3
Building the Framework for Evaluating Health Hazard

Ultimately, the purpose of aerosol biodetector systems is to provide for early warning of an airborne health hazard. Biodetection should, therefore, focus on those aerosol particles that have the potential to cause deleterious biological effects in humans. Since biological activity is the principal focus, bioaerosol measurements should be reported in terms that quantify this characteristic. A measure based on anything but biological activity would focus attention away from the central purpose of biodetection—identifying and quantifying health hazard. Thus, biological activity provides a convenient starting point for developing a framework for evaluating risk given the wide range of biological agents and detection methods. The approach described here is general and applicable to research beyond the DoD T&E community.

The health impact of BWA exposure is affected by both the physical and biological characteristics of the aerosol. To be of value across the full spectrum of biological threats, an evaluation framework should include the principal physical and biological factors affecting health hazard. In the previous sections, it was concluded that the two main aerosol cloud characteristics affecting health hazard are (1) physical properties that determine site of deposition in the respiratory tract and (2) biological activity at the deposition site. A framework that incorporates information about these key determinants would allow an analytical assessment of health hazards. It would also provide an appropriate and interconvertible standard for the testing of biodetectors.

In the sections that follow, an approach for quantifying the health hazards from aerosols is presented. Existing measurement technologies that support the use of this framework are described. Finally, an approach to detector testing and the connection between health risk and detector testing is explored.

3.1
THE FRAMEWORK IN WORDS

Biological activity can be described as the concentration of biologically active aerosol particles (BAP) or total biologically active units (BAU) in a volume of aerosol cloud. The BAU can be made up of virions, bacteria, toxin molecules, or a mixture. Aerosol particles may also contain nonactive substances that affect assessment of biological activity, such as assay inhibitors, growth media, or salts. As discussed in the previous chapter, current research suggests that the number of BAU has greater influence on health risk than the number of BAP. However, this conclusion is based on minimal data for only one bacterial agent. The possibility exists that future research will identify threat agents for which BAP may strongly influence health risk (e.g., by determining the number of lesions formed). For this reason both primary and secondary units of bioaerosol activity measure are recommended below.

The uncertainty in relative importance of the number of aerosol particles deposited at a



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 29
3 Building the Framework for Evaluating Health Hazard Ultimately, the purpose of aerosol biodetector systems is to provide for early warning of an airborne health hazard. Biodetection should, therefore, focus on those aerosol particles that have the potential to cause deleterious biological effects in humans. Since biological activity is the principal focus, bioaerosol measurements should be reported in terms that quantify this characteristic. A measure based on anything but biological activity would focus attention away from the central purpose of biodetection—identifying and quantifying health hazard. Thus, biological activity provides a convenient starting point for developing a framework for evaluating risk given the wide range of biological agents and detection methods. The approach described here is general and applicable to research beyond the DoD T&E community. The health impact of BWA exposure is affected by both the physical and biological characteristics of the aerosol. To be of value across the full spectrum of biological threats, an evaluation framework should include the principal physical and biological factors affecting health hazard. In the previous sections, it was concluded that the two main aerosol cloud characteristics affecting health hazard are (1) physical properties that determine site of deposition in the respiratory tract and (2) biological activity at the deposition site. A framework that incorporates information about these key determinants would allow an analytical assessment of health hazards. It would also provide an appropriate and interconvertible standard for the testing of biodetectors. In the sections that follow, an approach for quantifying the health hazards from aerosols is presented. Existing measurement technologies that support the use of this framework are described. Finally, an approach to detector testing and the connection between health risk and detector testing is explored. 3.1 THE FRAMEWORK IN WORDS Biological activity can be described as the concentration of biologically active aerosol particles (BAP) or total biologically active units (BAU) in a volume of aerosol cloud. The BAU can be made up of virions, bacteria, toxin molecules, or a mixture. Aerosol particles may also contain nonactive substances that affect assessment of biological activity, such as assay inhibitors, growth media, or salts. As discussed in the previous chapter, current research suggests that the number of BAU has greater influence on health risk than the number of BAP. However, this conclusion is based on minimal data for only one bacterial agent. The possibility exists that future research will identify threat agents for which BAP may strongly influence health risk (e.g., by determining the number of lesions formed). For this reason both primary and secondary units of bioaerosol activity measure are recommended below. The uncertainty in relative importance of the number of aerosol particles deposited at a 29

OCR for page 29
30 site in the respiratory tract versus the total number of biologically-active units deposited is a minor aspect of the inherent uncertainty associated with characterizing the biological threat. A detailed discussion of the effect of biothreat uncertainty on biodetector testing is outside the scope of this study, however, a summary is provided in Box 3.1. Briefly stated, evaluating biodetector accuracy is severely constrained by inherent threat uncertainty (e.g., which strain will be encountered, what the particle distribution will be, what the agent will be suspended in). However, high precision in biodetector testing can be achieved by well-controlled system and component testing with simulants and well-chosen threat agent prototypes. 3.1.1 Normalization of Agent Concentration to Health Hazard Biological activity with respect to the ability to cause an adverse health response is routinely characterized using controlled biological assays. The observed biological end point BOX 3.1 Precision and Accuracy in Biodetector Testing The number of different bacteria, viruses, and toxins that can cause adverse health effects is great. For each bacterial and viral type of agent (e.g., the genus and species B. anthracis, VEE virus), myriad strains with varying biological and physical properties occur naturally; anthropogenic effects, from agricultural use of antibiotics to genetic engineering, may alter natural strains. Some infective strains are culturable using artificial media; others will not grow in such an environment. Which specific microorganism will a biodetector actually be confronted with on the battlefield? The mode of aerosol dissemination (e.g., dry or wet dissemination, particle size distribution), suspension media (e.g., addition of fluidizers to prevent clumping, chemicals to absorb killing UV light and potentiate agent biological activity), and natural background materials (e.g., chemical pollutants, pollen, and ambient microbial flora) can greatly affect the nature and health effects of the aerosol cloud. It is not presently possible to fully test and characterize the accuracy of a detector against this vast array and combination of potential agents and aerosol presentations; nor could current detectors identify the entire range of agents, even if such testing were possible. Therefore, biodetector accuracy testing is inherently limited, in contrast to the great precision that can be achieved by employing well-controlled test procedures. A practical and useful level of accuracy can be approached by carefully selecting prototypical strains and bounding key aerosol parameters, such as particle size. A robust understanding of pathogenesis and phylogenetic diversity (e.g., selecting strains with conserved virulence epitopes) for a pathogen is critical to constructing such a practical level of accuracy in biodetection testing. varies, depending on the mechanism(s) of pathogenicity for the particular agent and the assay system. Results of these assays are often expressed as a statistical function, such as the dose that is lethal, infective, or effective for 50 percent of an exposed population (LD50, ID50, or ED50, respectively). Expressing biological activity as a statistical unit of health consequence provides the opportunity to “normalize” activity units across the spectrum of biothreat agents. One can compare the effects of a set number of LD50 units for several disease-causing microorganisms. For example, in vaccine efficacy testing, vaccine treatment groups are often challenged with 100 LD50 of agent. When articulating the performance of a biodetector, the threshold of detection for any number of agents can be expressed in terms of LD50 units. Converting the actual number of bacteria, viruses, or mass of toxin to units of biological activity provides for greater information

OCR for page 29
31 content and facile appreciation of relative hazard. In addition, normalized units of adverse biological activity enable comparison of health hazards across diverse modes of exposure (e.g., pulmonary, cutaneous, ingestion). A statistical representation of dose response, such as LD50, is of great use in describing health hazard, and some effective models and framework for dose response analysis have been developed. However, there are practical limitations in utilizing such statistical representations. Limitations include lack of knowledge of the shape of the dose response curve (e.g., probit slope) and fidelity of animal model data in representing human response. Lack of knowledge of the dose response curve can be a significant concern when low incidence effects are important. For example, a commander may be concerned about an LD1 if the infectious agent is highly contagious, since a single infected person could pass the disease to others. For pathogens with large LD50 values (e.g., B. anthracis), LD1 can vary greatly depending on the shape of the population dose response curve as well as specific host susceptibility (e.g., age, health, and immune status). Biological activity assays are surrogates for human health effects. The assays employed for different biothreat agents vary in fidelity for predicting human disease. Therefore, the robustness of predicting health effects in humans for a particular agent, as well as for comparative assessments across agents, varies with the fundamental knowledge base for the agent as well as availability of suitable animal or ex-vivo models. The importance of understanding the mechanism(s) of pathogenicity for each biothreat agent, whether for estimating aerosol health risks, medical countermeasures development, or biodetector testing cannot be overstated. One of the benefits of adopting a detector evaluation framework based on biological activity is that it will focus research attention on areas where understanding of pathogenicity is poor. Use of in-vivo and ex-vivo data from multiple animal species and challenge modes (e.g., pulmonary, oral, cutaneous) can provide greater fidelity for estimating human health effects and hazards. There is currently a paucity of data for some pathogens for converting detector output to human health hazard. Knowledge gaps in human dose response and modes of action of pathogens, of which there are many, will be brought into clear view when BAULA is implemented. Currently, due to this lack of knowledge, field commanders “finesse” interpretation of biodetector outputs to guide operational decisions. Rapid progress is being made in characterization of pathogens of importance to DOD, due to investments by several federal departments, but generating detailed and accurate non-human dose response curves for all bioagents will be difficult and costly. Even relatively crude estimates of pathogenicity and infectivity, however, are arguably more useful than using a measurement system that implicitly assumes that all agents are equally hazardous. 3.2 PROPOSED FRAMEWORK FOR EVALUATING AEROSOLIZED BIOTHREAT AGENTS The committee recommends that the following units be adopted as Department of Defense-wide standards: BAULADae. A primary unit of concentration is recommended for biological aerosols that quantifies Biologically Active Units per Liter of Air as a function of aerodynamic diameter (BAULADae). This unit permits incorporation of a predictive model of the site of deposition sites in humans. Biological activity is itself a compound term that incorporates the quantity of active

OCR for page 29
32 agent, strain identity, and some probability of physiological response. The BAULADae provides key information needed to estimate health hazard. Biologically active units are measured and calculated differently for each threat agent, to provide a common measure of hazard for agents with disparate modes and levels of biological action, and to account for different activity levels for particles that deposit in different regions of the respiratory tract or otherwise affect human response. The biological activity is normalized to a 50 percent population activity end point. An example of the use of this unit in estimating the health hazard for an aerosol challenge with Bacillus anthracis spores would be as follows. An aerosol would be generated having a particle size distribution of 1-3 µm number median aerodynamic diameters; Dae is 1-3µm. A volume of aerosol might be collected into a liquid and the original aerosol particles disrupted to yield individual spores. The spores could be quantified on agar plates yielding CFU/ml of collection fluid volume. The number of LD50 units contained in the original aerosol would be calculated based on air volume sampled, volume of liquid in which the spores were suspended, volume plated on agar plates, and estimated number of CFU in an LD50 for B. anthracis. BAPLADae. The committee proposes a secondary framework for bioaerosol measurement to quantify the size distribution of particles in an aerosol that contain biologically active material. This measure is the size-resolved number of biologically active particles per liter of air for an aerosol of a defined size distribution (BAPLADae). This secondary unit of measure could be used to provide a comparative unit for historical comparisons with ACPLA measurements and in anticipation of the discovery that the number of sites where aerosol particles deposit biologically active units in the respiratory tract may need to be known to predict health hazard for some agents. As BAULADae represents total agent units per liter rather than units per particle, the site information is lost when using BAULADae. BAPLADae would be determined in a manner similar to current methods for ACPLA, with the exception of employing a defined challenge aerosol particle size distribution and normalizing biological activity for each agent. In the case of a B. anthracis challenge, a defined volume of air would be impacted over time on a rotating slit impactor sampler. The number of CFU growing on the plate would be used to estimate the concentration of aerosol particles containing spores and adjusted for the number of CFU in an LD50 of B. anthracis. As with BAULADae, measured or estimated values of LD50 for agents are needed to determine BAPLADae values. In addition, the combination of these two units, BAULADae and BAPLADae, can also provide an estimate of the mean number of biologically active units per particle of size Dae that contains any biological activity. Over time, more will be learned about the relative health risks of BWA exposures, so that the quantity of an agent that represents one BAULADae may change for any given agent. Linking BAULADae to the probability of a negative health outcome, as will be discussed in the next section, allows for an adjustment of test requirements without fundamentally altering the unit. 3.3 A MATHEMATICAL DESCRIPTION OF BIOAEROSOL HEALTH RISK The health risk presented by a bioaerosol includes two components: the hazard posed by the agent and the physiological responses of the individuals exposed to it. Thus, a cloud of smallpox virus is a hazard, but it may pose no health risk if all of the exposed individuals are effectively immunized against smallpox. What the committee is proposing in BAULADae is a unit that quantifies the hazard posed by a particular bioaerosol. The BAULADae hazard

OCR for page 29
33 framework (see Figure S.1, p. 7) can then be embedded in a full health risk framework that places the hazard measurement in the context of the likely physiological impact on the particular population in question. The final result provides an accurate assessment of health risk. FIGURE 3.1 A framework for evaluating the health risk posed by aerosolized biological warfare agent exposure. The committee performed a survey of common units of measure and concentration used in the bioaerosol community and found that these fell into four general categories. Table 3.1 summarizes the comparison of BAULADae, BAPLADae, and other common units. Note that the new units are the only ones that take into account the factors necessary to evaluate the actual health hazard posed by any BWA aerosol as described in Equation 3.1. In all other cases, the concentration measures do not provide enough information to adequately evaluate the health hazard. The committee asserts that a framework that uses health hazard to compare biowarfare agents best addresses the charge presented by the statement of task to propose a unified method for comparing bioaerosol threats. However, as the equation describing the recommended framework shows, converting to a measure of health hazard is complex. Even a minimal estimate of health hazard requires accurate measures of lethality and infectivity, which have not been established for all possible biological agents. Furthermore, in a field environment, there is likely to be incomplete knowledge of how many other factors might affect health risk, from weather conditions, to host immune status, to precise strain characteristics. The committee is aware that even BAULADae and BAPLADae represent incomplete measures of health hazard, but propose that their application would focus attention on the most important factors affecting troop protection, factors that are not included in the current unit of measure.

OCR for page 29
34 TABLE 3.1 Comparison of Common Units of Measure for Biological Material Physical Characteristics Biological Characteristics Total Amount LD50 Particle Size Activity of agent of Active Agent Agent containing No No No No particles per liter of air Particles/unit No No No No volume of air Mass/unit No No No No volume of air Viable No No Yes Yes organisms/unit volume of air BAPLA No Yes Yes Yes Dae BAULA Yes Yes Yes Yes Dae The additional information provided by the units described above is apparent when viewed in the broader context of health risk evaluation. Figure 3.1 provides a framework for exploring the feasibility of developing a standard approach to measurement that directly addresses the primary concern, risk of adverse health outcomes. In the discussion that follows, we develop a mathematical framework for this representation of hazard. This mathematical framework will then be used to examine the performance of sensors in light of their ultimate purpose, which is to reduce risk. Airborne biological agents exist as aerosols (i.e., as suspensions of particles: liquid, solid, or mixed-phase) in air. No aerosol, even the best calibration aerosol, is truly monodisperse (i.e., consists only of identical-size particles); all aerosols contain particles that are distributed over some range of sizes. To understand the dose response to aerosol exposure we must address the contributions of particles of different sizes by introducing a particle size distribution function. The number of particles per unit volume of air with aerodynamic diameters between Dae and Dae+dDae at a point in space and time is dN = n(Dae , x, y,z,t)dDae . (1)

OCR for page 29
35 Equation (1) describes the distribution of particles without reference to their agent content. n(Dae,x,y,z,t) is the differential particle size distribution with respect to aerodynamic diameter of the particles (hereafter labeled the particle size distribution), and is determined by the formulation of agents, the nature of the release, and the atmospheric processes involved in transport and transformation of the agent cloud, as well as all other sources of particles. Definitions of terms used in this section may be found in Box 3.2. Even when an agent is present at dangerous levels, many or even most aerosol particles will be common, agent-free ambient particles, such as dust, pollen, fungal spores, other natural aerosols, emitted primary particles, and secondary aerosols formed from gas phase chemical reactions. The units of n are {number of particles}/{unit volume of air}/{unit of aerodynamic diameter}( i.e., m-3µm-1). Dae is the aerodynamic diameter of the particle, which corresponds to the diameter of a spherical particle with unit specific gravity (density of water) that would sediment at the same velocity as the particle. It is selected as the basis for defining the particle size distribution since, for all but the smallest isolated virions or toxin molecules, the aerodynamic diameter is the particle characteristic that determines the mechanism and efficiency with which ambient exposure translates into dose in the different regions of the respiratory tract or other parts of the body. dDae is a differential interval in aerodynamic diameter. To determine the dose or instrument response, we will eventually integrate over particle size. We will, however, retain the differential form in order to maintain rigor as we consider the many factors that may influence the dose and human response. While present measurement methods may not be able to fully resolve this differential distribution, this mathematical formulation provides a base from which the approximations inherent in different measurement methods can be elucidated. The amount of agent in the air at a point in space and time is best expressed as a concentration or quantity of agent per unit volume of air. Some of the sensors that will be examined in the discussion that follows may not directly measure the biological agent; instead, they estimate or infer exposure from the concentration of detected particles in the air. The human response to exposure to airborne agent depends on the particles that deposit into different regions of the body, especially the respiratory tract. As described in section 2.2.2, the International Commission on Radiological Protection (ICRP 1994) has generated a detailed human respiratory tract model that describes particle deposition into different regions of the respiratory tract (extrathoracic (ET), tracheobronchial (TB), and pulmonary (P) regions) as a function of particle size and breathing rate. That model forms a basis for estimating the probability that a particle of size Dae will deposit into each of these regions, including effects of nasal or oral breathing as discussed in Chapter 2. The number of particles per unit volume of air in the differential size interval that will deposit into any one of the regions of the respiratory tract if inhaled is given by the products of the probability of deposition in that region and the number concentration of particles in the size interval, that is., dN Pe = Pe (Dae ,Q)dN = Pe (Dae ,Q)n(Dae , x, y, z, t) dDae Extrathoracic region: dN Pt = Pt (Dae ,Q)dN = Pt (Dae ,Q)n(Dae , x, y,z,t)dDae Tracheobronchial region: dN Pp = Pp (Dae ,Q)dN = Pp (Dae ,Q)n(Dae , x, y, z, t ) dDae Pulmonary region: (2) where Pi(Dae,Q) denotes the probability that a particle with aerodynamic diameter Dae will

OCR for page 29
36 BOX 3.2 Variables in the Health Risk Equation Dae Aerodynamic diameter t Time Q(t) Volumetric inhalation rate ν (Dae) Number of agent units of aerodynamic diameter Dae [0, inf) ν (Dae) Pν (Dae) Probability that a particle of aerodynamic diameter Dae contains agent units [0,1] Pi(Dae,Q) Probability that a particle with aerodynamic diameter will deposit in region i of the respiratory tract when inhaled at a rate Q where i=1: Extrathoracic region; i=2: Tracheobronchial region; i=3: Pulmonary region [0,1] PA( ν ,Dae) Probability that a biological agent in a particle of size Dae is active (µm) n( Dae , x, y, z , t ) Differential particle size distribution as a function of aerodynamic diameter Dae space and time ({number of particles} / {m3} / {µm}) P I( ν ) Probability of an adverse health outcome originating in region i dNPi Differential probability that a particle with aerodynamic diameter Dae will deposit in region i of the respiratory tract when inhaled at rate Q dNCi Differential number of cells that can be expected to deposit into each region i of the respiratory tract per unit volume of air inhaled if the probability that a particle of diameter Dae contains ν cells (or other agent entities) is Pv (Dae) dNAi Differential number of active biological agents per unit volume of inhaled air that will deposit into region i of the respiratory tract if the probability that a cell in a particle of size Dae is active as PA (v,Dae) dNSi Differential number of sites that will be produced in region i of the respiratory tract where infection may develop from deposition of a particle that contains one or more active cells per unit volume of air inhaled dNIi Differential probability of an adverse health outcome originating in region i of the respiratory tract deposit in region i of the respiratory tract when inhaled at a volumetric inhalation rate Q. In the discussion that follows, we will consider separately the three deposition regions; the forms of the equations for the different regions are similar, so we will use the subscript i to denote the region of interest, dN Pi = Pi (Dae ,Q)dN = Pi (Dae ,Q)n(Dae , x, y,z,t)dDae (i=1,2,3) (3)

OCR for page 29
37 (i=1: extrathoracic region (ET); i=2: tracheobronchial region (TB); i=3: pulmonary region (P)). While exposure through the respiratory regions is among the most important, and is illustrated explicitly here, other modes of exposure are easily accounted for in the context of this approach. For example, many particles deposited in the ET region will be suspended in mucus and ultimately ingested, which may be the primary route of infection. The mathematical formula presented here may be modified to include this and other modes of exposure as appropriate for a given agent. Each deposited particle may carry one or more cells, spores, virions, or other irreducible agent entities, hereafter described as cells. The total number of cells deposited into any region can be considered to be the cell dose. If the probability that a particle of diameter Dae contains ν cells (or other agent entities) is, Pν (Dae ) the number of cells that can be expected to deposit into each region per unit volume of air inhaled is dNCi = ∑ν(Dae )Pν (Dae )Pi (Dae ,Q)n(Dae , x, y,z,t)dDae (i=1,2,3) (4) ν ν avg ,i Dae = ∑ν (Dae )P(Dae ) . (5) ν Not all cells that deposit will be biologically active. The probability that a cell remains active may vary with particle size, time since release, temperature, relative humidity, and other factors; cells in the core of an aggregate of cells may be protected by those that surround it. Denoting the probability that a cell in a particle of size Dae is active as PA (ν,Dae ) , the number of active biological agents per unit volume of inhaled air that will deposit into region i of the respiratory tract becomes dN Ai = ∑ν(Dae )Pν (Dae )PA (ν,Dae )Pi (Dae ,Q)n(Dae , x, y,z,t)dDae (i=1,2,3). (6) ν If no agent is present in a particle, ν =0. If none of the agent is active, PA=0. In either case, the agents may be counted by some detectors, but will not produce infection. A deposited particle that contains at least one active agent creates a site at which infection may develop. Some agents may be sufficiently infectious that a single active cell can induce infection. In that case the probability of adverse health outcome would depend on the number of sites where any viable agent deposits. The number of such sites that will be produced in the respiratory tract where infection may develop from deposition of a particle that contains one or more active cells is, per unit volume of air inhaled, dN Si = ∑ Pν (Dae )PA (ν,Dae )Pi (Dae ,Q)n(Dae , x, y,z,t)dDae , (i=1,2,3). (7) ν ≥1 The total numbers of cells deposited and potentially infectious sites created in different regions of the respiratory tract depend on the time of exposure to the agent-containing cloud and the respiration rate, and must include the contributions of all particle sizes. The resulting total number of biologically active cells deposited becomes: ∞ T ∫ Q(t) ∫ ∑ν(D N Ai = )Pν (Dae )PA (ν,Dae )Pi (Dae ,Q)n(Dae , x, y,z,t)dDae dt , (i=1,2,3). (8) ae ν 0 0

OCR for page 29
38 Similarly, the numbers of potential infection sites created in the different regions become: ∞ T ∫ Q(t) ∫ ∑ Pν (D N Si = )PA (ν,Dae )Pi (Dae ,Q)n(Dae , x, y,z,t)dDae dt , (i=1,2,3). (9) ae ν 0 0 Health outcomes can be expressed in terms of probabilities of infection, incapacitation, or death. These depend upon the numbers of agent cells or potential infection sites that result from exposure in each of the deposition regions. We denote the probability of an adverse health outcome originating in region i as Pi(NSi,NAi). The cumulative probability of an adverse health outcome originating in any region of the respiratory tract, or other deposition, can be determined by considering the independent probabilities of no adverse outcomes in any of the R deposition regions, i.e., ⎛Probability of any ⎞ R ⎟ = 1− Π (1− Pi ) . ⎜ (10) ⎝adverse health outcome ⎠ i=1 The foregoing analysis provides a formalism through which the hazard, measured as a probability of an adverse health outcome, can be related to measurable properties of a bioaerosol cloud, taking into account the size distribution of the aerosol particles, the variation of the composition with particle size, the behavior of particles when inhaled, and the probability that deposited particles will induce an adverse outcome. In spite of the level of detail presented, the BAULADae unit does not reach the level of representing risk as opposed to hazard. Risk will depend on the actual population exposed and, therefore, will have to include such factors as variation in deposition probability among individuals, and variations in responses of different subpopulations. Nonetheless, this analysis provides a reasonable basis for examining the suitability of monitors to assess the hazard posed by bioaerosol releases, to probe the meanings of the presently employed units of measure for bioaerosols, and to develop an evaluative framework that is applicable to all agents and all sensing technologies. 3.4 APPLICATION OF THE EQUATION TO DETECTORS The probability of adverse health outcome is a function of the probabilities of disease initiation for each physiological region of the body. We have made the assumption that these regions are countable and finite and that disease initiation is a function of NAi and NSi—the number of biological agent units (bacteria, virions, toxin units) deposited and the number of potential infection sites (particle deposited), respectively. While it would be convenient to use only NAi, there is currently insufficient scientific evidence to rule out the effect of NSi in disease causation. Terms in the equations that define NAi and NSi (Equations [8] and [9], respectively) include those in Box 3.3. In theory, a detector can measure or estimate each of the terms. In practice, the detector is used to estimate specific parameters that can be extrapolated through the equation as an estimate of hazard. For example, a biological point detector can use an aerosol collector and antibody or nucleic acid identifier to estimate such actions as inhalation and uptake. The detector itself will not directly measure BAULADae, rather the size and pathogen distributions in the aerosol would be linked to the best available probability distributions for the likelihood of disease. A single

OCR for page 29
39 point detector can determine the minimum quantity of aerosol agent material present, but unless the measurement is made proximal to the source, the detector is likely to underestimate the quantity released due to dispersion and transformation. A network of point detectors or a remote sensing system could provide more data that might integrate with meteorological measurements and algorithms for estimating transport and transformation. BOX 3.3 Definition of variables in Equations 8 and 9 Dae Aerodynamic diameter (µm) t Time (min) Q(t) Volumetric inhalation rate (L/min) ν (Dae) Number of agent units of aerodynamic diameter Dae [0, inf) ν (Dae) Pν (Dae) Probability that a particle of aerodynamic diameter Dae contains agent units [0,1] Pi(Dae,Q) Probability that a particle with aerodynamic diameter will deposit in region i of the respiratory tract when inhaled at a rate Q where i=1: Extrathoracic region; i=2: Tracheobronchial region; i=3: Pulmonary region [0,1] ν,D PA( ae) Probability that a biological agent within a particle of size Dae is active (µm) n( Dae , x, y, z , t ) Differential particle size distribution as a function of aerodynamic diameter Dae, space and time ({number of particles} / {m3} / {µm}) In short, the detector is used to estimate the physical, temporal, and spatial parameters of the aerosol as well as provide some compositional properties or actual biological properties that are used to indicate whether the aerosol present could be hazardous or not. In a generalized perspective, the detection system is a stand-in for a human, or alternatively, a human could be regarded as a specific type of detector. In either case, detector characteristics will need to be accounted for in order to deduce the aerosol description as expressed in the equation terms described in the previous section. The mathematical description of the aerosol given above is in its most fundamental form: as an instantaneous snapshot of all space. All detection systems will have their own specific volumetric sampling rates and efficiencies as well as response times. In some cases the response times will be fast enough to follow the natural temporal variations of the aerosol size distribution and concentration so that the instrument results could be directly substituted into the formulas as presented. In other cases, the response times will be much longer and appropriate integrations of the formulas will be required to incorporate the instrument data. Finally, identification of the pathogen or toxin allows extrapolation from scientific foundations (not all of which exist today) to prediction of probable health risk. A great strength of the proposed framework is its adaptability to new developments in detector technology. As with new developments in the understanding of disease response in humans, BAULA has the flexibility to incorporate new detector capabilities into its framework. To demonstrate this flexibility, it is useful to anticipate the types of information that detectors will provide in the coming years. Box

OCR for page 29
40 3.4 presents four scenarios describing the data provided by a detector and the assumptions that must be made to determine BAULADae in each case. Note that in all the scenarios listed in Box 3.4, it would be possible to determine an appropriate value of BAULADae assuming “worst-case” probabilities until the medical research and detector technology could provide more specific information. BOX 3.4 Calculation of BAULADae with Available Data Scenario 1 Identified agent is present KNOWN: UNKNOWN: How much of the agent is active Particle size distribution LD50 Worst case scenario: All detected agent is active It can cause illness wherever it deposits Agent is extremely virulent (low dose causes illness) Calculate BAULADae assuming: 1) all detected agent is active; 2) all deposit sites have equal risk; and 3) any exposure at all is a health risk. Consequence: Detector will sound alert for minimum detectable concentrations of agent Scenario 2 Identified agent is present KNOWN: How much is needed to cause disease (e.g. LD50) UNKNOWN: How much of the agent is active Particle size distribution How much agent is needed to cause disease at a specific site of deposition

OCR for page 29
41 Worst case scenario: All detected agent is active It can cause illness wherever it deposits Calculate BAULADae assuming: 1) all agent is active; 2) all deposit sites have equal risk; and 3) correct LD50 value. Consequence: Detector will sound alert if agent concentration>LD50 (or other dose level, depending on specific scenario requirement) Scenario 3 Identified agent is present KNOWN: Particle size distribution How much is needed to cause any form of disease How much agent is needed to cause disease at a specific site of deposition UNKNOWN: How much of the agent is active Worst case scenario: All detected agent is active Calculate BAULADae assuming: 1) all detected agent is active; 2) correct risk of infection at sites where particles of the detected size range will deposit; and 3) correct LD50 value. Consequence: Detector will sound alert if agent concentration >LD50 AND particles are the right size to reach sites vulnerable to infection (or other dose level, depending on specific scenario requirement) Scenario 4 Identified agent is present KNOWN: How much of the agent is active Particle size distribution How much is needed to cause any form of disease How much agent is needed to cause disease at a specific site of deposition Calculate BAULADae assuming: 1) correct proportion of agent that is active; 2) correct risk of infection at sites where particles of the detected size range will deposit; and 3) correct LD50 value. Consequence: Detector will sound alert if concentration of ACTIVE agent >LD50 AND particles are the right size to reach sites vulnerable to infection (or other dose level, depending on specific scenario requirement) 3.5 COMPARISON AND CONVERSION OF BAULADae TO OTHER UNITS As shown in Figures 3.2, 3.3, and 3.4, the unit of measurement used to describe a biological aerosol reflects the method by which a sample was collected and analyzed. Once the particles in the sample have been separated into different size ranges, the means by which they are immobilized for further analysis becomes the critical determinant of the final unit of measurement. In the case of a bacteria-containing aerosol, if the particles are impacted onto the surface of an agar plate and the number of resulting colonies is determined, the aerosol concentration will be expressed as the number of particles containing viable bacteria in each liter of air (ACPLA). If no additional samples are examined by other methods, no further description of the biological content of the aerosol is possible.

OCR for page 29
42 FIGURE 3.2 Methods of analyzing a bacteria-containing aerosol. If the sample is collected by impaction and culturing onto the surface of an agar plate (B), the bacterial content of the aerosol will be expressed in ACPLA. Because the particles have not been disrupted, the colony forming units only represent a count of particles with agent present. If the sample is collected into liquid (A), the unit of measure will differ depending on the subsequent analysis. In this example, the colony forming units (Ai) can be used to determine BAULADae with a known LD50 for the agent. For genome equivalents (Aii), the measure includes viable and nonviable bacteria and fragments of bacterial DNA in the sample. Signal contribution from antigen detection (Aiii) comes from viable and nonviable bacteria and bacterial antigens in the sample. Quantitation in total colony-forming units (Ai) provides the information needed to assess health hazard because it includes a measure of activity.

OCR for page 29
43 FIGURE 3.3 Methods of analyzing a virus-containing aerosol. Impaction onto an agar plate (B) is specific to bacteria and is not appropriate for viral agents. The initial sample must be collected into liquid medium (A). The subsequent analysis will determine the final unit of measure. In this example, the plaque forming units (Ai) can be used to determine BAULADae with a known LD50 for the agent. For genome equivalents (Aii), the measure includes viable and nonviable viruses and fragments of viral DNA in the sample. Signal contribution from antigen detection (Aiii) comes from viable and nonviable viruses and viral antigens in the sample. Quantitation in total plaque-forming units (Ai) provides the information needed to assess health hazard because it includes a measure of activity.

OCR for page 29
44 FIGURE 3.4 Method of analyzing a toxin-containing aerosol. Impaction onto an agar plate (B) is specific to bacteria and is not appropriate for toxins. The initial sample must be collected into liquid medium (A). The subsequent analysis will determine the final unit of measure. Measurement in ACPLA provides the lower bound for estimates of the amount of viable bacteria delivered to the respiratory tract. As demonstrated in Figure 3.2, it correctly determines that dose only in the case of an aerosol of single bacteria or spores, and increasingly underestimates the delivered dose as the mean number of viable bacteria increases above one per particle. Because only bacteria are capable of replicating and producing colonies on an agar plate, this method fails to detect any viruses or toxins that may be present. ACPLA, therefore, cannot be used formally as their unit of measure. ACPLA also cannot be converted to units of mass/L, because those are only employed in quantifying certain toxins. A main alternative to collecting particles by agar impaction is collection into a liquid medium, using an impinger, wet-walled cyclone, or other device (Figure 3.2A). Because the particles may be partially or completely disrupted as they are driven into the liquid and interact with their new environment, examination of the resulting suspension no longer permits the characterization of the original aerosol in units of ACPLA, but instead allows the determination of the total number of viable bacteria per liter of air by standard culture methods. That result, expressed in colony-forming units (CFU), is directly related to health hazard, because it indicates the dose of bacteria delivered to the respiratory tract that can replicate and cause disease. Measurement of CFU/liter of air can therefore, be converted directly to BAULADae using known

OCR for page 29
45 LD50 values (in this example, one bacterium is assumed to cause death in 50 percent of an exposed population). Except in the case of an aerosol of single spores or vegetative bacteria, the number of units of biological activity will always exceed the number of agent-containing particles (ACPLA). The same liquid suspension can also be used as the substrate for determining the total number of bacterial genome equivalents or the total quantity of a particular bacterial antigen per unit volume of air. As shown in Figure 3.2(Aii, Aiii), the number of units measured will generally be greater than the number of biologically active (viable) bacteria present. It should also be noted that it is possible to underestimate the amount of viable bacteria present if the bacteria are nonculturable on artificial laboratory media. Should these bacteria still be capable of causing infection, a measure of CFU/L could result in an underestimation of, or even complete failure to detect, health hazard. Collection of aerosolized material into liquid medium also makes it possible to go beyond the bacterial content to characterize a number of other aspects of an aerosol. Standard culture methods can be employed to quantitate viable virus (Figure 3.3), while biological assays can be used to measure active toxin (Figure 3.4). Those results are expressed in PFU/L and toxin activity/L, respectively; and both can be used to determine BAULADae. In addition, the analysis of an aliquot of the collection medium by PCR can determine the total number of genome equivalents for a specific bacterial or viral species present in a unit volume of aerosol, and an immunologic test such as ELISA, can be used to measure the amount of specific bacterial, viral, or toxin antigen. However, these methods will almost always overestimate the value of BAULADae for the same aerosol because both active and inactive agent are detected. If a portion of the aerosolized material loses viability during release and transport, but its DNA is still intact, PCR will overestimate the amount of biologically active bacteria or viruses delivered to the respiratory tract. A similar result will be obtained when antigen derived from nonviable organisms remains in aerosolized particles and is measured by an immunologic assay. The measured number of genome equivalents or quantity of antigen per liter of air should, therefore, be seen as setting an upper limit to the total amount of live viruses or bacteria that the sampled aerosol could contain, while the number of viable viruses or bacteria provides a lower bound for the total number of genomes or amount of antigen that could be present. Any attempt to convert the results from one of these detection methods to another must recognize these limitations. Figure 3.4 shows the initial path for analyzing a sample of a toxin-containing aerosol. The initial sample must be collected into liquid medium. Testing may then be based on the biochemical characteristics of the material (mass spectroscopy), an antigen-based assay, or a direct measure of toxin activity, such as the mouse toxicity assay for botulinum toxin. The latter type of analysis will provide the most important information for determination of the predicted health effects of the aerosol.

OCR for page 29