Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
EXECUTIVE SUMMARY The National Research Council's Committee on Fire Toxicology was formed in December 1984 in the Board on Toxicology and Environmental Health Hazards (now the Board on Environmental Studies and Toxicology) in the Commission on Life Sciences. It was supported by a consortium of federal agencies (the Consumer Product Safety Commission, the Federal Aviation Administration, the Department of the Navy, and the Environmental Protection Agency) concerned with developing sound regulatory policy. The Committee's general task was to review the state of the art of combustion-product toxicity testing and fire hazard assessment. In addition, the Committee considered the relationship between the physiologic and behavioral end points currently used in combustion-product toxicity test systems and the performance capabilities of humans exposed to pyrolysis and combustion products. The Committee was also to evaluate fire hazard models (both available and in development), focusing on the use of toxicity as an input, and provide guidelines for their application. HAZARD ASSESSMENT VS . RI SK ASSESSMENT Risk is the product of an event's severity (i.e., its degree of hazard) and the probability that the event will occur. This report deals only with fire severity (i.e., fire hazard) and its quantification. The degree of fire hazard is a function of a number of factors, such as fuel load, building structure, ignition, and propagation of flames, but also including the amount and toxicity of the smoke, the exposure to the smoke before escape, and the 1
2 exposure to heat. The assessment of fire hazard concen- trates on combining the factors that affect the time available for escape (TAE). ASSESSMENT OF FIRE HAZARD Hazard assessment, whether based on full-scale simula- tions or on mathematical models, requires an array of quantitative information, such as: . The amount of material present. The amount of energy required to ignite the material and spread flame over its surface. . The mass loss and heat release rate of the material, both when it is burning alone and when it is exposed to known energy fluxes from external sources. · The toxic potency of its smoke, expressed in terms of concentration, such as lethal concentration, effective concentration, or lethal concentration-time product. · Similar information on whatever else is burning, in addition to the material of interest. . Ventilation in the fire environment. · The geometry and thermal characteristics of the compartment that contains the fire (the fire environment) Detection models calculate the size of a fire at the time the detector (of smoke or heat) is activated and therefore the extent to which smoke or heat has developed in the compartment that encloses or is adjacent to the detector. Fire growth and smoke transport models can be used to predict the buildup of heat and smoke, thereby permitting calculation of TAE. The most widely known is the Harvard fire model, which can predict the growth of a fire with time and the resulting buildup of smoke and heat in up to five interconnected compartments, all on one level. The model's outputs include the times of ignition of second or third objects, the rate of gas outflow from the compartment, and the concentrations of various species in the outflowing gas. The FAST model .
3 developed at the National Bureau of Standards embodies some of the characteristics of the Harvard code, but emphasizes prediction of movement of fire products. It is claimed to provide more "rugged" or ~robust" solu- tions--e.g., those with a smaller tendency to give grossly wrong answers or to fail to run to completion--for a wide variety of input parameters. Such models assume that each compartment divides into a hot upper zone and a cold lower zone, with no mixing--hence their frequent designation as "zone" models. In contrast, field models divide a com- partment into hundreds or thousands of zones in a three- dimensional array and can therefore predict fluid motion far more realistically than zone models can. Field models, however, require very powerful computers and are not yet practical for routine hazard analysis. Computer models now becoming available can, in prin- ciple, calculate the development of a fire in a compart- ment and the buildup of smoke at specific locations in it. Because toxicity data are relevant to the TAE, the fate of occupants of those locations cannot always be predicted unless the smoke toxicity is known: TAE can be compared with the time needed for escape or rescue (TNE) for a selected scenario. TNE in turn also depends on the ages and health of the occupants. Theoretically, non- lethal exposure to toxic fire products can affect the TNE (e.g., by impairing mental acuity and so hindering escape), so for some scenarios data on nonlethal effects would be more relevant than lethality data. The details of the computations vary with the scenario under consideration, but it should be clear that smoke toxicity data constitute only one ingredient. Toxicity data alone are insufficient for complete and accurate assessment of a fire hazard. The overall hypothesis of hazard assessment is that survival of any fire is likely if the TAE exceeds the TNE. TAE depends on how quickly the environment becomes untenable; this in turn is controlled by the material's flammability and smoke toxicity. TNE is largely inde- pendent of the material burning. Besides having some importance in predicting the outcome of a given scenario, TAE serves as a surrogate for a material's relative fire hazard in that scenario. A comparison of the TAES for a series of materials is
4 therefore considered to give an indication of their relative fire hazards under the selected conditions. THE TESTING OF COMBUSTION-PRODUCT TOXICITY At least two combustion-product test methods can be used to provide the toxicity data required for modeling hazard: the National Bureau of Standards (NBS) and University of Pittsburgh methods. These methods are relatively well documented and yield toxicity values in substantial agreement for most materials that have been tested with both methods. The Deutsche Industrie Norm. . , . tU1N) ~d 45b method, developed in Germany, probably would also he adequate, but has not been thoroughly evaluated in the United States. And it is to be expected that these bioassays will be improved and that others will be developed for specific uses (e.g., to measure effects on mental acuity). In general, the primary unit of toxicity is the LC50, which is the concentration of a toxicant that causes death in 50% of the exposed animals in a specified period. The L(Ct)50, a unit that combines concentration (of fire products) and time, where appropriate for integration into a numerical fire hazard model, would theoretically make more refined results possible. The few pathologic measures that have been used (e.g., lung weights and corneal opacity) have yielded only limited information on the biologic effects of exposure to fire products. No test providing data on incapacitation has yet been developed that is demonstrably more sensitive than the use of death as the end point, although for some fire scenarios accurate measurement of incapacitation or performance decrements could be important. If such a test is developed, one would wish to demonstrate that incorporation of an end point other than death into a fire hazard model improved the ability to assess hazard. In the NBS method, a quartz beaker is heated to above or below the autoignition temperature of a sample to be burned; the sample is then placed in the beaker. Gases are collected in an airtight chamber, where rats are exposed for 30 min. The test results (referred to as LC50s) are expressed as sample weight charged per chamber volume (mg/200 L). The animals are observed for
5 14 days after exposure, and the ~LC50" is based on the number of animals killed at each "concentration" (although these are not actually exposure concentrations or dosages, because the atmosphere in the exposure chamber is not characterized). The method has proved reliable and reproducible in both intralaboratory and interlaboratory tests with Douglas fir and various other materials. In the Pittsburgh method, the sample is placed in a furnace that is then heated at 20°C/min. Mice are exposed to the continuous smoke stream, which is diluted with chilled air, for 30 min. This method also generates reproducible results. Although it has not been subjected to as many interlaboratory tests as the NBS method, the data suggest good agreement among results from three laboratories for Douglas fir and two other materials. As is true of any small-scale test, these tests do not model a "real fire" accurately. If only the data per- tinent to mortality (LC50) are used in the estimation of hazard, both methods might be equally applicable. However, if mass loss rate and time to death were to be used, only the Pittsburgh method could provide this information. Although there might be exceptions to this generality, it appears on the basis of the limited comparative data available that the choice of one or the other method would not alter substantially the outcome of a fire hazard assessment. For purposes of predicting the fire hazard of different materials, the Committee believes that the required smoke toxicity data are currently best obtained with animal- exposure methods. However, chemical analysis of smoke might be useful in the process of measuring smoke toxi- city. The advantages of chemical tests are that many are quicker to perform than bioassays and that they avoid the use of test animals. The main advantage of biologic tests is that they produce data of high validity. The major potential danger of a chemical test is that it could "miss" unanticipated, and perhaps unusually toxic, com- bustion products (although unusually toxic combustion products whose formation was not predicted by chemistry have rarely been encountered); there is little danger of missing biologically relevant response in a bioassay. In addition, most current fire hazard models are designed to accommodate toxicity data in the form of LC50 or L(Ct)50 values; the use of chemical data alone in such a model
6 would require development and verification of an appro- priate scheme to summarize and add the various measured concentrations. A toxicity-testing strategy that avoids the uncer- tainties of a chemical analysis while exploiting its advantages could have the following steps: . Chemically analyze the test material's smoke for expected major toxicants, such as carbon monoxide, hydrogen cyanide, and hydrogen chloride. · Calculate an "expected" LC50 for the smoke, on the basis of the response of test animals to the toxicants identified in the chemical analysis. . Perform a bioassay of the material's smoke at, slightly above, and slightly below the expected LC50. If all the important toxicants have been identified in the chemical analysis, this test should be sufficient to confirm that identification and to yield an approximate LC50. If the observed LC50 is very different from the expected LC50, the difference will be apparent, and more extensive bioassays must be carried out. Beyond LC50 data, the routine measurement of carbon monoxide in smoke or of carboxyhemoglobin in the blood of exposed animals, however useful such measures are for research purposes, provides no information of utility to hazard assessment efforts that is not provided with more certainty by the LC50 itself. CONCLUSIONS AND RECOMMENDATIONS There is a strong need for additional research in combustion-product toxicity testing and fire hazard assessment. Indeed, knowledge in these fields is still quite incomplete. Thus, the approaches embodied in the following conclusions and recommendations should be viewed as being of an interim nature. The issues should be reviewed again in 5 years and the recommendations revised in the light of new knowledge. No model or test method comes close to reproducing the peril in which fire places human life. The results of mathematical models are only as good as the data used in
them, and those data, whether they are abstractions of smoke movement and ventilation or toxicity values deter- mined in rodents, are only representations of the reality that takes the lives of some 5,000 people a year in this country. Although efforts to reduce the likelihood of fire must continue, we must assume that there will always be fires. We must therefore work to improve our ability to cope with fire by improving the fire performance of the materials that furnish the places in which we live and work, while continuing to improve building designs, fire codes, and fire detection and suppression techniques. · Although mung the highest in the world, the number of fire fatalities in the United States has been declining for the last 30 years. The concern that the introduction of synthetic mater- ials into general use in homes has increased the risk associated with fire is not supported by data on recent fire-death trends. New materials are being used in homes and commercial buildings, and these materials have dif- ferent combustion characteristics, but the number or rare fatalities per 100,000 people has declined. This decline cannot be fully explained by improvements in firefighting equipment, sprinkler systems, or detectors, although they _ _ ~ are relevant. · The best-characterized threats associated with fire are the acute results of exposure to heat, the toxic agents and irritants that make up smoke, and perhaps oxygen depletion. Long-term effects of repeated exposure, such as would be encountered by firefighters, have not been conclusively characterized. Smoke inhalation is the cause of death in the majority of fire fatalities. The toxic components of smoke are largely carbon monoxide and other gases, such as hydrogen cyanide. Carbon monoxide is well accepted as a factor in 50-80% of all fire fatalities; the role of hydrogen cyanide is still under investigation. Other components, such as respiratory and sensory irritants, might con- tribute to the inability of people to escape from fire, as well as to long-term pulmonary complications in survivors. The cause of death of many fire victims is not fully understood. Even less is understood about the potential long-term health consequences in survivors of single exposures to fire. The influence of the type of
8 materials involved in a fire on mortality has not been established. To determine cause of death, postmortem study of fire victims would include both autopsy and blood-gas investigations (for carbon monoxide, hydrogen cyanide, etc.). Prospective studies of pulmonary and necrologic function in both single-exposure fire victims and occupationally exposed firefighters are needed to evaluate long-term health consequences. There seems to be no conclusive evidence that the long-term effects of repeated exposure to fire include an increased risk of developing cancer, although some groups of firefighters have been found to have excesses of some cancers. Finally, research on chemical and cellular biologic markers of combustion-product toxicity (e.g., with bronchoalveolar ravage) should continue, inasmuch as such markers might provide early indicators of pathologic effects of smoke toxicity. · The dynamics of a fire in generating heat and toxic products will determine the ability of people to escape; the use of fire hazard assessment to estia te ability to escape a given fire is currently the best approach to measuring the hazard associated with materials. Determination of the likelihood of escape from a burning building requires evaluation of the time avail- able for escape and the time needed for escape. TAE can be calculated from the time at which the fire is detected, the temperature and the quantity of smoke, and the growth curve of the fire or smoke. TAE is not to be used at face value as a measurement of real time, but as a tool for comparing materials under some set of conditions. a rapidly growing fire, temperature can increase so rapidly that the toxicity of smoke is irrelevant. In a fire that is growing slowly or if the potential victim is not in the same room as the origin of the fire, the toxicity of smoke might be the prime source of danger. In TNE is calculated from factors associated with the potential victim's ability to find a safe escape route, the nature of the building, the victim's age and related characteristics, etc. · If a combustion-product toxicity test is to be useful in a hazard model, it should have good inter-
9 laboratory reproducibility, should differentiate between materials on the basis of relative toxicities, should be able to determine dose-response relationships for a given material, and should yield data in units that are co" patible with hazard models. Just as there is no single set of fire conditions, there is no "correct" set of toxicity-test parameters. Different tests can be expected to yield different rankings of the same group of materials. This failure of agreement is unlikely to be resolved by additional research and test development; and total agreement might not even be desirable, in that use of a single "standard" test could lead to unnecessarily restricted sampling of combustion conditions. · Laboratory metbDds for measuring the toxicity of combustion products have been developed to the point where relative toxicities of materials can be reliably measured. Both the NBS and the Pittsburgh test methods provide a comparison of relative toxicities; when used by different laboratories, each has reasonable reproducibility. Neither method provides a complete model of a real fire, but each provides data on some aspect of combustion. Data from these methods can be used in hazard models. Currently used toxicity test methods use lethality as the end point; other end points remain to be remelted. The use of death as the end point provides a reliable index of smoke toxicity, but fails to provide information on the inability of people to escape fires. This inabil- ity could be due to sublethal exposure to toxic gases, such as carbon monoxide, or to the effects of irritants in impeding escape. No animal model of sublethal effects has been found more useful than measures of lethality in providing the desired information. A sensitive measure of sublethal effects would theoretically improve the ability to assess hazard; at the least, it would allow for a more logically consistent representation of ability to escape. In order to understand the effects of combustion products on mental acuity, specific tests of impairment
10 of judgment or performance in animals and in humans should be developed. The results of such tests could be used in a fire hazard evaluation in various ways, e.g., with respect to toxicity or human performance. . As a basis for judging or regulating materials' performance in a fire, co~bustionrproduct toxicity data must be used only within the context of fire hazard assessment. In determining overall hazard to people, toxicity data obtained from animal experiments should be incorporated into a fire hazard model. Given all the other factors that are relevant to fire hazard, such as rate of burning and heat generation, toxicity cannot be the sole criterion for defining the hazard. If products for some intended use have been shown to be very similar in composition and other fire properties, a pass/fail decision that depends on a toxicity test could be justified. Although this appears to be a screening test, it is in fact simply the final point of discrimination in a less formal hazard analysis. For uses with no regulatory component te.g., manufacturer's surveillance of products under develop- ment), any chosen test can be used for screening, with specific performance criteria set by the user. . Because of the possibility that new toxic chemicals will not be detected in chemical tests of combustion products, biologic tests Bust remain the ultimate toxicity assays. Chemical tests can be extremely useful in measuring concentrations of known chemicals in combustion products and thus might become a first screen for testing toxicity. An animal biologic model acts as the ultimate integrator of the combined toxicities of combustion products, whereas a chemical assay is a selective measure of specific chemicals and might or might not detect all toxic agents. Therefore, animal tests must remain as the final deter- minants of human hazard. This necessity could become even more evident as nonlethal measures of toxicity become available. . Although techniques will continue to improve, fire hazard assessment can already be used to answer fundamental questions about the suitability of materials.
11 Computational models require a knowledge of the burning rate of materials and critical concentrations of toxic combustion products, if TAE is to be calculated. TAE is used as a surrogate of fire hazard and makes it possible to compare relative fire performance of materials in a given application. Powerful computers and increasingly detailed analysis of more complex spaces will lead to better understanding of the dynamics of fires and to more realistic approximation of TAE. In the meantime, however, many regulatory questions can already be answered.
