5
Proposed Studies on the Safety of the Anthrax Vaccine

The congressional mandate to the Centers for Disease Control and Prevention (CDC) for research on the anthrax vaccine includes a call for studies to examine risk factors for adverse events, including differences in rates of adverse events between men and women, and studies to optimize the vaccination schedule and administration of the vaccine to minimize the occurrence of adverse events. The materials presented to the committee by CDC describe several studies that have been proposed to respond to these aspects of the congressional mandate. This chapter summarizes the relevant components of each of these studies and presents the committee’s findings and recommendations regarding each study.

The committee notes that the standard regulatory terms for any undesirable effect of a vaccine (or other biologic or drug) are adverse event or adverse reaction.1 Adverse events can range from mild to severe or life-threatening. The standard term used by regulatory agencies to describe the characteristic profile of adverse events associated with a product is the safety of the product. The committee emphasizes that the safety of a vaccine or other product is relative, not absolute. In general, the term safety reflects expectations of relative freedom from, but not necessarily the complete absence of, harmful effects when a product is used prudently, considering the condition of the recipient and the health risk the product is directed against.2

No single set of criteria defines acceptable limits on the frequency and severity of vaccine-related adverse events. Expectations for the safety of vaccines are especially high, however. In contrast to therapeutic agents, which are given when a disease is known to be present (or at least suspected), vaccines are usually given to healthy people to protect them against pathogens that they may or may not be exposed to in the future.

Thus, the committee reviewed the studies proposed by CDC with the expectation that they should be appropriate for producing knowledge that can aid in the evaluation of suitable uses of Anthrax Vaccine Adsorbed (AVA) and of possible new anthrax vaccine formulations.

1  

An adverse event includes any undesirable condition that occurs following vaccination, whether or not it is causally linked to the vaccine. An adverse reaction is an event considered causally related to receipt of the vaccine.

2  

The definition of safety used by FDA is “the relative freedom from harmful effect to persons affected, directly or indirectly, by a product when prudently administered, taking into consideration the character of the product in relation to the condition of the recipient at the time” (21 C.F.R. § 600.3[1999]).



The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement



Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 56
5 Proposed Studies on the Safety of the Anthrax Vaccine The congressional mandate to the Centers for Disease Control and Prevention (CDC) for research on the anthrax vaccine includes a call for studies to examine risk factors for adverse events, including differences in rates of adverse events between men and women, and studies to optimize the vaccination schedule and administration of the vaccine to minimize the occurrence of adverse events. The materials presented to the committee by CDC describe several studies that have been proposed to respond to these aspects of the congressional mandate. This chapter summarizes the relevant components of each of these studies and presents the committee’s findings and recommendations regarding each study. The committee notes that the standard regulatory terms for any undesirable effect of a vaccine (or other biologic or drug) are adverse event or adverse reaction.1 Adverse events can range from mild to severe or life-threatening. The standard term used by regulatory agencies to describe the characteristic profile of adverse events associated with a product is the safety of the product. The committee emphasizes that the safety of a vaccine or other product is relative, not absolute. In general, the term safety reflects expectations of relative freedom from, but not necessarily the complete absence of, harmful effects when a product is used prudently, considering the condition of the recipient and the health risk the product is directed against.2 No single set of criteria defines acceptable limits on the frequency and severity of vaccine-related adverse events. Expectations for the safety of vaccines are especially high, however. In contrast to therapeutic agents, which are given when a disease is known to be present (or at least suspected), vaccines are usually given to healthy people to protect them against pathogens that they may or may not be exposed to in the future. Thus, the committee reviewed the studies proposed by CDC with the expectation that they should be appropriate for producing knowledge that can aid in the evaluation of suitable uses of Anthrax Vaccine Adsorbed (AVA) and of possible new anthrax vaccine formulations. 1   An adverse event includes any undesirable condition that occurs following vaccination, whether or not it is causally linked to the vaccine. An adverse reaction is an event considered causally related to receipt of the vaccine. 2   The definition of safety used by FDA is “the relative freedom from harmful effect to persons affected, directly or indirectly, by a product when prudently administered, taking into consideration the character of the product in relation to the condition of the recipient at the time” (21 C.F.R. § 600.3[1999]).

OCR for page 56
OBJECTIVES AND CRITICAL RESEARCH QUESTIONS FOR CDC RESEARCH ON THE SAFETY OF THE ANTHRAX VACCINE CDC’s stated objectives for the safety component of its anthrax vaccine research program are displayed in Box 5-1. At the request of the committee, CDC also identified a set of critical research questions, shown in Box 5-2. BOX 5-1 CDC Objectives for Research on the Safety of the Anthrax Vaccine To investigate potential long-term sequelae of AVA. To gain a better understanding about the type, frequency, and gender differences of vaccine adverse events associated with AVA. To evaluate the completeness and accuracy of reporting of AVA adverse events in the military and to develop and implement interventions to improve AVA adverse events reporting and surveillance. To assess AVA administration practices and the military immunization health care system that may impact AVA adverse events, and to enhance AVA delivery practices (quality assurance of AVA administration services in the military). To evaluate concerns that military personnel may have about AVA and improve their knowledge and understanding about the risk benefit of AVA and other vaccines. To provide AVA information, education, and communication resources to the civilian public and to military personnel in collaboration with the Department of Defense. SOURCE: CDC, 2002d, p. 11. BOX 5-2 Critical Research Questions Regarding the Safety of the Anthrax Vaccine, as Identified by CDC How does changing the route of administration affect the safety profile of AVA? What are the important risk factors for the development of adverse events to AVA? Is gender an important risk factor for the development of adverse events to AVA? What is the overall safety profile of AVA immunization? Are there any chronic health or long-term problems associated with AVA immunizations? Are there specific syndromes or disorders associated with AVA? Does hormonal phase affect the occurrence of adverse events [in women]? Is AVA safe for children? Is AVA safe for use in the elderly? Is AVA safe for women, with respect to reproductive health? What is the safety profile of AVA when administered postexposure with antibiotics? What is provider knowledge of VAERS and compliance with reporting to VAERS? SOURCE: CDC, 2002e.

OCR for page 56
The committee found CDC’s research objectives and research questions regarding the safety of the anthrax vaccine to be responsive to the request from Congress. The committee notes that while Congress and CDC specifically address differences between men and women in their risks for adverse events following vaccination, other demographic characteristics such as age and race may also be related to the risk for adverse events. In addition, the committee cautions that while the investigation of potential chronic or delayed sequelae of the receipt of AVA is of interest, it is a challenging research task. Identifying particular conditions in the small study population that has been proposed for this purpose and establishing a biologic basis for connecting any conditions that might be identified with the receipt of AVA would be difficult. The research objectives and questions (and the related study proposals) concerning the reporting of adverse events and the acceptability of AVA, which go beyond the specific congressional request, are discussed in Chapter 6. CDC has included among the critical research questions determining the safety of AVA when it is administered to children or older adults, and when it is administered in conjunction with antibiotics following exposure to anthrax spores. However, CDC gave a lower priority to these questions than to the others it listed. The committee feels these are critically important questions, although it agrees that studies in these populations should be delayed long enough to be able to take into account the findings from the human clinical trial on the optimal route and number of vaccine doses for young and middle-aged adults. Planning for future studies in children and the elderly should be flexible enough to respond to changing circumstances, including the possible availability of a newer anthrax vaccine. The committee identified persons with chronic illnesses as another population that should be studied—again, after taking into account the findings on immunogenicity and reactogenicity in healthy adults from the clinical trial and allowing for modifications in response to changing circumstances. The committee’s views on the place of such studies within the CDC research program are discussed in Chapter 7. In the remainder of this chapter, the committee reviews the specific studies that have been proposed and presents its findings and recommendations concerning those studies. ANTHRAX VACCINE ADSORBED: HUMAN REACTOGENICITY AND IMMUNOGENICITY TRIAL TO ADDRESS CHANGE IN ROUTE OF ADMINISTRATION AND DOSE REDUCTION This study, referred to as the human clinical trial, is intended to compare the immunogenicity and reactogenicity of AVA when given under the currently licensed regimen—subcutaneous (SQ) administration of six primary doses of vaccine (at 0, 2, and 4 weeks and 6, 12, and 18 months) and annual booster doses—with the immunogenicity and reactogenicity of the vaccine when a reduced number of doses are given intramuscularly (IM) (see CDC, 2002a,f). The components of the study related to immunogenicity are described in Chapter 4. CDC proposed two hypotheses related to reactogenicity and adverse events: AVA administered by the IM route results in decreased local reactogenicity compared with SQ administration. Occurrence of adverse events following AVA administration is influenced by selected risk factors. Study Design The study is designed as a prospective, randomized, double-blind,3 placebo-controlled clinical trial to be conducted over a period of 43 months at five sites in the United States. As described in Chapter 4, the study population will consist of 1,560 healthy civilian adult men and women between the ages of 18 and 3   Unblinded staff will prepare and administer the vaccine or placebo, but CDC staff, the investigators monitoring and analyzing immunogenicity and reactogenicity, and the participants will remain blinded to study group assignment.

OCR for page 56
61 years. Recruitment efforts will focus on groups for whom AVA vaccination for bioterrorism preparedness has been considered, including emergency first responders, federal responders, and medical practitioners. Study participants will be randomly assigned to one of six study groups of 260 persons each, in a manner that ensures that at least 20 percent of the members of each study group will be women. One study group will receive eight SQ doses of AVA in accordance with the currently licensed schedule. A placebo group will receive eight injections of sterile saline; half of the group will receive SQ injections and half will receive IM injections. In the four other study groups, participants will receive either four, five, seven, or eight IM doses of AVA. Participants who receive fewer than eight doses of AVA will receive an injection of the saline placebo in place of an omitted dose of AVA.4 All vaccine doses are to come from AVA Lot FAV063, manufactured by BioPort Corporation as a “post-renovation qualification lot” (CDC, 2002a, p. 61). Study participants will be actively monitored for adverse events following scheduled injections. Each person will have a total of 22 clinic examinations for assessments of a set of predefined local and systemic adverse events (referred to as solicited adverse events) and of other health-related endpoints (see Box 5-3). In addition, participants will receive 14-day diaries after each of the first two injections and 28-day diaries after each subsequent injection to record adverse events. They will also receive digital thermometers to measure oral temperature at bedtime for the four days following an injection. Clear circular rulers will be given to participants so that they can measure the diameter of any injection-site reactions. The information from the clinical assessments and diaries will be used to determine the presence or absence of the specified adverse events, the presence or absence of any adverse event rated moderate or severe, and the total number of events. Local and systemic events will be considered separately. The number of days of restricted activity each study participant experiences as a result of adverse events will also be assessed. In addition, the study protocol calls for participants to complete the SF-36 v2 Health Survey (described below) at enrollment and at 12, 18, 30, and 42 months. Over the course of the study, each participant will also have a total of 16 blood samples drawn. Data will be collected from study participants for analysis of potential risk factors for adverse events. The risk factors to be considered include (but are not limited to) age, sex, body mass index, hormonal status (women only), known allergies, physical activity level, smoking status, perceived general health status, number of previous doses of AVA, pre-injection titers of anti-protective antigen immunoglobulin G and toxin neutralizing antibody, history of adverse events associated with previous doses of AVA or with doses of other vaccines, and study participants’ beliefs at the conclusion of the study as to whether they had received doses of the vaccine or the placebo. The analysis will compare men and women in terms of the reactogenicity of the vaccine and the influence of various risk factors on the occurrence of adverse events. Planned Analyses The study protocol specifies that the primary reactogenicity analyses will be conducted using the participants who can be evaluated according to protocol and that an intent-to-treat analysis will be used to assess whether deviations from the protocol biased the results.5 CDC will contract with statistical experts to examine deviations from the protocol in the form of dropouts, noncompliance, and loss to follow-up and to devise appropriate analyses. The proposed analysis will test for less reactogenicity with IM administration than with SQ administration in the two study groups receiving the full eight doses of AVA. In addition to the overall IM–SQ comparison, IM–SQ comparisons will also be performed separately for men and women. Per-dose and repeated-measures analyses will be performed, and a one-sided test of significance (alpha = 0.025) will be 4   See Table 4.1 in Chapter 4 for the schedule of injections for each study group. 5   See Chapter 4 for an explanation of the differences between according-to-protocol and intent-to-treat analyses.

OCR for page 56
BOX 5-3 Categories of Adverse Events to Be Identified During the Human Clinical Trial Solicited Adverse Events Within 28 days of vaccination: fever, fatigue, muscle ache, headache, temperature, axillary adenopathy, warmth, tenderness, itching, pain, arm motion limitation, erythema, induration, nodule, bruise Serious Adverse Events Any time during the study: death, life-threatening adverse event, initial inpatient hospitalization or prolongation of hospitalization (including pregnancy), significant or persistent disability/incapacity, congenital anomaly or birth defect, or any important medical event based upon appropriate medical judgment that may jeopardize the participant and may require medical or surgical intervention to prevent one of the other outcomes defined as a serious adverse event. Other Adverse Events Any other adverse event that cannot be classified as a solicited adverse event. SOURCE: CDC, 2002a, pp. 79–80. used. The occurrence of local and systemic effects will be considered separately, based on a dichotomous measurement of the presence or absence of specified adverse events as detected during clinical examinations or recorded in participants’ diaries. The total number of days of restricted activity per participant will also be compared. Separate analyses will compare all study groups to assess the effect on reactogenicity of a reduction in the number of vaccine doses along with the change in route of administration. Exploratory analyses will test for differences in the severity of adverse events and in perceived general health and well-being (based on responses to the SF-36). Other exploratory analyses will examine the association of various risk factors with the occurrence of adverse events. To test for significant differences between men and women, comparisons will be made within each of the six study groups (two-sided test, alpha = 0.05). Another exploratory analysis will test whether premenopausal women differ in the occurrence of adverse events depending on their hormonal phase (follicular or luteal) at the time of vaccination. One comparison will be restricted to women receiving SQ doses of AVA. The other comparison will include all eligible women. The study protocol notes that these analyses will focus primarily on women who are not using pharmacologic methods of birth control and that special efforts may be necessary to recruit adequate numbers of women meeting this requirement. A Data and Safety Monitoring Board will review quarterly progress reports on the study and will assist in the preparation of the interim analysis of the data from the first 7 months of the study. This board will also monitor the occurrence of any serious adverse events or procedural problems that might warrant a recommendation to terminate the study. Committee Comments On the whole, the committee found that the study, as described in the protocol, is generally appropriate for comparing the reactogenicity of SQ and IM administration of AVA. The basic analyses related to the association between the route of vaccine administration and the occurrence of adverse events should have, and do seem to have, the highest priority. The plans for monitoring adverse events should produce the most systematic assessment to date of those events that occur with relatively high frequency within the 42-month time frame of the study. It will, of course, be unable to provide insights regarding less frequent, albeit serious, events.

OCR for page 56
In addition, the use of vaccine from a lot manufactured following the completion of the renovation of the BioPort facilities will provide an opportunity for systematic documentation of the reactogenicity of the newly manufactured vaccine. The availability of such data will help guide the routine monitoring of the vaccine called for by the IOM review of the efficacy and safety of AVA for the Department of Defense (DoD) (IOM, 2002). The committee also commends CDC for assembling a strong and well-balanced Data and Safety Monitoring Board, whose members have substantial experience with studies of the safety and efficacy of vaccines. Finding: As described in the study protocol, the human clinical trial is generally responsive to the congressional mandate to evaluate the incidence of, risk factors for, and differences between men and women in local and systemic immediate-onset health effects associated with AVA and the effect of the route of vaccine administration on adverse events. The study will also provide a 42-month follow-up period during which to monitor the occurrence of later-onset health effects.6 The committee noted one area of concern related to the proposed statistical analyses of reactogenicity. The study protocol specifies that the primary analysis will test the hypothesis that IM administration of AVA results in reduced reactogenicity compared with SQ administration and that a one-sided statistical test of significance will be used. It appears that this approach was chosen to parallel the immunogenicity analysis to establish the “non-inferiority” of antibody response to IM administration compared with the response to SQ administration. However, it may not be justified for the analysis of differences in reactogenicity. To justify a change to IM administration of AVA, the immunogenicity analysis must show that IM administration of the vaccine results in an antibody response considered at least as good as the response with SQ administration. If IM administration does not perform as well as SQ administration, there will be no basis for changing the current practice of SQ administration and little reason to establish whether the antibody response for IM administration is significantly worse than that for SQ administration. Therefore, one-tailed statistical testing is appropriate. For reactogenicity, however, the hypothesis being tested is that IM administration of the vaccine is superior to SQ administration (i.e., less reactogenic) rather than not inferior. The proposed use of a onesided test of statistical significance does not allow for the detection of inferior performance, that is, whether any adverse events occur at significantly higher rates with IM administration of AVA than with SQ administration of the vaccine. Assuming that the study can provide satisfactory indications that the antibody response with IM administration of the vaccine is at least comparable to that with SQ administration, the decision as to whether to modify the route of administration should also take into account the likely impact on the frequency and severity of recognized adverse events. If IM administration of the vaccine proves to be significantly less reactogenic in terms of all the proposed indicators, use of a one-sided statistical test is sufficiently informative, and the decision to adopt IM administration is straightforward. However, if some indicators were to show that reactogenicity is not reduced with IM administration, it would be important to establish whether IM administration resulted in increased reactogenicity over SQ administration before deciding to change the licensed indications for use of AVA. A two-tailed statistical test is necessary to differentiate both decreased and increased reactogenicity from no difference. The plans for using one- or two-tailed statistical tests in the analyses of risk factors for reactogenicity are unclear. For these analyses, the committee sees no basis for one-sided tests unless a clearly articulated biological rationale can be provided. 6   The committee has adopted the terminology used by the Committee to Assess the Safety and Efficacy of the Anthrax Vaccine (IOM, 2002). The phrases “short term” and “long term” were not used to characterize adverse events because of the potential for confusion. Instead, the duration of an adverse event is characterized as acute or chronic; the timing of the onset of an adverse event is characterized as immediate or later.

OCR for page 56
Recommendation: The analyses of reactogenicity in CDC’s human clinical trial of AVA should use two-sided statistical tests. The likely loss of participants over the course of the study should also be considered. The study protocol notes that sample sizes were calculated to allow for the loss of up to 50 percent of participants. As noted in Chapter 4, the committee urges that CDC and the centers participating in the study ensure that those interested in participating are fully informed about the demands of the study, including the vaccination schedule and the time commitment involved. Also as noted in Chapter 4, the committee urges continued consultation with experts in the analysis of data from clinical trials with significant loss to followup or noncompliance, as well as consultation with FDA, to ensure appropriate analysis of the data from the clinical trial. The most important analyses are those necessary to establish the route of vaccine administration and the number of vaccine doses that produce immunogenicity at a level at least comparable to that of the current regimen while minimizing reactogenicity. The study appears adequately powered for these analyses. However, the committee is concerned that the size of the study population will not provide adequate statistical power for some proposed analyses. As noted in the committee’s interim report (IOM, 2001), it may not be possible to analyze risk factors for adverse events among men and women separately. Similarly, analyses of other demographic factors that are of interest, such as age and race, may be hampered by lack of statistical power. It may also prove challenging to perform the planned analysis of the possible association of hormonal phase with the occurrence of adverse events because of the need to recruit adequate numbers of premenopausal women who are not using pharmacologic methods of birth control. The committee also observes that the expectation reflected in the protocol that the inclusion of placebo groups will permit the evaluation of rates of rare adverse events is unrealistic because of the difficulty of detecting truly rare events (e.g., incidence less than 100 per 100,000) in a study population of this size. Other approaches, such as the analysis of data from large observational data sets like those of the Defense Medical Surveillance System (DMSS), appear far better for studying the incidence of rare adverse events. A final observation concerns the planned use of the SF-36 to assess whether vaccination is associated with differences in perceived general health and well-being. As in its interim report, the committee endorses the general goal of assessing the health status of vaccinees. It is essential, however, to recognize the limitations of the SF-36 for this purpose in the context of the human clinical trial. The SF-36 is a well-studied and extensively validated instrument for the self-assessment of health status, principally for producing a generic measure of the burden of disease (Ware et al., 1993). The SF-36 produces scores on eight broad aspects of physical and emotional functioning and well-being7 and two summary scores on physical and mental health. But because the SF-36 is designed primarily to detect substantial differences in health status as a result of disease or injury, it is unlikely to be sensitive enough by itself to make meaningful distinctions among small changes in a generally healthy population, such as the participants in the human clinical trial. The committee encourages CDC to supplement the SF-36 with other instruments, such as a symptom checklist or other validated assessment tools specifically related to possible adverse events or to specific complaints (e.g., fatigue, cognitive impairment, or reduced productivity). The additional information would facilitate interpretation of the SF-36 results through comparisons of persons with and without certain symptoms or would permit direct assessments of the impact of specific adverse events. The committee’s reservations about the proposed use of the SF-36 are also discussed later in the chapter in conjunction with the review of the studies to be based in the Vaccine Healthcare Center (VHC) Network. 7   The eight health concepts measured by the SF-36 are limitations in physical functioning, limitations in usual role (e.g., worker, student, etc.) because of physical health problems, bodily pain, general health, vitality, limitations in social functioning because of physical or emotional problems, limitations in usual role because of emotional problems, and mental health.

OCR for page 56
FOLLOW-UP STUDY OF TEXTILE MILL WORKERS VACCINATED AGAINST ANTHRAX CDC has proposed to study the possible chronic or later-onset health effects of AVA vaccination by examining the mortality experience and functional status of textile mill workers who received doses of AVA 10 or more years ago. A draft study protocol was submitted to the committee for review (see CDC, 2002b,g). Study Design A retrospective cohort study has been proposed. The study population will be drawn from former workers at one textile mill that processed goat hair from the mid-1960s through the mid-1990s. The mill required vaccination against anthrax for its entire workforce, which averaged 800 to 1,000 workers at any given time. Immunization records were maintained by a company-employed nurse but had not been located at the time the study proposal was provided to the committee. CDC is working with the Social Security Administration (SSA) to identify former employees of the mill. As of January 2002, CDC had received information on 3,172 persons employed at the mill or by its parent company during the period 1978–1996; of this group 337 may have worked at another site, making them ineligible for the study. Deceased workers will be identified using the National Death Index. CDC will obtain contact information for surviving workers through a process developed for and successfully used in studies by the National Institute of Occupational Safety and Health (NIOSH). CDC estimates that 2,605 of the 3,172 workers identified from the SSA records are still alive. Assumptions that 15 percent of the survivors will be lost to follow-up and that 70 percent of those located will participate suggest a vaccinated study cohort of 1,550 persons. Preliminary calculations indicate that for outcomes with a prevalence of 1.0 percent this sample size will provide a power of 97 percent to detect a fivefold increase in risk and power of 57 percent to detect a twofold increase in risk. For outcomes with a prevalence of 0.5 percent, the power to detect a fivefold or a twofold increase in risk is estimated to be 74 percent and 28 percent, respectively. Two comparison groups of unvaccinated persons are planned. One group will be drawn from the community in which the goat hair mill was located. Participants will be recruited from the census tracts in which vaccinated workers now live and frequency-matched in terms of age, sex, and race. An occupational comparison group will consist of persons who worked in other kinds of textile processing mills in the same region of the country as the goat hair mill and during the same time period as the members of the vaccinated study group. Participants will be selected by frequency matching on age, sex, and race. Information will be collected on participants’ demographic and socioeconomic characteristics and on various health-related risk factors, including occupational history, personal and family medical history, use of medical care, and history of disability claims. The questionnaires and interview forms used to collect this information will, to the extent possible, be based on various national survey instruments (e.g., National Health Interview Survey, the Longitudinal Survey of Aging, and the Current Population Survey). Health outcomes will be assessed using measures that can provide objective evidence of pathology that could be examined further in subsequent studies. For this initial study, the measures, which are not necessarily linked to specific clinical endpoints, include excess mortality (overall and cause-specific), excess morbidity, and measures of current functioning. For the goat hair mill workers who have died, CDC will ascertain their dates of death and underlying and contributory causes of death (coded according to the ICD-9). CDC will also obtain the death certificates for these workers. Medical conditions for which data from the self-reported medical histories of the surviving goat hair mill workers show a statistically significant excess compared with the control groups will be verified by a review of participants’ medical records. The assessment of current functioning will include the following four components, measured using the specified tools:

OCR for page 56
Health status and health-related quality of life (HRQoL), measured using the SF-36 and the Health Utilities Index (versions 2 and 3) Energy or activity level (no tools specified) Cognitive function, measured using the Halstead-Reitan Neuropsychological Test Battery (selected subtests), the Wechsler Memory Scale (Third Edition), and continuous performance tests and measures of attention Immunologic function, measured using a complete blood count with differential, lymphocyte subsets, immunoglobulin levels, complement levels, T-cell proliferation assays, skin tests for anergy, and thyroid hormone levels. The criteria considered by CDC in selecting these measurement tools included the availability of population-based norms, a demonstrated usefulness in clinical and research studies, and logistically feasible and tolerable requirements for administration. To the extent possible, all participants will complete all assessments. An extramural advisory panel will be convened before the study begins to advise CDC on the study design, the selection of tests of current functioning, and criteria for the interpretation of test results. The same panel (or a similar one) will reconvene periodically to provide advice during the collection and analysis of data. Planned Analyses The analysis of the mortality experience of the vaccinated population will be based on the calculation of standardized mortality ratios (SMRs) for overall and cause-specific mortality. The SMRs will compare observed mortality with that expected based on death rates for the United States as a whole, for the state in which the goat hair mill was located, and for the occupational comparison group. Computer programs developed by NIOSH for life table analysis and Poisson regression analyses will be used to account for person-years at risk. The study protocol notes that if there is an indication of excess deaths among the vaccinated workers, it will be necessary to allow for possible survivor bias in the analysis and interpretation of other data. The assessment of health-related life experiences and current functioning among vaccinated workers and the comparison groups will make use of descriptive statistics (e.g., means, frequency distributions) to characterize study participants. Various parametric and nonparametric analytical techniques will be used to compare health outcomes between the vaccinated population and the unvaccinated comparison groups. Risk associated with exposure to AVA will be measured using multivariate techniques (e.g., linear or logistic regression, Cox proportional hazards models). Matching criteria will be included as covariates in all statistical models. Committee Comments This study is intended to address the questions of whether receipt of AVA is associated with chronic health impairments or with adverse health effects that become apparent only after several years (later-onset health effects). The committee agrees that these questions deserve attention. Vaccinated workers in textile mills that processed goat hair are clearly an appropriate population to consider studying. They represent one of the primary target populations for vaccination during the 1960s, 1970s, and 1980s. Some of the textile mill workers from the proposed study population may have begun receiving AVA on a regular basis from the time it became available in the 1960s, offering the possibility of as much as 30 years of follow-up. In addition, the plans to include two comparison groups—a community group and an occupational group—reflect an awareness of some of the challenges of studying occupational health risks. Nevertheless, the committee has serious reservations about the proposed study. The most important concern is the risk of a selection bias: there is no truly comparable control group available. The use of an

OCR for page 56
occupational comparison group is an appropriate effort to compensate for the “healthy worker effect” that is often found in comparisons between working populations and community controls. But it may be impossible to establish the true comparability of the workplace exposures in the textile mills that processed goat hair with those in other types of textile mills in the region. As such, any finding that emerges, whether positive or negative, has too high a risk of being spurious, due to making comparisons between groups that are not equivalent in ways other than their exposure to AVA. Another important fact is that the study is too small and therefore lacks sufficient statistical power for a meaningful assessment of differences in mortality and disease risks and in other health outcomes between the vaccinated population and the comparison groups. The estimates of statistical power in the study proposal use 0.5 percent (500 cases per 100,000) as a minimum estimate of the prevalence of a hypothetical outcome of interest. The committee notes that cohort studies usually rely on measures of incidence rather than prevalence. In basing its calculations on prevalence, CDC may have underestimated the difficulty of detecting events of interest in a cohort of the size anticipated for this study. Regardless, the events of interest regarding AVA are much less common than the rates evaluated in CDC’s sample size calculations. An analysis of DMSS data, for example, found that postvaccination hospitalization rates among military personnel who received AVA were 92.5 per 100,000 for all neoplasms (ICD-9-CM codes 140–239); 32.5 per 100,000 for all endocrine, nutritional, metabolic, and immunity disorders (ICD-9-CM codes 240–279), and 67.2 per 100,000 for all diseases of the nervous system (ICD-9-CM codes 320–389) (AMSA, 2001). Within these summary categories, hospitalizations for individual diseases—which are more plausibly associated with vaccination—were much rarer. While these rates are based on a shorter period of observation than would be the case in the proposed study, they illustrate the incidence of medically significant events observed in a large population of vaccinees. Furthermore, the committee questions the validity of the assumptions used to project the likely size of the study population. The expectation that only 15 percent of the survivors will be lost to follow-up and that 70 percent of those contacted will participate seems overly optimistic. Greater loss to follow-up or lower participation rates will only increase the probability of a spurious association (or a spurious lack of an association) as it is quite likely that the few outcomes of interest would be missed. Such losses will also exacerbate the problem of low statistical power. In addition, the study also poses the statistical risk of generating spurious associations, either positive or negative, between vaccination with AVA and various health outcome measures because of the large number of proposed outcome measures and analyses described in the study protocol. It is reasonable to expect some statistically significant associations to arise by chance alone. Unidentified confounding factors might also serve to obscure true differences in health outcomes between the vaccinated and unvaccinated populations. The study might produce incorrect positive findings due to this uncontrolled confounding or selection bias, and it could miss true positive findings, even in the unlikely event that the signal was strong enough to be detected in a study this small. Another serious concern is the potential for misclassification bias. The validity of the analysis would be weakened if workers who were assumed to be vaccinated were not or if the number of doses they had received was not accurately recorded. The study protocol noted that the immunization records for workers at the goat hair mill had not yet been located. Once again, this bias could result in a masking of true findings. Given the limitations of the proposed study, the committee is concerned that participants face inappropriate risks. The study process will be intrusive, and asking participants about their health status and medical histories could be construed as meaning that health problems are anticipated. In addition, the false signals that random error or bias in the data may generate could result in unnecessary anxiety or medical tests and interventions. Instead, the committee urges consideration of alternative approaches for investigating whether receipt of AVA is associated with chronic or later-onset health effects. In particular, the committee encourages the development of plans that will permit follow-up of military personnel who received the vaccine. This

OCR for page 56
might be done through DMSS or the Millennium Cohort Study, which will follow up to 140,000 military personnel during and after their military service. The use of data from DMSS and other sources is discussed in more detail later in this chapter in conjunction with other study proposals. Finding: The committee concludes that the preliminary exploration of a study of possible chronic or later-onset adverse events related to anthrax vaccination among goat-hair textile mill workers, with community and occupational comparison cohorts, was appropriate. That effort, however, has produced sufficient information to indicate that the study (1) poses the risk of generating spurious associations or masking real associations, in part because of the difficulty of identifying suitable comparison groups, and (2) would not have sufficient statistical power to detect conditions of interest. Furthermore, with these limitations, conducting the study poses the risk of generating unwarranted health concerns among the participants. Recommendation: CDC should not continue work on the proposed follow-up study of textile mill workers who received AVA. STUDIES BASED IN THE VACCINE HEALTHCARE CENTER NETWORK The committee reviewed draft protocols for three studies to be conducted through the VHC Network (CDC, 2002c,h,i). The VHC Network is a collaboration between DoD and CDC to address issues of safety and acceptability of all types of vaccines administered within the military health care system (CDC, 2002h). The first VHC was established at Walter Reed Army Medical Center in Washington, D.C., in September 2001. Plans call for a total of 10 to 12 VHCs to be opened over the next 5 years. The goals for the network are to serve as a platform for studies of vaccine-related adverse health events and to enhance the immunization-related health care of military personnel. DoD is to focus on the clinical management of vaccination services and the care and follow-up of service personnel who experience vaccine-associated adverse events. CDC is to focus on observational research on vaccine safety, pilot tests of vaccine safety and acceptability activities and interventions, and assessments of the impact of the VHC Network on vaccine safety and acceptability. Concerns related to AVA will be the initial focus of these activities, but the VHC Network is expected to address issues related to other vaccines, as well. The three study proposals reviewed by the committee use an observational study design in a military population to replicate certain components of the human clinical trial (discussed above). Specifically, these studies are to examine (1) the effects of the route of AVA administration on adverse events that occur soon after vaccination, (2) the effect of AVA on HRQoL, and (3) the effect of hormonal phase on the occurrence of adverse events in women receiving AVA. Each study proposal is discussed in more detail below. The proposal notes that these studies will complement the human clinical trial by overcoming some of its limitations, in particular, the trial’s low statistical power to test some risk-factor associations and the need to wait until the completion of the study (43 months) to perform some of the analyses. The drafts of the study proposals reviewed by the committee specify that participants are to be recruited from random samples of military personnel (active duty and reserves) scheduled to receive AVA vaccinations. Vaccination of personnel scheduled for deployment to high-risk areas has the highest priority. Study subjects will be permitted to participate in more than one of the proposed studies, but the initiation of the studies and the continued recruitment of subjects will depend on the pace at which service members are vaccinated under DoD’s new anthrax vaccination policy, announced in June 2002 (Wolfowitz, 2002). In addition, two of the proposed studies plan for cohorts that receive IM doses of AVA or that receive a new recombinant anthrax vaccine. The proposal notes that the inclusion of these cohorts will depend on the Food and Drug Administration’s (FDA) licensure of AVA for IM administration and of a new anthrax vaccine product when it becomes available. Study participants will not necessarily be vaccinated at a VHC location, and plans are being developed for VHC oversight of AVA vaccinations administered at other sites. Military medical staff will be

OCR for page 56
Committee Comments Given the concerns that have been expressed by some about the adverse impact on overall health and well-being of vaccination with AVA, the committee agrees in general with the concept of a study to assess differences in health-related quality of life associated with receipt of AVA. There are, however, several concerns about the specific study that has been proposed. First, as noted in the comments on the related component of the human clinical trial, the SF-36 generally performs poorly in detecting meaningful differences in the health status of generally healthy populations. Because the proposed study participants will be scheduled for military deployment, they can be expected to be among the healthiest members of the military population. Second, the fact that the study participants are to be recruited from among military personnel scheduled for deployment raises a concern about the feasibility of plans to readminister the SF-36 at 6-month intervals over a 2-year period. The study proposal does not address the procedures that will be used to locate study participants or to administer the SF-36 once they are deployed. Third, the committee is concerned that it will be impossible to validly associate any observed differences in HRQoL with receipt of AVA because the SF-36 results will be confounded by the study partic ipants’ exposure to other vaccines and medications administered prophylactically in preparation for deployment, as well as by the potentially substantial effects of a variety of factors resulting from deployment itself. Furthermore, the effects of deployment might vary depending on the deployment duty station (e.g., South Korea versus Southwest Asia). Even with a control group that is deployed but has not received AVA, an adverse impact on health status specifically associated with receipt of AVA would have to be substantial and widespread for differences between groups or changes over time to be detectable and distinguishable from the effect of other confounding factors. On the basis of these concerns, the committee concluded that the proposed study is unlikely to accomplish its intended purpose. Finding: The SF-36 is designed to detect large changes in health status. It is not suitable for distinguishing differences in health-related quality of life among basically healthy people such as the military personnel who will receive AVA. Furthermore, in the proposed study population, the confounding effects of exposure to other vaccines and particularly of the experience of deployment are likely to make it difficult to discern any unique effect associated with the receipt of AVA. Recommendation: CDC should not conduct the proposed VHC-based study of the effect of AVA vaccination on health-related quality of life. Effect of Hormonal Phase in the Female Population on the Occurrence of Adverse Events Following Immunization with AVA Menstrual cycle phases and levels of progesterone are considered possible risk factors for adverse events in women following vaccination with AVA. CDC reports in the study proposal that oral contraceptives are prescribed for approximately 80 percent of women in the military, and the effect of these contraceptives on hormone levels must also be taken into account. The study is intended to test the following hypotheses: The occurrence of local adverse events in the 3-day period following vaccination with AVA for women in the luteal phase of the menstrual cycle differs from the occurrence of local adverse events for women in the follicular phase of the cycle. Common pharmacologic birth control methods have an indirect effect on the occurrence of adverse events in women following vaccination with AVA.

OCR for page 56
Study Design CDC proposes a prospective observational study. The study population will be recruited from a random sample of premenopausal military women scheduled to receive their first dose of AVA. The sample size will be determined once a sampling frame is provided by DoD. Participants will be asked to provide a blood sample (1 ml of sera) and a pregnancy test before each of their first three doses of AVA. The blood samples will be analyzed at a commercial laboratory to determine the serum progesterone concentrations. Study participants’ hormonal phase (luteal or follicular) will be determined on the basis of progesterone levels. If a participant’s progesterone level is indeterminate, the starting date of her last menstrual period will be used to help classify the participant’s hormonal phase. Study participants will be given a diary at the time of each vaccination and asked to record all adverse events occurring during the following 3 days. The diaries will also be used to record the use of medications, including prescription contraceptives. Demographic information, such as age and body mass index, will be obtained for each participant. The draft study protocol notes that for this study, it will be necessary to ensure that a trained phlebotomist is available at the sites where AVA vaccinations will be given. Planned Analyses Analyses for this study will first be conducted for women who are not using pharmacologic methods of birth control. They will be repeated for all women in the study, with contraceptive use included as a binary covariate. The analysis of the effect of hormonal phase on the occurrence of adverse events will use a dichotomous dependent variable, defined as the presence or absence of individual local adverse events. A repeated-measures analysis will be used, with the logit model and a binomial distribution. Additional covariates will be evaluated for inclusion in the model. Progesterone concentration will be analyzed as a continuous dependent variable using a mixed model methodology and repeated measures procedures. Several covariance structures will be evaluated. Committee Comments As with the other VHC-based studies, the committee agrees that it is reasonable to plan an observational cohort study as a complement to the related analysis of the association of hormonal status with adverse events that is included in the human clinical trial. However, the committee received limited information on the plans for this study and is concerned that the complexity of the subject will pose serious challenges because of the potential difficulty in identifying adequate numbers of suitable participants and the large number of variables that should be included in the analysis. In determining the number of participants for each cohort, adequate allowance must be made for the apparently widespread use of prescription contraceptives by women in the military. These contraceptives will affect the physiologic indicators of hormonal phase, and different contraceptives can be expected to have different effects on those indicators. Therefore, accurately determining whether study participants are using prescription contraceptives and which contraceptives they are using will be essential for valid analysis. In addition, it will be important for the analysis to take into account various other factors that could confound the relationship between hormonal phase and adverse events. Some of these other factors include age, race, and parity. The draft protocol indicated that the size of the study population had not yet been determined. The committee concluded that the sample sizes needed to study these issues will probably be quite large. Finding: The VHC-based study of the effect of women’s hormonal phase on the occurrence of adverse events following receipt of AVA would address a complex subject with many potentially confounding factors (e.g., age, race, parity).

OCR for page 56
Recommendation: As currently described, the VHC-based study of the relationship between women’s hormonal phase and the occurrence of adverse events following receipt of AVA should have a low priority in the CDC research program. Guidance for VHC-Based Research Activities Given the complexities evident in the three draft proposals for VHC-based research studies, the committee is persuaded that regular consultation with a standing panel of outside scientific experts will be important for the success of VHC-based research activities related not only to the anthrax vaccine but also to any other vaccine. This group should be able to advise on matters ranging from study design to data analysis. In particular, biostatistical expertise in propensity analysis will be important because the factors that affect the likelihood of vaccination may also be related to health status, as they have been with AVA. Other areas of expertise that may be valuable include health care outcomes assessment, pharmacoepidemiology for guidance on postmarketing surveillance, and possibly clinical epidemiology of medically unexplained symptoms. The committee is aware that CDC described plans for an advisory panel that will include members from DoD and the military services who can provide valuable advice on the feasibility and logistics of proposed studies. However, it was unclear from the information available to the committee whether the advisory panel was to include members from academia and the private sector and whether it would be a standing group consulted on a regular basis for all VHC research activities. Recommendation: An external scientific advisory group should be constituted to provide guidance to CDC and DoD on all research undertaken through the VHC network. Given the draft study proposals reviewed by the committee, the advisory group should include, among others, experts in biostatistics (propensity analysis), health care outcomes assessment, pharmacoepidemiology (postmarketing surveillance), and clinical epidemiology (medically unexplained symptoms). ENHANCED SIGNAL DETECTION AND HYPOTHESIS TESTING FOR ADVERSE EVENTS FOLLOWING ANTHRAX VACCINATION CDC provided the committee with information on plans to use data from the Vaccine Adverse Event Reporting System (VAERS) and DMSS to investigate adverse events that might be associated with receipt of AVA (CDC, 2002j,k). These investigations will include efforts to detect signals suggesting a possible association between receipt of AVA and an adverse event and efforts to test whether that association might be causal. CDC specified the following objectives: Enhance the capacity to identify adverse events signals from VAERS, the Vaccine Safety Datalink (VSD), and DMSS. Evaluate the association of adverse health events with anthrax vaccine using VAERS and DMSS data. The committee refers to activities related to the first objective as hypothesis generation, and those related to the second objective as hypothesis testing. VAERS is the nation’s principal system for collecting spontaneous reports of adverse events following the use of any vaccine licensed in the United States. It is jointly administered by CDC and FDA. Anyone can submit a report to VAERS, including vaccine recipients or their family members, and more than one report can be submitted about the same adverse event. Reporting is encouraged for any clinically significant event following vaccination, and health care providers are required to report certain specified events (VAERS, 2001). Limitations of VAERS include duplicate reporting and underreporting of an unknown extent that can vary over time and among various kinds of adverse events, as well as incomplete

OCR for page 56
and sometimes inaccurate information in submitted reports. In addition, VAERS lacks data on event rates in the unvaccinated population and on the number of vaccine doses administered. DMSS is a system of DoD-wide databases of health-related information for military personnel on active duty. Data are submitted by the individual armed services, and the system is coordinated by the Army Medical Surveillance Activity (AMSA). The databases include records on inpatient care in military medical facilities since 1990 and records on ambulatory care in military facilities since 1996. Records are also available for all military services on AVA immunizations administered since 1998. Because DMSS captures only events that require ambulatory or inpatient medical care or result in the loss of time from duty, it lacks information on mild adverse events. It also lacks information on care received by military personnel from civilian hospitals or physicians and care that they receive once they leave active duty. AMSA routinely screens DMSS data for signals of adverse events following receipt of AVA but has lacked the resources to conduct studies to investigate associations that may be identified. To be able to conduct such analyses of DMSS data, CDC is establishing a formal agreement with AMSA for the creation of an Analytic Unit that will be based at AMSA. FDA and the Anthrax Vaccine Immunization Program Agency in DoD will collaborate with CDC and AMSA. Plans call for the Analytic Unit to be operational by August 1, 2002. The initial 3-year agreement is renewable. As presented to the committee, the Memorandum of Understanding establishing the collaboration between CDC and AMSA specifies that CDC will cover costs of up to $500,000 for the first fiscal year and up to $225,000 in each subsequent year. Hypothesis Generation Planned Activities The hypothesis-generating activities described by CDC are to focus on the application of automated exploratory statistical tools and processes, including those referred to as data mining, to the VAERS and DMSS databases to identify groups of adverse events that might be associated with receipt of AVA. Data mining is described as a class of techniques that allow rapid extraction of information from large data sets with many variables. CDC proposed the following specific approaches: Bayesian analysis of VAERS reports Techniques and software tools that can be used to identify and rank associations among multiple vaccines or multiple adverse events are being developed by AT&T under contract to FDA and CDC. The initial product was a methodology referred to as the multi-item gamma-Poisson shrinkage estimator (MGPS). Software that is easier to use is being developed, and the analytical procedures are being evaluated through sensitivity and specificity estimation. Bayesian analysis of DMSS reports The methodology developed for use with VAERS data is being adapted for use with DMSS data. Changes in the analytic procedures are necessary to account for the differences between a spontaneous reporting system (VAERS) and longitudinal administrative databases (DMSS). Association analysis of VAERS reports This approach, also referred to as market basket analysis, is used with VAERS data to search for combinations of one or more vaccines with one or more adverse events. Demographic characteristics can also be incorporated in the analysis. The results are expressed in terms of association rules, and a measure of the strength of the association is produced. The analysis requires relatively little computing time and can be repeated at regular intervals (e.g., quarterly, yearly) to assess changes. Association analysis of DMSS reports As with VAERS data, the analysis of DMSS data is expected to identify associations within a large, complex data set. The nature of the DMSS data will allow for consideration of longitudinal information. A hypothetical association rule might be “anthrax vaccination is associated with being a serviceman aged 18–24 years.” CDC notes that the challenge in using this technique is interpreting the association rules and identifying those that merit further investigation.

OCR for page 56
Factor analysis and clustering of VAERS reports Factor analysis is another statistical technique that CDC is assessing for use in identifying clusters of adverse events that follow receipt of AVA or other vaccines. The occurrence of a cluster of adverse events reported by military personnel who received AVA can be compared with the occurrence of the same cluster of events in military personnel who received other vaccines. Measures of association can be obtained with chi-square or Fisher’s exact tests. To decide whether the hypotheses of possible associations between receipt of anthrax vaccine and certain adverse events generated by these analyses should receive further investigation, a systematic literature review will be conducted to determine whether the association has been reported and to evaluate the biologic plausibility of the association. Committee Comments The committee agrees that both VAERS and DMSS are valuable and essential resources for generating hypotheses regarding the occurrence of adverse events following vaccination, and that they should be routinely monitored for signals of adverse events related to use of AVA and all other vaccines administered to the military and civilian populations. The committee was pleased to see greater attention being given to DMSS as a tool for hypothesis generation. Because the medical care records in DMSS databases are collected in a more systematic and complete fashion than are reports to VAERS, the DMSS databases may be better suited to the proposed application of data mining and other techniques of statistical analysis. Some of these analytic tools are relatively new, and their validity when applied to DMSS data must still be carefully tested. But the analyses proposed by CDC may provide a valuable opportunity to learn more about data mining and about the analytic uses of DMSS data. One approach to testing data mining with DMSS data might be to conduct retrospective analyses to determine whether known associations between exposures and health outcomes can be detected. Such analyses would have to involve exposures that are systematically documented in a DMSS data set. Opportunities to study vaccine-related adverse events are currently limited because DMSS has complete vaccination data only for AVA. But the Air Force has recently completed work to enter into DMSS all immunization data from medical records for airmen on active duty. These data extend back as far as is recorded in individual medical records. Plans also call for eventually adding immunization records from the other services (Personal communication, J. Brundage, Defense Medical Surveillance System, July 11, 2002). However, the committee has serious reservations about the proposed plans for various screening analyses of the VAERS data set using data mining and other statistical techniques. As the committee noted in its interim report, statistical analysis of VAERS data is challenging. As a spontaneous reporting system, VAERS is inherently incomplete and subject to reporting biases that are difficult to assess. The data may also be duplicative or inaccurate. The committee is concerned that statistical analysis of data from such systems is of questionable validity. CDC notes that other efforts are planned to increase the completeness and accuracy of VAERS reporting, especially as related to AVA vaccination, but the emphasis on increased reporting will not overcome the fundamental limitations of VAERS for certain types of analyses. While there is growing interest in and use of data mining and related analytic techniques with VAERS (e.g., Niu et al., 2001) and other spontaneous reporting systems, these techniques still require further study to establish their validity and reliability when used with large, automated, multipurpose data systems such as DMSS, and even more so with a system like VAERS. The committee considers routine application of data mining techniques to VAERS data inappropriate unless those techniques are thoroughly evaluated in other, more complete data sets such as those in DMSS and are shown to be effective even in the face of the kinds of biases inherent in the VAERS data. If the use of data mining in DMSS can be validated, the availability of data on health outcomes following exposure to AVA in both DMSS and

OCR for page 56
VAERS may provide an unprecedented opportunity to use associations identified in DMSS in subsequent efforts to validate the use of data mining in VAERS. It is possible that data mining techniques may be found to have the ability to detect more subtle or more complex associations than simpler analyses can. The committee notes, however, that AMSA’s routine screening of DMSS data is already generating hypotheses related to AVA that have not yet been investigated. Finding: The application of data mining and other statistical analysis techniques to screen data from VAERS and from DMSS data sets is still experimental. Recommendation: Hypothesis generation using data mining and other statistical techniques for screening data should be tested and validated in DMSS or other structured data sets before being considered for use with VAERS. Only if these techniques can be validated with a structured data set and then with VAERS data should they be used to generate hypotheses from VAERS concerning adverse events and AVA. Hypothesis Testing Planned Activities Hypotheses generated by the analysis of VAERS and DMSS data will be tested by the collaborative Analytic Unit being established at AMSA. The Analytic Unit will have access to the DMSS databases and to other medical records for military personnel who have received AVA and those who have not. These data can be used for various types of hypothesis-testing studies, including case-control or cohort studies or case-series analyses. Because the DMSS data resources appear similar to those of the civilian managed care organizations participating in CDC’s collaborative VSD project, the Analytic Unit may be able to apply methods, such as survival analysis, that are being used in VSD studies. CDC notes the need to develop methods to map the coding system used in VAERS to identify adverse events (COSTART) to the ICD-9-CM9 diagnosis codes used in the DMSS databases. Committee Comments The committee is pleased to see the evolving collaboration between CDC and DoD to facilitate the use of DMSS data for hypothesis testing. The DMSS databases are unique in that they contain relatively complete data on health care, AVA immunization, and personnel status for the active-duty military population—the group in whom AVA has been used most widely. Furthermore, as a product of the routine collection of administrative data, the DMSS data are fundamentally different from VAERS data, making them a good resource for testing hypotheses that may emerge from VAERS. But because DMSS will also be used to generate hypotheses, the committee notes the need for suitable plans to test those hypotheses with other data. Preliminary investigations, though, might include more detailed analysis of data from DMSS, including efforts that might involve the review of medical records. Finding: DMSS is a uniquely valuable resource for testing hypotheses regarding medically significant health effects of exposure to AVA or other vaccines, especially those that might arise several months after vaccination but within the period of active duty. In the committee’s view, CDC should be placing much greater emphasis on the hypothesis-testing aspect of this portion of the Anthrax Vaccine Safety and Efficacy Research Plan. Reviews of VAERS and DMSS data have already generated hypotheses that require further investigation. Although AMSA conducted some preliminary analyses at the request of the IOM committee that reviewed the safety and efficacy of the anthrax vaccine for DoD (IOM, 2002), resource constraints limited the extent of those analy- 9   ICD-9-CM refers to the International Classification of Diseases, Ninth Revision, Clinical Modification.

OCR for page 56
ses. Moreover, AMSA’s primary mission is surveillance, not hypothesis testing. Therefore, CDC’s establishment of the Analytic Unit at AMSA represents a real and valuable addition of resources to the effort to ensure that possible health risks associated with receipt of AVA are adequately studied. Recommendation: CDC should work with DoD to follow up the signals regarding AVA that have already been generated by the review of VAERS reports and preliminary analyses of DMSS data on hospitalization and outpatient visits (see IOM, 2002). The DMSS databases provide the opportunity to monitor the inpatient and outpatient health care provided to military personnel over the entire period of their active service. For those who receive AVA, this generally means that health care information is available for the period before vaccination, as well as for several months and potentially several years after vaccination.10 In addition, comparable information is available for military personnel who do not receive AVA but do receive other vaccines. This makes DMSS data particularly valuable and appropriate for studies of medically significant adverse events that might be associated with AVA. DMSS, however, does not routinely capture information on milder adverse events for which medical care is not sought or that do not result in time lost from duty. Recommendation: Analysis of DMSS data should be the primary approach for investigation of possible AVA-related health effects of medical significance that occur within the typical period of active duty following vaccination (perhaps as much as 3 to 4 years on average). Investigation of concerns about adverse events that occur in the months and years after vaccinated military personnel leave active duty will require access to data sources beyond DMSS. The committee urges CDC, in support of its hypothesis-testing activities, to explore other data sources and possible ways to link them with DMSS data. The Millennium Cohort Study, for example, may provide a framework for studying morbidity and mortality in a defined group of military personnel, if it includes adequate numbers of participants who received AVA. In addition, concerns about premature or cause-specific mortality might be investigated using data from the Department of Veterans Affairs (VA). Deaths of military personnel could be tracked through resources such as the Beneficiary Identification and Records Locator Subsystem of the VA, the Social Security Administration, and the National Death Index. The committee also emphasizes the importance of exploring the use of data from the VA health care system as an adjunct to DMSS for studies of morbidity. Although the VA health system serves only a small portion of the population of persons who have left military service, that population may be particularly well suited for identifying cases of unusual health problems. A new agreement between DoD and VA for greater exchange of health information (DoD, 2002; MacKay and Chu, 2002) may facilitate link-ages between DMSS and VA health records. Recommendation: To allow for analysis of health effects of AVA that might arise following the completion of active duty, CDC should investigate the use of DMSS data in conjunction with morbidity and mortality data from the Millennium Cohort Study and the health system of the Department of Veterans Affairs. Deaths of military personnel identified through DMSS could be tracked through resources such as the Beneficiary Identification and Records Locator Subsystem of the VA, the Social Security Administration, and the National Death Index. The committee is concerned that insufficient priority is being given to the hypothesis-testing work that is to be done through the Analytic Unit being established at AMSA. The proposed funding level reflected in the version of the CDC–DoD Memorandum of Understanding provided to the committee (a total of $950,000 over 3 years) does not appear adequate to support the kind of analysis of DMSS data that will be necessary to investigate hypotheses that have already been generated by AMSA’s routine screening of DMSS data and by the work of the IOM committee that reviewed the safety and efficacy of the 10   Typical Army enlistment is 2 to 6 years (Grabenstein, 2001).

OCR for page 56
anthrax vaccine for DoD (IOM, 2002). Some of these analyses are likely to require time-consuming collection and examination of individual medical records, not just automated analysis of electronic records. Moreover, CDC’s planned hypothesis-generating activities will only add to the demand for additional hypothesis-testing analyses of DMSS data. Recommendation: Adequate resources (substantially more than can currently be identified from the CDC–DoD Memorandum of Understanding) should be made available to support the use of DMSS data for testing hypotheses regarding health effects related to AVA or other vaccine exposures. Management and Oversight of Activities Related to Hypothesis Generation and Hypothesis Testing Committee Comments The committee also had concerns about the management and oversight of the hypothesis-generating and hypothesis-testing activities that CDC has described. These activities should be guided by an overall study plan or strategy to ensure an appropriate balance and coordination between hypothesis-generating and hypothesis-testing activities. Comprehensive oversight, based in CDC but allowing for coordination with DoD, is needed to establish priorities for this set of activities as a whole and for specific activities as well. Such oversight is also needed to address other matters, including ensuring timely and systematic hypothesis testing when hypotheses emerge. In the committee’s view, the considerable emphasis placed on hypothesis generation through the use of experimental methodologies of data mining compared with the limited attention given to hypothesis-testing activities illustrates the lack of overall guidance and priority setting for these activities. The materials from CDC do not provide a clear indication that this component of the AVA research program currently has the overall planning and guidance that the committee sees as necessary. Because DMSS is such a critical resource for both hypothesis generation and hypothesis testing, ongoing coordination between CDC and DoD seems essential. In addition, the committee believes that CDC’s hypothesis-testing activities would benefit from a periodic assessment by outside experts who could provide advice on matters such as study design and analytic techniques, as well as the priorities for the analyses. The project timeline (CDC, 2002j) refers to the establishment of an Analytic Unit Advisory Committee in July 2002, but no information was provided to the committee concerning the anticipated membership or activities of this group. Finding: An overall study plan or strategy is needed to guide CDC’s use of VAERS, DMSS data sets, and other data sources for hypothesis-generating and hypothesis-testing activities related to AVA. Recommendation: CDC, working with DoD, should establish a staff team with overall responsibility for the review and analysis of VAERS and DMSS data for both hypothesis generation and hypothesis testing related to AVA. Recommendation: A committee of nongovernmental experts should be established to periodically advise CDC on plans and priorities for the analyses of data from DMSS and other sources to test hypotheses regarding health effects related to AVA.

OCR for page 56
POSSIBLE ROLE OF ALUMINUM HYDROXIDE ADJUVANT IN AVAASSOCIATED ADVERSE EVENTS Potential Research Topics CDC presented to the committee a brief review of issues related to aluminum-containing adjuvants and adverse events (CDC, 2002l). Aluminum is one of the most abundant elements on earth, and humans are routinely exposed to it through sources such as drinking water, medications (e.g., antacids), and deodorants. Little is known about the toxicology of injected aluminum, and human studies of the clearance of aluminum have been limited to patients with chronic renal insufficiency. Aluminum compounds are used as adjuvants in some vaccines to enhance the immunogenicity of the product. AVA uses aluminum hydroxide as an adjuvant, and several other vaccines routinely administered to children and adults also use adjuvants containing aluminum. These vaccines include the diphtheria and tetanus toxoids and pertussis vaccine (DTP), other vaccines containing tetanus toxoids, and hepatitis B vaccine. While vaccine adjuvants containing aluminum have been used for many years, they have been associated with some local adverse events, such as erythema, subcutaneous nodules, skin allergy, and skin inflammation at the injection site. Adverse events of this type also occur in persons who receive AVA, but studies have not been done to establish whether the adverse events observed following receipt of AVA are related to the adjuvant. Recently, it has been suggested that aluminum-containing vaccine adjuvants might be associated with a condition that has been labeled macrophagic myofasciitis (MMF) (Gherardi et al., 1998, 2001). The symptoms attributed to this condition include myalgias, arthralgias, muscle weakness or tenderness, and fatigue. Aluminum has been found in tissue biopsies of persons considered to have the condition, but tissue biopsies from suitable control groups have not been tested. Thus, it remains uncertain whether the presence of aluminum in tissue biopsies of persons said to have MMF is a sign of pathology or only a coincidental finding. CDC (2002l) listed five possible research questions that might be investigated: Are subcutaneous nodules following AVA vaccination caused by subcutaneous accumulation of aluminum hydroxide adjuvant? Is the gender differential in the occurrence of adverse events observed following AVA vaccination associated with impaired local clearance of aluminum hydroxide adjuvant? Are individuals (particularly women) with iron deficiency anemia more prone to develop injection-site adverse events, including subcutaneous nodules, following AVA vaccination? Are individuals (particularly women) with iron deficiency anemia more prone to develop systemic adverse events following AVA vaccination? Is macrophagic myofasciitis (MMF) a condition associated with AVA? CDC assigned this topic a lower priority than the other proposed studies and noted in the materials provided to the committee that no study proposals or protocols had been developed. Committee Comments The committee acknowledges that there are concerns that the aluminum adjuvant in AVA might contribute to certain types of adverse events or adverse events in certain people. However, a summary of a May 2000 workshop on aluminum in vaccines (Eickhoff and Myers, 2002) indicates that the pervasiveness of aluminum in the environment and the limited understanding of the toxicology and kinetics of injected aluminum adjuvants pose serious scientific and practical challenges to efforts to investigate health effects that might be associated with a specific source of exposure, such as AVA. Furthermore, the preliminary nature of the evidence concerning MMF suggests that it is premature to investigate whether re-

OCR for page 56
ceipt of AVA is associated with that condition. Given these constraints, the committee does not consider it appropriate for CDC to pursue research on any of the proposed questions as part of the Anthrax Vaccine Safety and Efficacy Research Plan. Finding: Widespread environmental exposure to aluminum makes it difficult to conduct a study of potential adverse effects of exposure to the aluminum hydroxide adjuvant/adsorbant in AVA. Finding: The significance of the presence of aluminum in tissue biopsies of persons diagnosed with the condition called macrophagic myofasciitis has not been established. Recommendation: The study of the possible role of the aluminum hydroxide adjuvant in adverse events following receipt of AVA should be eliminated from the CDC research program. REPRODUCTIVE HEALTH CDC included among its research questions whether receipt of AVA is associated with adverse effects on women’s reproductive health. The committee agrees that this is an important concern if the vaccine is to be routinely administered to women in the military, or if circumstances should require vaccination of civilian women. CDC notes that several DoD studies are investigating aspects of this topic, including fertility rates following vaccination and rates of birth defects among vaccinated women.11 The committee notes that none of the CDC research studies directly address the relationship of AVA vaccination with reproductive health. REFERENCES AMSA (Army Medical Surveillance Activity). 2001. Surveillance of Adverse Effects of Anthrax Vaccine Adsorbed: Results of Analyses Requested by the Institute of Medicine Committee to Assess the Safety and Efficacy of the Anthrax Vaccine. Washington, D.C.: Army Medical Surveillance Activity, U.S. Army Center for Health Promotion and Preventive Medicine. CDC (Centers for Disease Control and Prevention). 2002a. Protocol 1: AVA human reactogenicity and immunogenicity trial to address change in route of administration and dose reduction. Anthrax Vaccine Safety and Efficacy Plan. Atlanta: Centers for Disease Control and Prevention. CDC. 2002b. Protocol 6: study protocol of long-term adverse effects from anthrax vaccination among civilian work-ers (draft). Anthrax Vaccine Safety and Efficacy Plan. Atlanta: Centers for Disease Control and Prevention. CDC. 2002c. Protocol 7: the VHC platform for AVA related research (draft). Anthrax Vaccine Safety and Efficacy Plan. Atlanta: Centers for Disease Control and Prevention. CDC. 2002d. Section 1: anthrax vaccine safety and efficacy plan. Anthrax Vaccine Safety and Efficacy Plan. Atlanta: Centers for Disease Control and Prevention. CDC. 2002e. Section 3: critical research questions table. Anthrax Vaccine Safety and Efficacy Plan. Atlanta: Centers for Disease Control and Prevention. CDC. 2002f. Section 6: study summary: AVA human reactogenicity and immunogenicity trial to address change in route of administration and dose reduction. Anthrax Vaccine Safety and Efficacy Plan. Atlanta: Centers for Disease Control and Prevention. CDC. 2002g. Section 11: study summary: study protocol of long-term adverse effects from anthrax vaccination among civilian workers. Anthrax Vaccine Safety and Efficacy Plan. Atlanta: Centers for Disease Control and Prevention. CDC. 2002h. Section 12a: Walter Reed National Vaccine Healthcare Center (VHC) network summary. Anthrax Vaccine Safety and Efficacy Plan. Atlanta: Centers for Disease Control and Prevention. CDC. 2002i. Section 12b: study summary: the VHC platform for AVA related research. Anthrax Vaccine Safety and Efficacy Plan. Atlanta: Centers for Disease Control and Prevention. 11   DoD policy exempts from vaccination women known to be pregnant, but some women with early unconfirmed pregnancies may be vaccinated inadvertently.

OCR for page 56
CDC. 2002j. Section 13a: study summary: enhanced signal detection and hypothesis testing for adverse events following anthrax vaccination. Anthrax Vaccine Safety and Efficacy Plan. Atlanta: Centers for Disease Control and Prevention. CDC. 2002k. Section 13b: the memorandum of understanding (MOU) between AVSA/NIP/CDC and AMSA/USACHPPM/DoD. Anthrax Vaccine Safety and Efficacy Plan. Atlanta: Centers for Disease Control and Prevention. CDC. 2002l. Section 14: study summary: possible role of aluminum hydroxide adjuvant in AVA-associated adverse events, potential areas for future research. Anthrax Vaccine Safety and Efficacy Plan. Atlanta: Centers for Disease Control and Prevention. DoD (Department of Defense). 2002. News release: agreement on federal health information exchange announced. Department of Defense, Washington, D.C. Eickhoff TC, Myers M. 2002. Workshop summary: aluminum in vaccines. Vaccine 20:S1–S4. Gherardi RK, Coquet M, Cherin P, Authier FJ, Laforet P, Belec L, Figarella-Branger D, Mussini JM, Pellissier JF, Fardeau M. 1998. Macrophagic myofasciitis: an emerging entity. Groupe d’Etudes et Recherche sur les Maladies Musculaires Acquises et Dysimmunitaires (GERMMAD) de l’Association Francaise contre les Myopathies (AFM). Lancet 352(9125):347–352. Gherardi RK, Coquet M, Cherin P, Belec L, Moretto P, Dreyfus PA, Pellissier JF, Chariot P, Authier FJ. 2001. Macrophagic myofasciitis lesions assess long-term persistence of vaccine-derived aluminium hydroxide in muscle. Brain 124:1821–1831. Grabenstein JD. 2001. VAERS information; average length of active duty. E-mail to Joellenbeck L, Institute of Medicine , Washington, D.C., November 13. IOM (Institute of Medicine). 2001. CDC Anthrax Vaccine Safety and Efficacy Research Program. Interim Report. Washington, D.C.: National Academy Press. IOM. 2002. Joellenbeck LM, Zwanziger LL, Durch JS, Strom BL, eds. The Anthrax Vaccine: Is It Safe? Does It Work? Washington, D.C.: National Academy Press. MacKay L, Chu DS. 2002. Memorandum of agreement for federal health information exchange governance and management. Department of Veterans Affairs and Department of Defense, Washington, D.C. Niu MT, Erwin DE, Braun MM. 2001. Data mining in the U.S. Vaccine Adverse Event Reporting System (VAERS): early detection of intussusception and other events after rotavirus vaccination. Vaccine 19(32):4627– 4634. VAERS (Vaccine Adverse Event Reporting System). 2001. A federal program for surveillance of vaccine safety. Brochure. Rockville, Md.: Vaccine Adverse Events Reporting System. Ware JE, Snow KK, Kosinski M, Gandek B. 1993. SF-36 Health Survey Manual and Interpretation Guide. Boston: New England Medical Center, The Health Institute. Wolfowitz P. 2002. Reintroduction of the Anthrax Vaccine Immunization Program (AVIP). Memorandum for Secretaries of the Military Departments; Chairman of the Joint Chiefs of Staff; Undersecretaries of Defense; Assistant Secretaries of Defense; General Counsel, Department of Defense; Inspector General, Department of Defense; Directors of Defense Agencies; Commandant of the U.S. Coast Guard. Department of Defense, Washington, D.C.