention and precise evaluation of what information will be gleaned and how it will be interpreted anc! applied to anthrax vaccine recipients. The CDC also plans two substudies in association with the Human Clinical Trial or Part A. These two studies seem not to be intencled to support a labeling change, but rather to contribute to the wider research context. Again, the problem is that it was not clear to the committee at this stage in the development of the overall research program how the CDC envisions these particular stucties fitting into that wider context. One of these would be to investigate at one study site, the
30 CDC ANTHRAX VA CCINE SAFETY & EFFICA CY RESEARCH PROGRAM Mayo Clinic, any influence of human leukocyte antigen (MEA) on immune response to the anthrax vaccine. The committee fount! that the rationale for the HEA substudy was not yet complete and recommencis further explanation of its role if it is to be part of the CDC's research program. The other substudy would address whether there is any association between progesterone levels and vaccine adverse events among women. The committee found that it hac! not yet heard the plans in sufficient detail to make a full assessment of the study. The committee did express doubt about the likelihood of recruiting sufficient female volunteers to carry out the investigation with adequate power, as was ctiscussec} under Part A above. NIP RESEARCH ACTIVITIES The NIP has begun work on a variety of types of research, including studies of the personnel actually involved in the vaccination program, either as intended! recipients or as health care providers, and also studies to make farther use of existing data. As previously mentioned, these studies appear to be intended as contributions to the wider context of anthrax vaccine research, not to the particular immediate project of changing the labeling. The NIP has proposed studies that are quite varied in technique and rationale, and at the time of writing the interim report, also varies! widely in degree of development. The committee kept in mind the preliminary nature of the proposals while assessing the program. On the basis of the information presented to slate, the committee looks forward to further review of comprehensive, and of aciclitional cietailect, information as soon as it is available. Knowledge, Attitudes, and Beliefs of Military Personnel about the Anthrax Vaccine The NIP has requested proposals to carry out a survey of the knowledge, attitudes, ant! beliefs (KAB) prevalent among military personnel regarding the anthrax vaccine. This study, which has not yet begun, will be designed to describe the KAB of military service personnel in general and also of military health care providers regarding immunization against anthrax. The resulting descriptive information will, it is hoped, provide the basis to assess the reasons for concern about anthrax vaccination and the eclucational needs pertaining to the anthrax vaccinations of the populations stuclied. The first step of the study wouIcl be to gather information from a series of focus groups, in order to develop a survey instrument informer} by a broad and representative example of individuals from the study population. The actual survey was planned to include a sample of some 14,000 individuals using cluster sampling to assure representation of all service and regional components of the military population. In all cases, participants would be asked to complete the survey in a classroom setting (and rewarded with a non-monetary incentive), in order to avoid the consistently low rate of returns from mailer! surveys. After the first survey and a sufficient period for planned interventions (expected to be two to three years), a follow-up survey would be administered to a newly selected group (again, selected by cluster sampling, and preceded by a focus group exercise). Finally, in order to evaluate more reliably the effectiveness of the interventions in affecting the KAB of services personnel, the overall study design incorporates a longitudinal subgroup a small cohort that will be identified in the first large survey and then relocated and resurveyed in the second part of the study (see Figure 2-~. The committee found that the rationale for investigating the KAB of service personnel was appropriate. The committee found, however, that the plans for the study were incomplete with respect to plans for the design of the educational interventions and survey instrument, and
INTERIM FINDINGS & RECOMMENDA TIONS 31 commented that entire stucly might be strengthened by investing more in its design phase. For example, in addition to careful use of focus groups, the educational interventions could be refined through cognitive tests in small groups so that both the educational interventions and the survey instrument were more sharply focused in purpose and therefore more effective. The committee also would appreciate farther discussion of how the protocol design will address the turnover in the military between surveys. The committee anticipates that plans for this study could undergo significant further development, and will be very interested! in any further information. In public hearings the committee heard from current and former military service members outside the health care system about their perceptions of the military health care system. The KABs of both patients anc3 providers are crucial in any health care program; in the anthrax vaccine program in particular, the need to unclerstand the KAB of all involved parties is pressing. The committee heard expressions of significant concern from current and former military service personnel, supporting the importance of obtaining reliabe KAB information. The committee also found that the KAB of health care providers would be likely to differ from military personnel in general and to play an especially important role in forming perspectives on the use of the anthrax vaccine, indicating possible need for a different survey instrument in the case of this group. Such a survey might be carrier! out in combination with another planned survey of military vaccine providers, discussed further below. Atlditionally, both surveysor perhaps even a third, specifically directed surveymight profitably be designecI to generate important information about perceptions of the military health care system. Concerns about the AV1P, and the use of civilian as well as military health) care facilities, may have effects on the KAB of military personnel, as the military may be perceived by some service personnel (however unfairly) as taking a less direct interest in the health of its troops. That perception might then affect the confidence and trust of military patients in their health care providers and the military health care system. Finally, the committee found that there was currently insufficient justification for the sample size suggested. l Phase I Focus Group | Phase 11 Focus Group 1 . 1 Time passes; educational inter- ventions and other events occur that ~ could affect KAB Longitudinal Longitudinal Cohort Cohort KAB AT BASELINE KAB AT FOLLOW-UP Figure 2-l KAB Survey of Military Personnel Regarding the Anthrax Vaccine . The committee recommends that the CDC consider expan[ling the (resign phase of the KAB study of military personnel regarding the anthrax vaccine to include cognitive and psychometric tests and a pilot survey in order to design both the educational interventions and the survey that will relate to them, in order to refine the sampling plan.
32 CDC ANTHRAX VACCINE SAFETY & EFFICACY RESEARCH PROGRAM Knowledge, Attitudes, and Beliefs of Vaccine Providers about VAERS Reporting The NIP has requested proposals to carry out another type of survey of the KAB prevalent among military ant! civilian providers of military vaccines regarding reporting to the Vaccine Adverse Event Reporting System (VAERS). This study, which shares some elements of its ctes~gn with the study ctescr~nea just above, also Is not yet underway. in this case, the descriptive information gathered in the study should allow the assessment of the level of awareness and knowledge of VAERS among vaccine providers, factors affecting reporting of adverse events among vaccine providers, and providers' beliefs regarding both the ease and the utility of reporting adverse events. Again, the study is planned as two surveys. These surveys will be conducted by telephone interview to avoid the low return rate of mailed surveys. The populations surveyed will include military physicians and nurses; other military health care providers, especially vaccine providers' civilian physicians and nurses; and public health clinic staff members, again using cluster sampling to assure diversity in the sample. The committee heard testimony from some concerned service members and other citizens indicating that they believe that the anthrax vaccine may be associated with more adverse events than reflected in VAERS, due at least in part to aspects of military life that may tend to discourage the reporting of adverse events. Based on the information available at the time of this report, the KAB study of health care providers with respect to VAERS reporting seems to be directed primarily toward investigating that concern, that is, toward enhancing VAERS reporting, especially in the military. The committee also observed, however, that as a passive surveillance system, VAERS can never be expected to serve as a source of information about the true rate of adverse events. Specifically, the rate of reporting adverse events is typically low relative to the true rate of adverse events, and fluctuates for many reasons. In the particular case of the anthrax vaccine, it may be that military culture has in some way tendec! to discourage reporting, but at the same time it may also be that publicity from news coverage and congressional hearings has tended to encourage reporting. In order to use information from VAERS regarding the anthrax vaccine, it would therefore be very important to compare reporting related to the anthrax vaccine, to reporting related to other vaccines such as tetanus ant! influenza. Incidence rates cannot be determiner} from spontaneous reports, so low numbers of VAERS reports about a product cannot clefinitively establish the safety of the product. Such low numbers reflect an absence of a signal of a problem and can never substitute for data from a formal pharmacoepidemiology study evaluating safety. Low numbers are thus consistent with (but not proof oily safety, and high numbers might likewise signal the lack of safety (but cannot definitively establish that either), as generally nothing definitively causal can be concluded from a passive surveillance system (TiTson, ~ 9921. There is a further clifficulty indirectly implied in the plans for this stucly. The study includes surveys at two time points. It is the committee's impression from this and other aspects of the CDC's program that the CDC expects to undertake educational and other activities in the hope of increasing VAERS reporting, presumably by changing the KAB of health care providers who do most of the reporting. The committee certainly has no objection to education about VAERS, nor to increasing the reporting rate for this or any other procluct. But if there were to be a change in the absolute number of VAERS reports following such educational and other activities, both the change and the number would be uninterpretable in the absence of some reference standard for the rate of . . ... .. . . . .. . . . . ~ , . . . .. .
INTERIM FINDINGS & RECOMMENDA TIONS 33 adverse events themselves. Without some inclepenclent measure of the rate of adverse events, it would be impossible to clecicle whether a change in the number of adverse event reports occurrent as a result of a change in the number of adverse events, or a change in the KAB of health care providers regarding reporting. That would be a serious problem, because the value of VAERS is as a signaling system: a sullen increase in adverse event reports associated with a particular vaccine may indicate a problem with a particular lot of the product, for instance, and thus serves to alert both regulators and industry to investigate. To the extent that such a change might reflect changer! attitudes about reporting rather than a change in the product administered, VAERS could no longer serve its signaling function. The centralized military health care system may offer ways to clevise a solution. For example, the committee also received information regarding the Defense Medical Surveillance System (DMSS), which receives records of medical encounters on both inpatient and outpatient visits for all service members. The DMSS database may thus provide an inclependent source of information on vaccinations and medical encounters and complaints following vaccinations. The committee found that the rationale for investigating the KAB of health care providers toward VAERS was appropriate, but cautioned that since this activity and associated educational interventions could affect VAERS reporting, it should be associated with an independent source of information about adverse events. The committee commented that data from the DMSS, which tracks the medical encounters and other pertinent data on military personnel, offers such an independent comparison at least for moderate and severe adverse events, and noted farther that this is an example of an area of potential collaboration between the CDC and the DoD. Furthermore, the committee found that the survey of military vaccine providers regarding VAERS might be coupled with the segment of the survey regarding the stance of health care providers toward the anthrax vaccine, as mentioned above. Finally, the committee found that previous research on VAERS reporting would justify reconsideration of the sample size, perhaps reducing the number of subjects. . . . The committee recommends that the CDC consider expanding the design phase of the knowledge, attitudes, and beliefs (KAB) study of military vaccine providers regarding the Vaccine A(lverse Event Reporting System (VAERS) reporting to include cognitive and psychometric tests and a pilot survey, in order to design both the educational interventions and the survey that will relate to them and possibly reduce the number of subjects. The committee recommends that the CDC consider including a study of the knowledge, attitudes, and beliefs (KAB) of health care providers regarding the anthrax vaccine in the study now designed to assess only KAB on the Vaccine Adverse Event Reporting System (VAERS) reporting. The committee recommends that the CDC make use of independent sources of information concerning vaccine adverse reactions in the military, such as the Defense Medical Surveillance System (DMSS), when assessing any monitoring of, or modification to, Vaccine A(lverse Event Reporting System (VAERS) reporting practices and VAERS analyses. Data Mining Using the VAERS Database Despite the limitations inherent in a passive surveillance system, the VAERS database is a rich locle of information. and the CDC has designed a data-mining study to exploit that resource.
34 CDC ANTHRAX VA CCINE SAFETY & EFFICA CY RESEARCH PROGRAM Although passively accumulated, the data may well yield information from systematic clusters of symptoms that could be iclentif~ec! through careful and structured analysis of the data available. The CDC has reviewed several software packages especially clesigned for seeking patterns in the data, and was pleased to find the SAS Cluster package, a familiar tool, among the top contenders. Another approach to local structure representation is the detection of self-organizing maps (this effort is not being funded as part of the CDC's anthrax vaccine research, but the results may nevertheless be relevant). As well as local structure, the database may exhibit some global patterns, which could be a basis for building statistical mociels of the factors involved. Various methods for global structure representation exist. The CDC is working mostly with empirical Bayes models. Finally, the CDC is pursuing several hybrid approaches such as using factor analysis anct logistic regression to attempt to reduce the number of variables that could possibly be relevant through the identification of structure underlying adverse events, then developing a logistic mode! to predict from those factors which vaccine had been given. Since VAERS is a passive surveillance system, as has been discussed previously, it is virtually guaranteed to inclucle reports that are duplicative, incomplete, or inaccurate, in addition to the fact that the rate of reporting typically will be low. As a result, data-mining techniques must be employed with care, and even more, any results should be reported so as to reflect the uncertainties implied by the data themselves. That is, the committee is concerned that the evaluation of data mining techniques applied to VAERS data is challenging, and encourages the CDC to explore thoroughly the use of complementary data bases such as DMSS in this context. For example, while VAERS is most useful for detecting trends emerging over time, the data- mininc techniques described to date clo not have a time dimension. The committee has not Yet O r - -a - - -, C7 ~ heard enough about this project, which is still in a very early phase of development, to make many specific recommendations about the CDC's plans. It is important to consider complementary analyses of the VAERS and DMSS databases because of the different characteristics of the two systems. The VAERS is an entirely voluntary system that receives any reports that are submitted depending on the level of interest and concern on the part of the vaccine recipient, the parent, or the health care provider. Its main value lies in its signal- capturing capacity: either by recording unusual events from large numbers of vaccine recipients during the post-marketing phase or by noting obvious changes in the trend of adverse event (AK) incidence that might suggest problems with the vaccine product or particular lot. DMSS is a database that collects records of all medical encounters in the military health care system, as well as anthrax vaccinations. It would therefore include data on visits that would be within some given time after an anthrax vaccination, and the ICD-9 Cole of those visits. For the purpose contemplated here, one of the main values of DMSS is its relative completeness with respect to medical encounters. Another reason the two systems may be complementary is that their different characteristics mean that the reports that are in each database reflect a different set of factors influencing what is to be reported. Certainly there are adverse events of varying severity and frequency that are caused by vaccines in some cases. Also there are surely symptoms suffered by individuals who have recently been vaccinated but which are in fact caused by something other than the vaccine. Decisions either to report to VAERS or to seek medical advice can be influenced by many factors. So both the VAERS ant! the DMSS systems would be expected to capture some, but not all, of the true vaccine-caused adverse events, and both would also be expected to capture some, but not all, symptoms following vaccination that were actually caused by something other than the vaccination. But each wouIc} likely capture a different subset, or a subset influenced by . , . . . .. _ ~q . , , ~
INTERIM FINDINGS & RECOMMENDS TIONS 35 different factors. And both would be expected to capture a greater percentage of the serious AEs than the percentage they capture of the milder AEs. Figure 2-2 represents these relationships; please note, however, that Figure 2-2 is not drawn to any scale; it depicts only logical, not quantitative, relations. Severity 1 .. I- .. #., ~ , a , ~ ~ ~ ~ ~~.~ , ~ FIGURE 2-2 Schematic (not quantitative) illustration of potential relationships be- tween the set of adverse events actually causer! by the product itself (labeled "true"), ant! the subsets captured by DMSS ant! by VAERS. Unshaded space enclosed in the DMSS and VAERS segments represents symptoms that occur following vaccination, but were caused by something other than the vaccination. Meta-Analysis of Existing Studies on the Anthrax Vaccine Since the data and information pertaining to the anthrax vaccine come from different types of studies over the decades, the CDC also intends to carry out a meta-analysis of a wide selection of safety and efficacy studies. The term "meta-analysis" here means a two-stage process of first ~leveloping and carrying out a structured search and systematic review of the available literature, followed by a combination, also structured and systematic, of the findings to produce a single overall estimate of the safety and the efficacy of the vaccine. The search for literature will begin with the safety review of the anthrax vaccine carried out by the DoD's AV]P, and additional Medline searches and consultation of references and specialists. After the first step (the structured collection, review, and scoring of each paper), then the data from each paper must be extracted and converted to some common scale in order to permit combination of all data to obtain overall estimates. As neither populations nor stucly methods may be homogeneous, this step of the meta-analysis involves a number of assumptions, which must be made explicit. The first step of a meta-analysis, namely, a systematic review of the literature, can certainly be undertaken and, if it has not Greatly been done, certainly should be pursued as soon as possible. The next step, however, not only requires multiple assumptions as noted but, in order to be at all useful, requires many studies to be reviewer! for combination. Prior to carrying out the combining of results, it will be important to determine whether the literature even contains enough studies of the safety and efficacy of the anthrax vaccine to make combination necessary or worthwhile, as well as reliable. The committee has not yet heart! enough about this project,
36 CDC ANTHRAX VA CCINE SAFETY & EFFICA CY RESEARCH PROGRAM which is still in a very early phase of development, to make any specific recommendations but will await the assessment of the number of studies, and then further discussion of how or whether to proceed with a meta-analytic combination. In conclusion, the committee recognized great progress in the clevelopment of specific scientific studies over the course of the lOM committee's review to date. The committee looks forward to receiving further information and protocols as the CDC continues to develop its overall anthrax vaccine safety and efficacy research program.