SAFETY
Ammonia under the safety spotlight again
In the US, EPA-proposed Acute Exposure Guideline Levels (AEGLs)
for ammonia have become a bone of contention between industry and government. In this article, Robert A. Michaels, PhD, CEP* examines the issues, and looks at the direction the debate may take.
catastrophically released from a
chemical plant in Bhopal, India. The release killed 2,000 local residents and caused irreversible injuries to the eyes and lungs of approximately 20,000 others. Governments around the world recognized the need to identify extremely hazardous substances (EHSs) and assist communities to prepare for chemical emergencies®. In the US, the Environmental Protection Agency (EPA) identified 366 EHSs, as required by the Community RightTo-Know Act in the 1986 Superfund Amendments and Reauthorization Act. Community emergency planning requirements were imposed under the Risk Management Program (RMP), in the 1990 Clean Air Act Amendments, Section 112(7).
Some nitrogen-containing chemicals such as ammonia are included among identified EHSs. This means that nitrogen industry facilities in the US; such as ammonia production, fertilizer production, and ammonia-for-refrigeration facilities; are among an EPA-estimated 66,000 industrial facilities potentially subject to the RMP. Their foreign counterparts can expect to be included in similar programs likely to emerge globally.
[: 1984; methyl isocyanate was
*Dr. Michaels is President of the RAMTRAC Corporation (3100 Rosendale Rd; Schenectady, NY 12309 USA; telephone: 518/785-0976; e-mail: 76517. 747 @compuserve.com).
July-August 1997
Emergency planning zones
Industrial facilities exceeding threshold quantities of EHSs on site are subject to the RMP. The RMP defines Emergency Planning Zones around facilities, and requires facilities to submit for EPA approval a risk management plan aimed at preventing catastrophic chemical releases from occurring, and averting Bhopal-type consequences of releases that do occur. The cost of community emergency planning for a facility is likely to depend roughly upon the area of its Emergency Planning Zone, which in turn depends upon the square of the radius around the facility required to attenuate a worst-case chemical release to specified airborne concentrations. The Committee on Toxicology of the National Research Council termed these airborne concentrations Community Emergency Exposure Levels (CEELs)®, but the term has been replaced by Acute Exposure Guideline Levels (AEGLs). AEGLs for each EHS are under development by the National Advisory Committee on AEGLs (NAC AEGL), with technical support documents provided by a contractor, currently the Oak Ridge National Laboratory? (ORNL).
Acute exposure guidelines
Three types of AEGLs were defined by NRC’s Committee on Toxicology®. AEGL-1 is defined to protect against nuisance exposure;
AEGL-2 against irreversible injury; and AEGL-3 against death. Each type of AEGL is defined by multiple combinations of airborne concentration and exposure duration to protect against the adverse health effects of concern over a range of accidental release scenarios potentially lasting from minutes to hours.
Ammonia AEGL proposals
Ammonia was among the first EHSs considered when NAC AEGL began meeting in June 1996. Ammonia remained on NAC AEGL’s Agenda even at its fifth (quarterly) meeting, on 9-11 June 1997, though it had been removed from the March agenda because the committee and its contractor, ORNL, required more time to respond to comments submitted by RAM TRAC*!2 and others?. Comments on AEGL-1, -2, and -3 values derived in ORNL’s draft technical support documents were almost all negative, criticizing ORNL and NAC AEGL for overestimating the toxicological potency of ammonia’. However, comments notwithstanding, NAC AEGL decided in June to formally propose very low AEGL concentrations for ammonia (See Table 1) corresponding to exposure durations of five minutes, 30 minutes, one hour, four hours, and eight hours. These AEGL values will be subject to two stages of review. The first review will be undertaken by NAC AEGL itself, in re-
NITROGEN No. 228
SAFETY
sponse to public comments which may be submitted about its Notice of Proposed Rule Making, which will be published in the Federal Register. The second review will be external, undertaken by the NRC’s Committee on Toxicology.
Objections
AEGL values to be proposed by NAC AEGL are a patchwork of conflicts with AEGL definitions and procedures, and inconsistencies with available data. A day after setting ammonia AEGL values, the Committee evaluated itself. Members suggested producing a document setting forth its standard operating procedures (SOPs). A member suggested that SOPs should be “artfully constructed to be defensible in court, and soon.” Another indicated that the Committee had not even “done lip service” to its SOPs or definitions, and that its future actions would have to be more defensible. One suggested that SOPs “should not specify AEGL derivation in a strict way” and, similarly, another suggested that SOPs be written in a sufficiently flexible manner so the Committee would not have to follow them, “to permit consensus to evolve.”
Ammonia AEGL-2 and AEGL -3 values to be proposed in the Federal Register are roughly four to five times lower than they should be. Given the dependency of Emergency Planning Zone area on the square of the emergency planning radius, rather than on just the radius, the cost of emergency planning may be 16 to 25 times higher than necessary. Further, inappropriate inclusion of residents and residences within Emergency Planning Zones can be expected inevitably to increase public opposition to permitting new facilities and re-permitting existing ones, as people become concerned and possibly alarmed — in many if not most cases unnecessarily — about the safety of their families and value of their property. Specific concerns about AEGL values to be proposed by NAC AEGL are addressed as follows.
NITROGEN No. 228
Time: 5 min 1 hour 4 hours 8 hours NAC AEGL Proposals
AEGL-1 25 25 25 25, 25 AEGL-2 380 160 110 110 110 AEGL-3 3,800 1,600 1,100 550 390 RAM TRAC Recommendations
AEGL-2 1,704 696 492 246 174 AEGL-3 16,869 6,887 4,870 2,435 Abyss
AEGL-3 values: NAC AEGL
violates AEGL-3 definition NAC AEGL and NRC define the AEGL-3 as the airborne concentration which could cause members of the general population, including ‘susceptible’ but excluding ‘hypersusceptible’ individuals, to experience life-threatening effects or death®’. However, NAC AEGL? translates this mandate to a reductio ad absurdum, stating the following policy with respect to ammonia: “No verified lethal concentrations for ammonia were found in the available literature. However, Silverman, et al. (1949) reported that 1,000 ppm induced an immediate urge to cough. Legters (1980) noted that coughing may indicate that the absorptive (scrubbing) capacity of the upper respiratory tract has been exceeded and that ammonia is penetrating to the lower respiratory _ passages... Therefore, concentrations of ammonia that exceed the scrubbing capacity of the upper respiratory tract and cause coughing, which indicates lower respiratory effects, have lethal potential”.
NAC AEGL inappropriately rejects air modelling
Early ammonia lethality reports (for example; 10,000 ppm for five-to 10-minute exposures) apparently have their origin at a time predating World War I, when ammonia was used as a war gas. Its producers competed to sell ammonia to their respective military forces, which had an interest in gases demonstrated to exhibit high toxic potency against people engaged in trench warfare. These
reports are cited by ORNL’ for lethality range-finding. Such suspect ‘data’ should have been accorded a low threshold for refutation, whereas NAC AEGL gave it undue credence relative to more meritorious studies, described below, involving industrial accidents
Several relatively recent industrial accidents involving ammonia release, refuting the early values, were available for AEGL-3 derivation*5, but inappropriately rejected for use by NAC AEGL?’. A tank failure in Potchefstroom, South Africa killed 18 people, but demonstrated survival of all outdoors who were exposed to ammonia at 33,737 ppm-v or higher for five minutes. A truck accident in Houston, Texas revealed similar results. An accident involving a front end loader (‘Bobcat’) breaching an overhead anhydrous ammonia tank in Bartow, Florida revealed survival of an individual who briefly inhaled ammonia close to its release point (estimated five-minute exposure concentration: 89,400 ppm-v)?. An incident in which an individual refilled a tank of 25 % ammonia solution while failing to wear respiratory protection revealed sporadic exposure to a high fraction of the concentration of saturated vapour at 330,000 ppm-v displaced from the tank being refilled®. The individual died of heart failure secondary to lung injury. His death six hours after the exposure incident was apparently unnecessary, given his failure to seek medical attention, and his return to work for three hours after the incident.
‘The Emergency Planning Zone radius around an industrial facility depends upon applicable AEGL
July-August 1997
SAFETY
values coupled with air modelling to quantify the distance required to attenuate a worst-case release to within AEGL concentrations. Thus, air modelling is central to the RMP, based upon the implicit assumption that modeled airborne concentrations and associated exposure durations accurately quantify worst-case exposure levels. Accordingly, air modelling has been applied to reconstruct airborne concentrations and exposure durations in accidental ammonia releases?. The premise of such reconstructions is that air modelling is a valid method for this purpose and, indeed, ORNL cited accident investigations using the WHAZAN model to reconstruct ammonia releases in Houston, Texas and Potchefstroom, South Africa to support early AEGL-3 value derivations.
RAM TRAC upwardly adjusted ORNL/’s concentrations’ estimated. via the decade-old WHAZAN model by applying results produced by the state-of-the-art HGSYSTEM model, which more effectively simulates the transition of cold, heavier-than-air, cryogenically released ammonia as it equilibrates with ambient air temperature and rises out of the breathing zone. When the air modelling, now revised, associated observed mortality rates with higher ammonia concentrations than those originally assumed by ORNL, NAC AEGL hired an air modelling consultant, who rejected air modelling as a basis for determining AEGL values for ammonia: “Atmospheric concentrations cannot serve as a surrogate for exposure, and therefore should not be used to derive AEGL values”’.
The above finding, adopted by NAC AEGL as a conclusion, seems self-serving. It is inappropriate in view of the centrality of air modelling to facility permitting generally, and to the RMP specifically. To address the valid concern that modeled air levels are spatially homogeneous in an accident zone, whereas actual air levels following a release are inevitably heterogeneous, RAM TRAC applied an uncertainty factor of two to compensate for ‘benign bubbles’. Ironically, ORNL’ termed this
July-August 1997
uncertainty factor the benign bubble ‘hypothesis’, claiming that it lacked scientific proof, and should be rejected!
Accident reconstruction represents the only source of human lethality data involving very short durations of exposure to ammonia. Rejection of ammonia accident reconstruction data is thus also ironic in the context of NRC’s mandate to NAC AEGL: “workplace or ambient air quality ... limits are not easily or directly translated into the kind of limits required for emergency exposures, which typically involve exposure at high levels but of short duration, usually less than one hour, and only once in a lifetime”®.
Mice and rats
NAC AEGL; citing Ten Berge, et al.!4; derived AEGL-3 values by using data from studies of mice. However, Ten Berge, et al. explicitly indicated that this is invalid: “The plots of concentration against exposure period... reveal that mice are often more sensitive than other mammals... These findings suggest that experiments using mice do not provide an appropriate basis for predicting quantitatively the mortality response in humans”. Indeed, rat data are more relevant to humans than mouse data. NRC’s Committee on Toxicology addressed the issue of the appropriateness of species as a model for humans as follows: “This subject area has been reviewed (Hakkinen and Witschi, 1985) and various mammalian species (rat, hamster, and rabbit) have been identified as appropriate species for extrapolation from several perspectives”. Mice are conspicuously absent from the list of species appropriate for extrapolating to humans. However, as acknowledged by ORNL, mice are approximately three times as sensitive as rats to ammonia. Thus, use of rat data instead of mouse data would appear to multiply AEGL-3 values by a factor of three. Lest failure to do this might seem biased, NAC AEGL demonstrated how derivation of AEGL-3 values based upon the rat data could produce
approximately the same AEGL-3 values. NAC AEGL simply added an interspecies ‘uncertainty factor’ of three to restore essentially the same values as those derived from the mouse data. However, such an uncertainty factor is needed only in the absence of human data. In the case of ammonia, accident data are available and suggest comparable tolerance in humans, which is also reasonable given the greater diameter of the respiratory tree in humans than in smaller species. Further, the primary mechanism of toxicity is lower respiratory irritation, and ORNL has verbally acknowledged that “it doesn’t matter whether the cell being irritated came from a rat or a human.”
Scaling from animals to humans
NAC AEGL failed to scale ammonia concentrations used for exposure of mice to human equivalent concentrations (HECs). This violates procedures specified by the NRC Committee on Toxicology for “cross-species extrapolation: specific dose adjustment” in AEGL development®. Specifically, NRC states: “The extrapolation of quantitative animal data to humans, whether for threshold toxicity or for assumed non-threshold toxicity (eg, genotoxic carcinogens) requires cross-species extrapolation.” NRC cites procedures specified by the US EPA for developing inhalation reference concentrations (RfCs) under the Clean Air Act: “The RfC methodology requires conversion by dosimetric adjustment of the NOAEILs and LOAELs observed in laboratory animal experiments... to human equivalent concentrations (HECs) ...The dosimetric conversion to an HEC is necessary before the different adverse effects in the data array can be evaluated and compared”.
Resolution of AEGL-3 differences
Resolution of the 4.4-fold difference between RAM TRAC and NAC AEGL recommendations can be achieved via two steps, restoring consistency of NAC AEGL
NITROGEN No, 228
SAFETY
procedures with those specified by the US EPA! and NRC’s Committee on Toxicology®. First, conversion of mouse or rat data to HEC values would multiply AEGL-3 values by a factor of 2.7. If only this adjustment were implemented, this would bring AEGL-3 values to within two thirds of RAM TRAC recommendations, which were de-rived based upon half the five-min-ute zero-mortality level of 33,737 ppm-v in the Potchefstroom accident reconstruction. ‘This leaves a difference of 1.6-fold.
Virtually all of the above difference can be eliminated by a second step, that is, adopting rat data (approximately mouse data times three) with an interspecies uncertainty factor of two rather than three proposed by NAC AEGL. This produces an approximate 1.5-fold increase in NAC AEGL proposed values. This adjustment is technically justified based upon accident reconstruction data, which suggest higher human tolerance to ammonia than assumed by NAC AEGL. Accident reconstruction data indicate the absence of any need to apply an interspecies uncertainty factor relative to rat data. The resulting harmony of AEGL-3 values with RAM TRAC recommendations would decrease the area of Emergency Planning Zones around ammonia facilities by a factor of approximately 20fold, with commensurate savings in emergency planning costs and reductions in the difficulty of permitting or re-permitting ammonia facilities.
AEGL-2 values: NAC AEGL violates AEGL-2 definition
The AEGL-2 is defined to protect against irreversible or long-term injury; including incapacity to escape from further, potentially injurious exposure®’?. However, NAC AEGL was unable to quantify this endpoint empirically. Instead of estimating it, NAC AEGL applied a different criterion:
“Reversible effects may become irreversible and irreversible effects may become lethal due to delays in medical treatment as well as to
NITROGEN No. 228
continued exposure. For these reasons, the AEGL-2 levels for ammonia are based on the concentrations considered to be “intolerable” rather than those causing irreversible or long-term effects”’.
NAC AEGL used a study by Verberk (1977) involving exposure of volunteers to ammonia at concentrations up to 140 ppm for up to two hours as one basis for AEGL-2 derivation’. According to the ORNL support document: “The study showed also that humans can breathe these concentrations for one hour without becoming incapacitated. However, some irritation to the eyes, nose, throat, and chest along with a disagreeable odour were reported by the subjects”. The Verberk study appears to be irrelevant to AEGL-2 derivation, given the mild effects reported by volunteers relative to defined AEGL-2 criteria.
Another study which NAC AEGL deemed relevant for quantifying the AEGL-2 value of 160-ppm at 30 minutes (and other values at other time frames), was a ‘sensory irritation test’ (Barrow, et al.)! in which a 30-minute concentration (termed the RD-50) of 303 ppm reduced the breathing rate of mice by 50 percent. According to ORNL, “Barrow, et al. predicted that the RD-50 concentration would elicit intense sensory irritation and is expected to be rapidly incapacitating to humans”’. However, Barrow did not state that 303 ppm also would be the human RD50. Indeed, the human equivalent concentration, or HEC, would be 2.7 times 303 ppm, or 818 ppm (18 % higher than RAM TRAC’s recommended 30-minute AEGL-2 value of 696 ppm). However, a study by Silverman, et al. (1949; 13) of 30-minute exposure of volunteers to ammonia at 500 ppm concluded that depression of the human breathing rate did not occur in this concentration range: “[t]he most significant physiologic change in response to ammonia was the increase in respiratory minute volume, amounting to 50 to 250 % over control values”.
NAC AEGL is thus incorrect that 303 ppm represents a plausible RD-50 concentration for humans.
More fundamentally, and contrary to NAC AEGL’s judgment that the RD-50 is relevant for AEGL-2 derivation, the US EPA indicated that sensory irritation tests (the RD-50 concentration) are invalid for quantifying acceptable exposure levels for humans: “A test for sensory irritation in laboratory animals was developed [the RD-50], based on the premise that if sensory irritation can be prevented then systemic effects will be prevented as well... [However...] [t]he relationship of sensory irritation to airway irritation is unknown... An evaluation of the sensory irritation test ... found that quantitative evaluation with respect to human data was not possible due to ... factors, including .... intra- and interspecies inconsistencies in response. For these reasons, the suitability of the sensory irritation test results is limited to serving as an indication of the potential for respiratory tract irritation” !5.
Exposure beyond 30 minutes
The proposed AEGL-2 is 160 ppm at 30 minutes and 110 ppm thereafter is based upon the need to preserve the ability to escape. The 30-minute concentration should not be reduced, however, for longer-term exposures of one hour, four hours, and eight hours. These values should remain at 160 ppm, consistent with NAC AEGL policy. Thirty minutes is a traditional period assumed for escape from chemical emergencies, and is embodied in the Immediately Dangerous to Life and Health (IDLH) parameter of the National Institute of Industrial and Occupational Health (NIOSH). Given the extreme volatility and fugacity of ammonia, ammonia can be expected to have completely dissipated within 30 minutes following a release. Thus, the generally reasonable 30-minute escape-time assumption is also specifically reasonable, if not more than reasonable, in the case of ammonia. As ORNL? indicated, the “immediate response to individuals exposed to severe irritating concentrations of ammonia is to escape from the area”. Thus, if 160 ppm is regarded as a concentration
July-August 1997
SAFETY
that would permit escape within 30 minutes, it should also permit escape within one hour, four hours, and eight hours; under the reasonable assumption that no more than 30 minutes would be required for escape (of both people and ammonia).
Accordingly, at the suggestion of the NAC AEGL’s ammonia chemical manager (at the 17th September 1996 meeting)!!, NAC AEGL decided to maintain AEGL2 levels constant beyond 30 minutes, except when toxic effects beyond 30 minutes could be demonstrated. However, in setting AEGL-2 values for ammonia at the June-1997 meeting, NAC AEGL denied having established this policy. However, during discussion of a constant value beyond 30 minutes, the Chair began to indicate that “this would be consistent with...,” and then stopped himself from referring to the Committee’s prior policy, apparently recalling its current denial that the prior policy had been made. However, all events eventually can be checked, because under rules of the U. S. Federal Advisory Committee Act (FACA), tapes of Committee deliberations are public documents. Such checking has just now become possible, because previously denied requests under the Freedom of Information Act for transcripts of earlier NAC AEGL proceedings finally have been produced in accordance with the law.
RAM TRAC recommendations
RAM TRAC’s recommendations for AEGL-2 (see Table 1) are based upon several factors®. One factor was the need to establish AEGL-2 at levels higher than 500 ppm over 30 minutes because the Silverman study!3 revealed signs neither of serious injury nor of incapacity to escape impending at this exposure level. A second factor was the desirability of being consistent with NIOSH procedures for deriving IDLH values, which are concentrations above which health and/or the ability to escape within 30 minutes may become impaired. The NIOSH IDLH for ammonia is cur-
July-August 1997
rently 300 ppm, but should be upwardly revised. The NIOSH procedure for deriving the IDLH is to establish a preliminary IDLH value of one tenth the (animal) LC50 (the concentration lethal to 50 % of those exposed for a specified duration). This value is then checked against a variety of criteria, such as any reported RD50 values, before being finalized. Rather than applying this procedure to an animal LC-50, to derive AEGL-2 values, RAM TRAC applied the NIOSH procedure to a human LC50 calculated by ORNL based upon published regression coefficients for populations unusually sensitive to ammonia.
Resolution of AEGL-2 differences
Resolution of differences between RAM TRAC and NAC AEGL recommendations should begin with NAC AEGL adherence to its definition of AEGL-2. AEGL-2 values should be derived to protect against irreversible injury or incapacity to escape. Further, NAC AEGL should adhere to its policy of maintaining AEGL-2 values constant beyond 30 minutes, except if necessary to reduce such values to protect against toxic effects associated with prolonged exposure. This step would bring AEGL-2 values to 160 ppm at exposure durations of 30 minutes or longer. A third step would be for NAC AEGL to adopt the Silverman, et al. 30-minute value of 500 ppm, which produced neither injury nor incapacity to escape among volunteers exposed to it. This would bring NAC AEGL into the range of RAM TRAC recommendations (500 ppm v. 696 ppm at 30 minutes), though somewhat higher values especially at five minutes and 30 minutes would be needed to complete harmonization. If NAC AEGL maintained 500 ppm at one hour, four hours, and eight hours, then its values would actually exceed RAM TRAC recommendations; harmonization at these time frames consistent with NAC AEGL definitions and policies should not be a problem. |
References
1. Barrow, C. S.; Y. Alarie, and M. F. Stock. ‘Sensory irritation and incapacitation evoked by thermal decomposition products of polymers and comparisons with known sensory irritants’. Arch. Environ. Health, 33:79-88, Mar/Apr. 1978.
2. Letter to NAC AEGL from James Skillen (TFD, transmitting report from R. B. Kapuscinski re AEGL values for ammonia, Feb. 1997.
3. Mazzola, C. A. ‘Potchefstroom dose reconstruction: uncertainties that significantly limit application to human health standards’. Stone & Webster Eng. Corp., May 1997.
4, Michaels, R. A. ‘Emergency planning issues relating to ammonia’. Pro. 19th Ann. Meeting, Intl Inst. of Ammonia Refrigeration pp. 14488, March 1997.
5. Mulder, J. S., and H. O. Van Der Zalm. ‘Een gaval van dodelijke ammoniakvergiftiging’ (Fatal case of ammonia poisoning). Tijdschrift Voor Sociale Geneeskunde, 45:458 -60, 1967.
6. NRC guidelines for developing community emergency exposure levels for hazardous substances. Washington DC; 1993.
7. “Draft AEGL for ORNL, June 1997.
8. ‘Acute inhalation risks potentially posed by anhydrous ammonia’.; RAM TRAC Corp., May 1996.
9. ‘Comments of R. A. Michaels, to NAC AEGL’. RAM TRAC Corp., Aug 1996.
10. ‘Supplemental comments by R. A. Michaels to NAC AEGL’. RAM TRAC Corp, Aug 1996.
11. ‘Residual issues of ammonia emergency planning: comments by R. A. Michaels to NAC AEGL, RAM TRAC Corp., Nov 1996.
12. ‘Compilation of previous comments still requiring response regarding draft ORNL ammonia support documents submitted to NAC AEGL’. RAM TRAC Corp., June 1997.
13. Silverman, L., J. L. Whittenberger, and J. Muller. ‘Physiological response of man to ammonia in low concentrations’. J. Industrial Hygiene and Toxicology, 31:74-8, 1949.
14. Ten Berge, W. F.; A. Zwart, and L. M. Appelman. ‘Concentrationtime mortality response relationship of irritant and systemically acting vapours and gases’. 7 Hazardous Materials, 13:301-9, 1986.
15. US EPA.‘Methods for derivation of inhalation reference concentrations and application of inhalation dosimetry’. Washington DC; EPA/- 600/8-90/066F, Oct. 1994.
Ammonia’;
NITROGEN No. 228
Publication citation
Michaels, RA. Ammonia under the safety spotlight again. Nitrogen (London), 228:27-31, July/ August 1997.