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 1
Assessment of Exposure-Response Functions for Rocket-Emission Toxicants SUMMARY WHEN deciding to launch a rocket under prevailing weather conditions, commanders at Vandenberg Air Force Base (VAFB) in California and at Cape Canaveral Air Station (CCAS) in Florida must evaluate the possibility that toxic concentrations of wind-blown rocket emissions might reach military or civilian populations. To assist commanders in estimating the risk of such exposures, the Air Force is developing the Launch Area Toxic Risk Analysis (LATRA) model. It contains two major components: (1) a dispersion model that predicts downwind exposure concentrations and (2) exposure-response functions (ERFs) that relate the estimated exposure concentrations to expected health effects. In 1995, the Air Force Air Space Command asked the National Research Council (NRC) for an independent review of the ERFs in LATRA. The NRC was asked to focus on the toxicity of the three major rocket emissions—hydrogen chloride (HCl), nitrogen dioxide (NO2), and nitric acid (HNO3)—and several characteristics of LATRA-ERFs, including the identification of sensitive populations; definition of severity of effects; selection of independent variables in each exposure-response model; choice of appropriate analytic form for the ERFs (e.g., lognormal or probit); quantification of ERFs for each of the emissions; and representation and propagation of uncertainties associated with the LATRA-ERF model. The NRC assigned this project to the Committee on Toxicology (COT), which convened the Subcommittee on Rocket-Emission Toxicants to respond to the request. Subcommittee members were chosen for their expertise in inhalation toxicology, pharmacology, biostatistics, risk assessment, and environmental health, and they worked with-
OCR for page 2
Assessment of Exposure-Response Functions for Rocket-Emission Toxicants compensation in national service, as do all NRC committee members. This report presents the subcommittee's evaluations, conclusions, and recommendations. DESCRIPTION OF THE LATRA MODEL The LATRA model is designed to estimate the probabilities of mild and serious health effects from exposing specified human subpopulations to estimated concentrations of specific rocket emissions. For each emission, exposure-response functions (ERFs) were developed to relate estimated exposure concentrations to expected health effects. At present, separate ERFs are derived for "sensitive" and "normal" segments of the general population. An ERF is specified by two points: a lower concentration assumed to be associated with a 1% incidence of a particular effect, and an upper concentration assumed to be associated with a 99% incidence of a particular effect. An ERF is fit to the two concentration versus-incidence points using a log-probit model (equivalent to assuming a lognormal distribution of the probability of effect). The resulting curve is then used to calculate the expected health effects and the risk profile for each population subgroup. A different procedure would be used to establish ERFs for carcinogenic emissions; however, the ERFs currently included in LATRA are not for substances known or suspected to be carcinogenic. To set the 1% effect levels for sensitive populations, the Air Force considered the National Research Council's short-term public emergency guidance levels (SPEGLs) and other published exposure concentrations estimated to be safe for exposures of the general public. In establishing the 1% effect levels for normal populations, the Air Force considered exposure concentrations independently estimated to be safe for workers. The 99% effect levels were set 5-fold higher than the 1% effect levels for sensitive populations and 10-fold higher than the 1% effect levels for normal populations to reflect the assumed greater range in sensitivity among members of the normal population. The subgroups considered sensitive to the rocket emissions modeled by LATRA are children (less than 15 years of age), the elderly (more than 64 years of age), and all persons with bronchitis, asthma, or other physiological stress, especially upper-respiratory ailments. The remainder of the population is considered normal.
OCR for page 3
Assessment of Exposure-Response Functions for Rocket-Emission Toxicants In the LATRA model, a mild effect is defined as temporary irritation with no organic damage, and a serious (or severe) effect is defined as organic damage requiring medical treatment. The LATRA model operates as a Monte Carlo simulation. A binomial model is used to simulate the variance (uncertainty) associated with the predicted number of people affected. The potential for combined effects of exposure to more than one compound is estimated by developing joint probabilities of effect from the individual toxicants' probabilities of effect. The LATRA model estimates of the total number of people at risk of health effects from a launch are based on (1) the risks associated with a normal launch, (2) the probability of a normal launch, (3) the risks associated with a catastrophic abort, and (4) the probability of a catastrophic abort. THE SUBCOMMITTEE'S CONCLUSIONS AND RECOMMENDATIONS In general, the subcommittee found the basic premise of the LATRA model—using exposure-versus-incidence-of-response models to predict the incidence of effects in humans—to be reasonable, but the available toxicological data on the specified rocket-emission toxicants are currently insufficient to support the ERFs used in the LATRA model. The subcommittee's specific conclusions and recommendations with respect to the toxicological components of LATRA, and possible alternative approaches recommended by the subcommittee, are described below. TOXICOLOGICAL DATA BASE The toxicity data available for HCl, NO2, and HNO3 are sufficient to identify no-observed-effect levels (NOELs) for humans and to indicate varying differences in sensitivity at low exposure concentrations between individuals with asthma and healthy individuals. The available data also are sufficient to estimate thresholds for mild, moderate, and severe effects for HCl and NO2, but not for HNO3. However, the only exposure-response data useful for predicting the proportion of individuals that might be affected by exposure to those compounds appear to be the data on mortality and severe effects in animals exposed to HCl and
OCR for page 4
Assessment of Exposure-Response Functions for Rocket-Emission Toxicants the data on mortality in animals exposed to NO2. Thus, the subcommittee found that the toxicity data on HCl, NO2, and HNO3 are insufficient to support the development of separate ERFs for mild and serious effects in sensitive and normal human populations. IDENTIFICATION OF SENSITIVE SUBGROUPS The subcommittee recognizes that interindividual differences in toxicological responses to chemical exposure are a major area of public-health concern. The causes of these differences include age, sex, genetic background, nutritional status, pre-existing diseases, and life style. The subcommittee does not believe that age (e.g., individuals over 64 years of age) should be the principal attribute to identify a segment of the sensitive population. The subcommittee believes that a more accurate assessment of the number of potentially sensitive individuals in the population near the launch sites can be obtained by basing sensitivity on the estimated prevalence of health conditions likely to render a person sensitive rather than by basing sensitivity on indirect measurements such as age. For adults, information on the age-specific incidence of diseases likely to increase individuals' sensitivity to rocket-emission toxicants can be used with information on the ages of the exposed individuals. Although sensitivity within the adult subpopulation might be due to the presence of certain diseases rather than to age, children might indeed be a potentially susceptible population, even when healthy. That potential could be due to such factors as differences in ventilation rate in children compared with healthy adults. The National Health Interview Survey (NHIS), available from the U.S. Bureau of Census, could be used to obtain information on age-specific disease incidence. The subcommittee recognizes that children might represent a potential susceptible population, even when healthy. The LATRA model includes separate ERFs for sensitive and normal populations. The subcommittee endorses explicit consideration of potentially sensitive subgroups; however, as mentioned above, it found the toxicity data available for the rocket-emission toxicants inadequate to define separate ERFs for the two subgroups. Available data support only the derivation of different thresholds of effect in sensitive and normal individuals. The toxicity information available for the three rocket emissions indicates that for short-duration exposures (i.e., 1 hr or
OCR for page 5
Assessment of Exposure-Response Functions for Rocket-Emission Toxicants less), sensitive individuals begin to respond at lower concentrations than normal individuals by a factor of 10 for NO2, a factor of 3 for HCl, and a factor of 20 for HNO3. DEFINITION OF SEVERITY OF EFFECTS The Air Force asked the subcommittee to consider how best to define three categories of severity of effects: mild, moderate, and serious. The subcommittee believes that categorizing specific effects into such severity categories is an acceptable approach. The subcommittee defined mild, moderate, and serious effects as follows. Mild effects are reversible within 48 hr and do not interfere with normal activity or require medical attention. Moderate effects are irreversible effects that do not alter organ function or interfere with normal activity, or they are reversible effects that alter organ function or interfere with normal activity. Persons experiencing moderate effects might seek medical attention. Severe effects are irreversible effects that alter organ function or interfere with normal activities. Severe effects usually require medical attention. Those definitions are specific for exposures to rocket emissions and might not be applicable to other exposure scenarios or toxicants. STRUCTURE OF THE LATRA-ERF MODEL In principle, the LATRA-ERF model is a valid concept, but the subcommittee does not endorse use of the LATRA-ERF model as it is currently constructed. The ERFs give the appearance of substantial accuracy; yet, they are not adequately supported by toxicological information. Consequently, a user of the LATRA-ERF model might believe that the model is more reliable than it actually is for estimating risk. In the interim, the subcommittee instead recommends that the hazard-quotient approach be used to characterize risks for sensitive and normal populations, as described below. However, if the Air Force wants to pursue the LATRA-ERF model, there are ways to improve components of the model, as described below. The ERFs in the LATRA model are currently based on 1-hr time-weighted-average concentrations and ceiling values. The subcommittee believes that 1 hr is too long because of the typical speed with which the
OCR for page 6
Assessment of Exposure-Response Functions for Rocket-Emission Toxicants ground cloud of emissions from a rocket launch passes over a given exposure location; increments of 10 to 30 min would be more representative of the exposure situation, covering the total duration of exposure. The LATRA model is capable of incrementing exposures. The subcommittee endorses the use of ceiling values for noncumulative effects. It also identified certain effects for which the product of exposure concentration and time (C × T) would be appropriate: for example, severe effects and mortality for HCl and NO2. For effects for which the relationship between effect and the product of C × T is unknown, the subcommittee recommends that sensitivity studies be conducted to determine how the selection of the independent variables for the ERF influences the LATRA model's output. If time-weighted-average concentrations for 10 and 30 min are used, for example, those results should be compared with the results of using C × T as the independent variable. A weakness of the current derivation of LATRA ERFs is that the dose-response model for predicting incidence (a log-probit model) is based on health-protective or "safe" levels that have no specified relation to the incidence of effects. The subcommittee does not believe that it is appropriate to interpret a safe level as a 1% incidence level for mild effects. The true incidence could be higher but is likely to be lower and presumably near or at zero. An accurate ERF could be used to predict the incidence below the 1% level. The subcommittee also does not believe that it is necessarily appropriate, in the absence of supporting data, to interpret a concentration that is 5- or 10-fold higher than that causing a 1% incidence level as the concentration at which all individuals are likely to show effects. That level provides a relatively steep ERF that might be conservative above a 1% incidence level, but might not be appropriate for the more likely scenario of exposure below a 1% incidence level. Combining the 1% and 99% incidence-versus-exposure values to construct a model for predicting the incidence is a judgmental process that lacks any direct measurements from either epidemiological or toxicological data. To the extent possible, the Air Force should use end-point-specific incidence data to develop end-point-specific ERFs. However, with the exception of mortality and a few other end points, incidence data for HCl, NO2, and HNO3 are not available. Without incidence data on humans or animals, it is difficult to endorse exposure-response models that predict incidence. Until end-point-specific data become available for
OCR for page 7
Assessment of Exposure-Response Functions for Rocket-Emission Toxicants HCl, NO2, and HNO3, the Air Force could attempt to validate the model against other compounds, such as chlorine gas and ammonia, that are likely to have adequate data on humans. The subcommittee recommends that the Air Force generate appropriate toxicity data to calibrate and validate the proposed model. The investments in appropriate testing procedures, at this time, would be worth the effort by improving the model's predictibility and reducing the uncertainty. Such studies should, at a minimum, examine concentration times, time responses, and include adequate histopathology. Appropriate toxicity data will allow the Air Force to calibrate its model on the basis of sound data. Until more data become available or an expert-elicitation process can be carried out to estimate incidence for end points with no incidence data, the subcommittee believes that a hazard-quotient model would be more appropriate. For the hazard-quotient model, estimates of the number of people at risk would be based on the number of people with exposures above a reference exposure level that is unlikely to cause adverse health effects. The ratio of the exposure concentrations to reference exposure levels might also be useful. The hazard-quotient model could be used to estimate how many people might be at risk of moderate or severe effects if the Air Force is willing to accept the level of uncertainty associated with exposure values identified as a threshold exposure for moderate and severe effects. Under the LATRA model, separate ERFs are developed for sensitive and normal populations. However, a properly constructed probit model can portray a wide variation in human sensitivity within a single exposure-response function. Moreover, available data are insufficient to quantify different ERFs for sensitive and normal populations. Thus, the subcommittee cannot support the use of specific ERFs for sensitive and normal populations, although it does support the use of different thresholds for effect if the hazard-quotient approach is used to characterize risk. In the absence of incidence data to construct ERFs for sensitive subgroups, the hazard-quotient approach could be used to characterize risk. When deriving a hazard quotient, the common practice is to use animal or human data to define a low-or no-effect level. That level is then divided by an appropriate uncertainty factor to yield an allowable exposure level. The hazard quotient is the ratio of an observed or pre-
OCR for page 8
Assessment of Exposure-Response Functions for Rocket-Emission Toxicants dicted exposure to an allowable exposure. The allowable exposure level would be set at a lower value by selecting an uncertainty factor that is sufficient to protect sensitive individuals. To avoid embedding value judgments in the scientific exposure-response analysis, ERFs should be developed first by health end point. After considering all the end points, decisions can then be made on which exposure concentrations to associate with mild, moderate, and severe effects. Incidence dose-response data are lacking for all but severe end points for HCl and NO2 and are altogether lacking for HNO3. It is possible, however, to estimate a reference exposure that is unlikely to cause mild effects for all three of the rocket emissions. In addition, reference exposures for moderate and severe effects can be estimated for HCl and NO2 (see Appendices D and E), although there are large uncertainties concerning the time-dependence of those estimates. Those reference exposures could be used with the hazard-quotient model. The subcommittee suggests that the Air Force be especially aware to avoid making certain value judgments based on an incomplete or limited data base. Such limitations make it difficult to evaluate or predict accurately the degree to which a specific human subpopulation might be more sensitive to air contaminants than others. The subcommittee does not recommend using the binomial model in LATRA to address uncertainty. The binomial model generates a variance that underestimates the variance associated with fitting the ERF to response data. If an adequate data base becomes available to support the development of ERFs, sensitivity analyses should be conducted to investigate the assumptions and procedures used to construct the ERFs. The subcommittee recommends that the Air Force evaluate potential health effects resulting from simultaneous exposure to more than one toxic rocket emission, assuming the potential for additive effects. That could be accomplished by using the hazard-index approach (i.e., adding the hazard quotients for individual chemicals) to characterize risk. Given the complex nature and extent and pattern of injury in the respiratory tract from exposure to airborne chemicals, it is important to understand interspecies differences in their response to inhaled substances. The ability to make interspecies dosimetric comparisons is critically important for judging the applicability of various toxicological results to human exposure conditions. Selected dosimetric experiments
OCR for page 9
Assessment of Exposure-Response Functions for Rocket-Emission Toxicants involving laboratory animals and humans can provide valuable data on the variability in uptake according to species and the specific region within the respiratory tract where the chemical might target. New experimental dosimetric approaches, such as those involving isotope ratio mass spectroscopy and cyclotron generation of gases, offer promise for improving the ability to make scientifically defensible predictions. The subcommittee recommends that the Air Force consider including interspecies dosimetric correction factors when applicable. Instead of presenting one risk estimate for a launch that combines the risks of a normal launch and a catastrophic abort, the subcommittee believes that it would be more appropriate for the Air Force to present separate risks for normal and aborted launches or to provide separate conditional risks and combined risks. The Air Force should ensure that any time-weighted-average exposure estimate used to determine risk is the maximum value possible. For example, the maximum 30-min time-weighted-average concentration passing over an exposure location should be compared with a 30min ERF or a reference exposure unlikely to cause an adverse health effect. The subcommittee also recommends that the Air Force evaluate the relative accuracies of the exposure estimates from the rocket-exhaust dispersion model and the estimates of incidence of effects from the ERFs (or reference exposures as suggested here). If the Air Force can determine whether the exposure component or the effects component of the LATRA model is the more serious limit to the model's accuracy in predicting risk, it can invest effort in improving the less accurate component. In summary, the LATRA-ERF model is a valid concept, but the current lack of toxicological data makes its implementation problematic. Some specific deficiencies have been noted above by the subcommittee, and some improvements in the LATRA-ERF model might be possible. In the interim, the subcommittee suggests that a hazard-quotient hazard-index approach be considered as a possible alternative. This approach would allow an estimate of the number of people exceeding a reference exposure level below which health effects are unlikely to occur. This approach would not attempt to estimate the incidence of health effects in an exposed population.
Representative terms from entire chapter: