REVIEW OF AEHA RISK-ASSESSMENT REPORTS ON ZINC CADMIUM SULFIDE
THE ARMY'S HEALTH risk assessment for Minneapolis, MN, Corpus Christi, TX, Fort Wayne, IN, and St. Louis, MO, addresses the potential adverse health effects resulting from the Army's atmospheric-dispersion tests with ZnCdS. Its conclusion was that because of the brief exposure period and concentrations used, the dispersion tests should not have posed any adverse health effect for people in the test areas. The Army recognizes that there was a large degree of uncertainty associated with its estimates of human health risk, as in any risk assessment, and that the calculated estimates should be taken not as "absolute estimates of risk but rather as conditional estimates."
The subcommittee's evaluation of the Army Environmental Hygiene Agency (AEHA) reports on risk assessment of ZnCdS is as follows.
APPROPRIATENESS OF USING CADMIUM TOXICITY IN RISK-ASSESSMENT REPORTS
In the absence of toxicity data on ZnCdS, AEHA used the toxicity data on cadmium in a worst-case scenario. For an accurate and defensible assessment of potential health hazards resulting from an exposure to a
chemical substance, it is important that particular characteristics of the substance be known. The strength and defensibility of the health assessment is greatly enhanced by the inclusion of information on the physical and chemical properties of the substance, its toxicokinetics and bioavailability, the type of toxic response that it elicits, the exposure concentration and duration necessary to produce the response, and the population at risk. Often, such data are lacking and the toxicologist must use other available toxicity data to assess the health risk accurately. An appropriate surrogate may be used—one of the individual components of the substance in question or a closely related substance with known toxicity. Use of such surrogates is especially appropriate if the physical and chemical properties of the two agents are similar, if the toxicity and toxicokinetics of the surrogate would be expected to mimic closely those of the test agent, and if judgment indicates that this approach would likely overestimate rather than underestimate the risk.
The National Research Council Committee on Toxicology (NRC 1988) and the US Environmental Protection Agency (EPA) have both used such an approach successfully. For example, chromium and chromium compounds have been used as surrogates for lithium chromate because little information was available on the toxicity of lithium chromate and because chromium is its most-toxic component, and benzene has been used as a surrogate for gasoline because benzene is its most-toxic component and EPA wanted to use a worst-case scenario.
The situation is similar for ZnCdS. There are few studies on the general toxicity, bioavailability, and toxicokinetics of ZnCdS, so it is necessary to consider the use of existing toxicity databases such as those on cadmium sulfide, zinc sulfide, and cadmium and zinc and their salts. Because zinc is an essential element for humans and animals and is generally considered to be relatively nontoxic in small amounts, the subcommittee concluded that the Army's approach of estimating the toxic potential of ZnCdS by using toxicity data on cadmium compounds as a surrogate was appropriate and could be considered to be conservative and prudent and to constitute a worst-case scenario for estimating risk associated with exposure to ZnCdS.
DISPERSION METHODS
CORPUS CHRISTI
ZnCdS was disseminated from an airplane flying parallel to the coast off shore of Port Aransas (AEHA 1994a). Nine single-release tests were performed over a 6-d period. Observed-concentration data on 2 of these tests are missing. ZnCdS was released at the rate of 1.5 lb/mile over a 30-mile course, producing a line source with a strength of 1.11 x 109 particles/ft (Smith and Wolf 1963). The tracer was released at an altitude of 500 ft in all experiments except for tests 15 and 17, whose release altitudes were 750 and 1,000 ft. Rotorod samplers were placed at 1-mile intervals along a 25-mile-long line downwind of and perpendicular to the line source at its midpoint.
MINNEAPOLIS
ZnCdS tracer was released from both stationary and mobile sources (AEHA 1994b). The winter program consisted of 23 field tests with 63 releases. Data are available on only 20 tests, which comprised 52 releases. The summer program consisted of 15 field tests with 39 releases, and data are available on all these tests. The winter program began on January 19, 1953, and lasted until April 28, 1953. Tests were performed in 4 areas of the city ("Able," "Baker," "Charlie," and ''Dog") that were selected because of their varied topographic and land-use characteristics; ZnCdS was dispersed from a point source either on a stationary vehicle or on a rooftop. Two tests were also performed on a citywide basis; ZnCdS was dispersed from a moving vehicle (line source). The summer program ran from August 21, 1953 to September 18, 1953. Point-source and dual-point-source releases were made from rooftops in 3 areas of the city ("Able,'' "Dog," and "Easy"), and 4 citywide tests were made with ZnCdS dispersed from moving vehicles. Filter samplers were placed outside and in residences, schools, and commercial buildings. The sampling methods and networks are not described in detail in the Army risk assessment (AEHA 1994b).
FORT WAYNE
There were 23 field tests comprising 75 releases during the summer of 1964 and the winter of 1965-1966 (Hilst and Bowne 1966). There were 2-4 releases for each test; 2 airplanes were used. The planes released ZnCdS at 2.5 lb/mile along a 20-mile straight line perpendicular to the mean wind, upwind of the city. The first plane released yellow ZnCdS (ZnCdS2267, lot H-395), and the second released green ZnCdS (ZnCdS3206, lot H-396, and one release of lot H-391). After the initial release, the plane returned to the field for reloading. Later releases were made after the plumes of the first releases had passed through the sampling area. Cassette-filter samples were collected at 25 random locations in the Ft. Wayne sampling network. Each filter sampler was set up to take 10 30-min sequential samples and 1 full-period integrated sample. Some 250 Rotorod samplers were used to collect integrated samples along 5 lines across the city: 1 at the upwind edge of the city, 1 at the downwind edge, and 3 dividing the city according to major land-use categories. The 50 samplers in each line were spaced as evenly as possible. In addition, Gelman paper tape samplers, programmed to collect sequential 5-min-average samples, were placed at 10 locations to detect the times of arrival and departure of the tracer plume.
ST. LOUIS
Sixteen field tests comprising 35 releases were carried out from May 20, 1953, through June 25, 1953 (AEHA 1994c). Tests were performed in 2 areas representing different topographic and land-use characteristics: the "How" area near the center of the St. Louis metropolitan section about 2 miles west of the Mississippi River and 1 mile from downtown, and the "Item" area in the middle eastern portion of the metropolitan area. They included 19 single-point and 10 simultaneous-dual-point releases from stationary vehicles or rooftops. In addition, a citywide test used 2 line-source dispersals from moving vehicles. The Army risk assessment does not describe the sampling methods and sampling network in detail (AEHA 1994c). Samples were collected with the membrane filter meth-
od (J. Kirkpatrick, US Army Center for Health Promotion and Preventive Medicine, Personal commun., Feb. 8, 1996). One release did not yield useful data, because the samplers were not downwind of the release; this release was not considered in the Army risk assessment.
EXPOSURE ANALYSIS
Army risk assessments of ZnCdS were based on direct transport from source to human receptor. Inhalation was the only route of entry that the Army considered important. The general approach was to estimate the exposure in each city from air-sampling data. The risks associated with these exposures were then examined. The methods used to estimate exposures were somewhat different for each city.
CORPUS CHRISTI
Only the integrated particle-number concentration in particle-minutes per cubic meter from the Rotorod samplers was available for analysis (AEHA 1994a). The Rotorod samplers used in the Corpus Christi experiments had a calculated sampling rate of 41.8 L/min. The average sampling efficiency was found to be about 75%, compared with a filter-sampling method assumed to have an efficiency of 100% (Smith and Wolf 1963). An average effective sampling rate of 30.8 L/min was calculated on the basis of the sampling efficiencies. However, the same sampling rate as that used in earlier Dallas Tower experiments, 33 L/min, was used in the calculations for determining ZnCdS concentrations in Corpus Christi because sampling rates determined at Corpus Christi were variable and on the average were not distinguishably different from the Dallas sampling rate (Smith and Wolf 1963).
The highest ZnCdS concentration was observed at the sampling station closest to the source. In the AEHA (1994a) risk assessment, the integrated concentration from this station summed over all 9 tests, 232.7 particle-min/L, was used as a conservative estimate of exposure throughout the area. The particle-count exposure was converted to mass exposure by
using the number of particles per gram from Smith and Wolf (1963), 2.16 x 1010 particles/g. The average mass concentration was obtained by dividing the integrated concentration by the sampling time. The actual sampling time was not available, so the sampling time used in a similar study, 300 min, was used in the Army calculations. The computed ZnCdS concentration was 3.59 x 10-5 mg/m3, or a cadmium concentration of 7.2 x 10-6 mg/m3, assuming that ZnCdS is 20% cadmium by weight.
The Army risk assessment estimated an integrated cadmium exposure equivalent to 0.029 µg by inhalation of particles transported directly from source to human receptor. That value assumes a breathing rate of 0.8 m3/h, 100% deposition efficiency in the lung (as a worst-case scenario; a more reasonable assumption would be that the deposition efficiency is 10-15%), and an exposure time of 300 min. The assumed sampling and exposure times cancel out and have no effect on the integrated-exposure estimate.
MINNEAPOLIS
Unlike the Corpus Christi study, the Minneapolis experiments varied in sources and receptor locations from test to test. The methods used to analyze concentration data for the Minneapolis experiments are described briefly in the Army risk assessment (AEHA 1994b). The approach taken was to plot isopleths of integrated concentration in particle-minutes per liter for each test on a Minneapolis map. The representative isopleth for the maximal concentration in each area and for each release was used to represent the dosage for that release. Isopleths used were from 100 to 100,000 particle-min/L with intervals for each power of 10 between (logarithmic intervals). For each test site, the areas of the regions between the 1,000- and 9,999-particle-min/L isopleths and between the 10,000 and 99,999-particle-min/L isopleths were reported (AEHA 1994b).
The Army then calculated the integrated exposure in milligram-minutes per cubic meter, the average concentration of ZnCdS in milligrams per cubic meter, and the average concentration of cadmium in milligrams per cubic meter corresponding to values of integrated count concentration
at each isopleth, shown here in Table G-1. The conversions from particle-minutes per liter followed the procedure used in the Army risk assessment for Corpus Christi, TX, discussed above, assuming a sampling duration of 60 min.
TABLE G-1 Army-Calculated Concentrations for Each Isopleth Value of Integrated Particle-Count Concentration, Minneapolis, MN (AEHA 1994b)
Count Concentration (particle-min/L) |
Sampling Time (min) |
Integrated ZnCdS Concentration (mg-min/m3)a |
ZnCdS Concentration (mg/m3)a |
Cadmium Concentration (mg/m3)a |
100 |
60 |
4.63 x 10-3 |
7.72 x 10-5 |
1.54 x 10-5 |
1,000 |
60 |
4.63 x 10-2 |
7.72 x 10-4 |
1.54 x 10-4 |
10,000 |
60 |
4.63 x 10-1 |
7.72 x 10-3 |
1.54 x 10-3 |
100,000b |
60 |
4.63 x 100 |
7.72 x 10-2 |
1.54 x 10-2 |
247c |
60 |
1.14 x 10-2 |
1.91 x 10-4 |
3.8 x 10-5 |
a Values rounded from 4 or 5 significant figures to 3 significant figures. b 100,000 isopleth occurred only 1 time in only 1 area. c Integrated count concentration reported at Clinton School. |
The Army risk assessment states that the isopleths were the only exposure data, except for data from Clinton School, available from the "Minneapolis report," presumably the Chemical Corps reports referred to in the risk assessment (AEHA 1994b). The risk assessment states that the maximal integrated concentration obtained outside the Clinton School was 247 particle-min/L; this value was used to represent exposure at the Clinton School.
It is unlikely that the maximal reported concentrations in Minneapolis represent an upper bound for the outdoor exposures by direct inhalation, inasmuch as some people might have been closer to the source or more directly downwind of the source than any of the sampling stations.
FORT WAYNE
The Army risk assessment for Fort Wayne gives estimates for 1-h, 8-h, and chronic (integrated) exposures (AEHA 1995). For the 1-h estimated exposure, the Rotorod sampler results were used because they gave the highest reported count concentrations. The highest reported integrated count, considering both yellow and green releases—yellow ZnCdS at 1,878 particle-min/L from the February 4, 1966, experiment—was selected for the 1-h exposure value in the Army risk assessment. The actual sampling times were not available, and the Army assumed a 38-min sampling time. That was the shortest period between releases, so it provides a conservative estimate of the individual sampling duration.
For the estimated 8-h exposures, the integrated concentrations for each release and for both colors of FP were summed to obtain the total daily maximal exposure. The maximal daily exposure over all the days was then selected as the 8-h exposure estimate: 2,035 particle-min/L from the December 5, 1965, experiment. The Army estimated the sampling time to be 150 min, but no rationale for this was presented.
To estimate chronic exposure, the Army plotted isopleths of integrated concentrations summed over the entire experimental program. The highest isopleth was 80 particle-min/L. The area on the high side of the 80-particle-min/L isopleths could be as high as 90 particle-min/L because the interval between isopleths is 10 particle-min/L. Thus, 90 particle-min/L was used as the chronic-exposure estimate.
The integrated concentration values in particle-minutes/liter for each of the 3 exposure times were converted to cadmium concentrations in milligrams per cubic meter by using the assumed sampling duration, a weight percentage of cadmium in ZnCdS of 20%, and the particles-per-gram values given in Hilst and Bowne (1966). The yellow ZnCdS had 1.32 x 1010 particles/g, and the green, 1.45 x 1010 particles/g. The mean of these, 1.385 x 1010 particles/g, was used for converting count to weight units when the sum of the yellow and green ZnCdS concentrations was used. The Army calculated the 1-h and chronic concentrations by dividing the integrated concentrations, shown in Table G-2, by the corresponding sampling times and particles-per-gram values. That approach yields a conservative value for the time-weighted average direct-transport inhala-
tion exposures because it assumes that the concentration during the sampling period persisted beyond the sampling period. It is more likely that the direct-transport exposure by inhalation persists only during the time required to pass by a person downwind of the source; this is roughly the same as the sampling time. The average concentration in terms of weight of cadmium per volume of air was calculated by multiplying by 0.2, the Army's assumed weight fraction of cadmium in ZnCdS.
TABLE G-2 Army Estimates of Time-Averaged Concentrations for Fort Wayne, IN, Atmospheric Tracer Experiments (AEHA 1995)
Type of Exposure |
Integrated ZnCdS Concentration (particle-min/L) |
Sampling Time (min) |
ZnCdS Concentration (mg/m3)a |
Cadmium Concentration (mg/m3)a |
1-h |
1,878 |
38 |
3.75 x 10-3 |
7.50 x 10-4 |
8-h |
2,035 |
150 |
1.47 x 10-4 |
1.96 x 10-4 |
Chronic |
90 |
38 |
1.71 x 10-4 |
3.42 x 10-5 |
a Values rounded from 4 significant figures to 3 significant figures. |
The 8-h average concentration reported in the Army risk assessment cannot be calculated with the approach applied to the 1-h and chronic concentration estimates. Also, the reported cadmium concentration is higher than the reported total FP concentration. It appears that there are errors in calculating the 8-h concentrations.
ST. LOUIS
In the Army risk assessment for St. Louis, air-monitoring data were analyzed with the same approach used for data from Minneapolis. The isopleths for St. Louis ranged from 10 to 10,000 particle-min/L (AEHA 1994c). The concentrations corresponding to these isopleth values are reproduced here in Table G-3.
TABLE G-3 Army-Calculated Concentrations for Each Isopleth Value of Integrated Particle-Count Concentration, St. Louis, MO (AEHA 1994c)
Count Concentration (particle-min/L) |
Sampling Time (min) |
Integrated ZnCdS Concentration (mg-min/m3)a |
ZnCdS Concentration (mg/m3)a |
Cadmium Concentration (mg/m3)a |
10 |
60 |
4.63 x 10-4 |
7.72 x 10-6 |
1.54 x 10-6 |
100 |
60 |
4.63 x 10-3 |
7.72 x 10-5 |
1.54 x 10-5 |
1,000 |
60 |
4.63 x 10-2 |
7.72 x 10-4 |
1.54 x 10-4 |
10,000 |
60 |
4.63 x 10-1 |
7.72 x 10-3 |
1.54 x 10-3 |
a Values rounded from 4 or 5 significant figures to 3 significant figures. |
ADJUSTMENTS TO EXPOSURE ESTIMATES
It was noted that the conversion from ZnCdS to cadmium concentration is in error. This computation assumed that 20% of the mass of ZnCdS is cadmium, whereas actually only 15.6% is. Thus, the exposure estimates based on this concentration are 28% too high. Smith and Wolf (1963) discuss the effect of disseminator efficiency on the number of particles per gram. The efficiency was determined to be 39%, so the number concentration of particles per gram of dust dispersed produced by the disseminator is 39% of the number concentration per gram produced in the laboratory. A truly conservative approach to estimating exposure would be to decrease the number of particles per gram used in these calculations by 39%, increasing the weight concentration by a factor of 2.6. Taking into account the corrections for the weight fraction of cadmium in ZnCdS and for the disseminator efficiency, we recommend that the Army direct inhalation-exposure estimates be multiplied by a factor of 2.0.
For the Ft. Wayne estimates of chronic exposure, the isopleths were drawn by using "average surface dosages." It appears that the sum, instead of the average, should have been used to plot these isopleths.
For the pathway of direct atmospheric convection and inhalation, Table G-4 summarizes maximal integrated exposures in the 4 cities considered in this report. In Corpus Christi, the maximal integrated concentration
for each release was reported at the station nearest the line of release. The maximal integrated-exposure estimate was taken as the sum of the integrated concentrations at that monitoring station.
TABLE G-4 Maximal Integrated Exposures and Lung Deposition of Cadmium by Direct Atmospheric Convection and Inhalation Pathway
City |
Area |
Maximal Integrated Exposure (particle- min/L) |
Cadmium (mg-min/m3) |
Lung Deposition Cadmium (mg) |
Corpus Christi |
|
2.3E+02 |
4.3E-03 |
7.2E-05 |
Minneapolis |
Able |
1.7E+05 |
3.2E+00 |
5.3E-02 |
|
Baker |
2.8E+04 |
5.3E-01 |
8.8E-03 |
|
Charlie |
2.1 E+04 |
3.9E-01 |
6.5E-03 |
|
Dog |
8.0E+03 |
1.5E-01 |
2.5E-03 |
|
Easy |
2.4E+04 |
4.4E-01 |
7.4E-03 |
|
Citywide |
1. 1 E+05 |
2.0E+00 |
3.3E-02 |
|
Clinton School |
2.5E+02 |
4.6E-03 |
7.6E-05 |
Ft. Wayne |
|
1.6E+04 |
2.9E-01 |
4.8E-03 |
St. Louis |
How |
4.9E+04 |
9.1E-01 |
1.5E-02 |
|
Item |
1.6E+04 |
2.9E-01 |
4.8E-03 |
|
Citywide |
1.0E+02 |
1.9E-03 |
3.1E-05 |
For the Minneapolis and St. Louis experiments, the maximal integrated-exposure estimates for each area were found by summing across all releases for each area the values of the isopleth above the highest observed concentrations. For Minneapolis, data from the Army risk-assessment table 7 were used (AEHA 1994b). For example, the "Baker" area had 2 maximums within the 10,000-particle-min/L isopleth, 8 within the 1,000-particle-min/L isopleth, and 4 within the 100 particle-min/L isopleth. Thus, the maximal integrated exposure estimate was taken to be
(2 x 10,000) + (8 x 1,000) + (4 x 100) = 28,400 particle-min/L.
That estimate is conservative in that it uses the isopleth level exceeding
the highest reported measurement and assumes that the same person was exposed to the maximum from every release within an area. For the St. Louis experiments, the data for the calculations were taken from table 6 of the Army risk assessment (AEHA 1994c).
For the Ft. Wayne experiments, the maximal exposure estimate was calculated by summing the maximal integrated concentration for each release from table 3 in the Army risk assessment (AEHA 1995). The maximal integrated exposure was then converted from particle-minutes per liter to cadmium in milligram-minutes per cubic meter by using the adjustments mentioned above. Using the data from Corpus Christi as an example:

The integrated exposures in terms of cadmium were then used to estimate the lung deposition of cadmium, assuming 100% deposition and a breathing rate of 1 m3/h, shown below for the Corpus Christi data:

ASSESSMENT OF NONCANCER HEALTH EFFECTS
The subcommittee's evaluation of the noncancer health effects addressed in the AEHA report is presented below.
The Army states that because the Environmental Protection Agency did not have a noncarcinogenic-toxicity comparison value for inhalation (a reference concentration, or RfC) for ZnCdS or for cadmium, a number of assumptions, regarding both exposure and toxicity, had to be made in its assessment of risk, namely, that cadmium is a proxy for ZnCdS, that ZnCdS is insoluble, and that inhalation is the exposure pathway. Major factors considered by the subcommittee include the appropriateness of using cadmium-toxicity data in the Army's assessment of the risk associated with exposure to ZnCdS; the assessment of both acute toxicity and
chronic toxicity as related to the dispersion tests, and the evaluation of the appropriateness of the conclusions drawn in the Army's report.
The risk assessment developed by the Army correctly emphasizes that it is the solubility of cadmium compounds that eventually will play a determining role in causing pulmonary or systemic effects. ZnCdS is considered to be much less bioavailable and therefore less toxic than many other cadmium compounds. Furthermore, it is pointed out that lung injury caused by inhalation of high concentrations of cadmium compounds, as seen in experimental animals, can range from signs of acute diffuse alveolar damage, with accompanying edema and regenerative changes during the recovery phase, to chronic degenerative lesions. Acute exposure to high concentrations of cadmium have caused death in humans, but no data are known on adverse pulmonary effects in humans exposed chronically to low concentrations. Under such conditions, renal toxicity, and not pulmonary toxicity, might be the adverse health effect that is of most concern.
Because of the lack of toxicity data on the health effects of ZnCdS, it was reasonable to assume, for the purpose of assessing risk, that cadmium sulfide, CdS, would be the most-toxic constituent. The health risk assessment conducted by the Army assumed a worst-case situation regarding the retention and absorption rate of the inhaled test substance. The assumption was that the airborne particles would be so small as to be 100% respirable. Considering the low airborne concentrations of ZnCdS aerosol measured in the Army's dispersion studies and the short exposure period, it is unlikely that such exposure to the substance would be sufficient to cause lung injury.
With regard to the assessment of possible noncarcinogenic pulmonary effects, the subcommittee concurs with the conclusion of the Army that, because of the low concentration and duration of exposure, the ZnCdS tests posed negligible pulmonary-health threats to residents in the test areas.
In the final risk assessment for noncancer effects, the Army document deals with kidney damage, and not with potential lung damage, caused by cadmium and then compares possible exposure to ZnCdS compounds with accepted standards for exposure of workers to cadmium compounds in industry. That approach seems to be justified.
The Army's health risk assessments for Minneapolis, MN, Fort Wayne, IN, and Corpus Christi, TX are adequate in addressing concerns regarding adverse health effects among those environmentally exposed populations. As stated in each of the 4 documents, several assumptions underlie the risk assessments (that cadmium is a proxy for ZnCdS, that ZnCdS is insoluble, and that inhalation is the exposure pathway). The Army states the limitations of its exposure estimates and appears to be conservative in interpreting results, given the many uncertainties associated with exposure estimation. Children were recognized to be more sensitive but no estimates of exposure were made for children except in Minneapolis.
The subcommittee concurs with the conclusions of the Army's health risk assessment for noncancer effects of the dispersion of ZnCdS. Because of the lack of toxicity data on the health effects of this chemical, the Army's assessment was based on available information on exposure to cadmium. That is reasonable because any expected toxicity due to ZnCdS exposure should be related to the cadmium available and not to the relatively nontoxic zinc. The subcommittee agrees that because of the low concentrations and short duration of the exposure, the ZnCdS tests posed negligible pulmonary-health threats to the residents in the test areas. With those considerations, the Army's health risk assessment is reasonable.
The subcommittee notes four concerns. First, there is no description of the population at risk in any document. From an epidemiologic perspective, coupled with appropriate attention given to ''sensitive'' subpopulations, a description of the population at risk would be helpful. Specifically, census data could be used to describe the population at risk by age, sex, population density, and economic status. The subcommittee's assessment also recognized some uncertainties that exist when applying the standards for healthy workers to the general population. The Army understands that in the general population, some people, such as the elderly and children might be more sensitive to the effects of cadmium than the healthy worker. As a general toxicologic practice, one does not assume that a safe level for adults is protective for children. The Army assumed that because the calculated air concentration was lower than the established occupational safety levels and the exposure was for a short period,
the air concentration in the testing area would not represent any undue hazard for the general population. Although the Army recognized that children could be more sensitive, no estimates of exposure were made for children except in Minneapolis, MN.
A second concern stems from the Army's inconsistent use of the terms used in the toxicity-assessment part of the documents. Such terms as "inconclusive studies," "limited data," "not generally associated," and "inconclusive data" are fairly close in meaning but subject to individual interpretation. Operational definitions for terms would be helpful in providing the reader with a sense of whether there are data, whether the data support an association (with or without statistical significance), or whether the studies are of sufficient quality. Regardless of the definitions chosen, consistent application of the terms to each organ system is warranted.
The third concern is that the executive summaries in the 4 documents express reproductive effects in slightly different, albeit subtle, ways. The executive summaries for Corpus Christi and Minneapolis state that there are no adverse reproductive outcomes among women occupationally exposed to higher concentrations of cadmium for longer periods. Later, in the toxicity-assessment sections, the Army notes that reductions in birthweight among offspring born to exposed women were observed in 2 studies. In Fort Wayne, "no deformities in offspring . . . were noted." The summary is correct in noting that reproductive risk has not been fully characterized. A spectrum of potential reproductive or developmental outcomes would need to be included to assess reproductive and developmental toxicity fully, especially at lower (background) levels of exposure.
Finally, the lung is a prime target for airborne substances, but other routes, such as dermal exposure and ingestion, should have been considered. The exposure profiles of adults and children can differ widely. Children tend to have more oral exposure, from hand-to-mouth contact, which could increase the potential for ingestion.
ASSESSMENT OF CANCER RISK
The subcommittee evaluated the cancer risk assessments for Corpus
Christi, TX, Minneapolis, MN, St. Louis, MO, and Fort Wayne, IN, conducted by the Army Environmental Hygiene Agency (AEHA 1994a,b,c, 1995). The general approach used was that of the Environmental Protection Agency, namely, the lifetime cancer risk was calculated as the product of cancer potency and average lifetime daily dose. Because the potency was calculated to represent an upper bound for low-dose cancer-risk estimation,

That procedure is widely used and is adopted for this report.
The Army used a cancer potency of inhaled cadmium of 6.3 mg/kg per day on the basis of studies in rodents (Integrated Risk Information System, Office of Health and Environmental Assessment, Environmental Protection Agency, Washington, DC). As discussed in Chapter 5, we have chosen to use a lung-cancer potency estimate of 11.7/(mg/kg) per day based on lung cancer observed in cadmium workers. That increases the Army estimates of risk by a factor of 11.7/6.3 = 1.9.
Arguments are presented in chapter 5 for increasing the cancer risk estimates by a factor of 10 to account for greater sensitivity at some ages, particularly for children.
The Army used an inhalation rate of air of 0.8 m3/h. For active people and workers, the National Research Council typically uses an inhalation rate of 1 m3/h. That increases the dose, and hence the risk, by a factor of 1/0.8 = 1.25.
The Army used an average lifetime of 70 yr, and the Research Council typically uses 75 yr. That averages out the cadmium dose over a longer period and reduces the average daily dose, and hence the cancer risk, by a factor of 70/75 = 0.93.
The Army considered cadmium to be 20% of ZnCdS. The subcommittee used a value of 15.6% that reduces the cadmium exposure by a factor of 15.6/20 = 0.78.
The efficiency of the particle disseminators used in some of the tests was 39%; that is, the weight per particle was 1/0.39 = 2.56 times as high as that used by the Army. Thus, the concentrations and risks need to be multiplied by 2.56.
The combination of the 6 factors makes the subcommittee cancer risk estimates higher than the Army estimates by a factor of
1.9 x 10 x 1.25 x 0.93 x 0.78 x 2.56 = 44.
The major difference is the factor of 10 introduced to allow for the possible additional sensitivity of particular age groups.
It appears that the Army estimates of cancer risk are based on the maximum average exposure at a site for 1 release in the Fort Wayne tests. There were 35 releases each of green and yellow cadmium compounds in the Ft. Wayne tests, so the cancer risk estimate should be increased by a factor of 35. The disseminator efficiency for the Ft. Wayne tests was 33%, compared with 39% at the other sites, so the overall modifying factor should be increased to 44 x (0.39/0.33) = 52, increasing the Ft. Wayne estimates by a factor of 35 x 52 = 1,820. Even with appropriate adjustments to the cancer risks calculated by the Army, the lifetime lung-cancer risks associated with the low doses of cadmium are low, and it is unlikely that anyone in these 4 test areas developed lung cancer as a result of direct inhalation of cadmium from the airborne releases.
REFERENCES
AEHA (U.S. Army Environmental Hygiene Agency). 1994a. Assessment of Health Risk, Corpus Christi, Texas. HRAS No. 64-50-93QE-94. U.S. Army Environmental Hygiene Agency, Aberdeen Proving Ground, Edgewood, Md.
AEHA (U.S. Army Environmental Hygiene Agency). 1994b. Assessment of Health Risk, Minneapolis, Minnesota. HRAS No. 64-50-93QE-94. U.S. Army Environmental Hygiene Agency, Aberdeen Proving Ground, Edgewood, Md.
AEHA (U.S. Army Environmental Hygiene Agency). 1994c. Assessment of Health Risk, St. Louis, Missouri. HRAS No. 64-50-93QE-94. U.S. Army Environmental Hygiene Agency, Aberdeen Proving Ground, Edgewood, Md.
AEHA (U.S. Army Environmental Hygiene Agency). 1995. Assessment of Health Risk, Fort Wayne, Indiana. HRAS No. 39-26-0467-95. U.S. Army Environmental Hygiene Agency, Aberdeen Proving Ground, Edgewood, Md.
Hilst, G.R., and N.E. Bowne. 1966. A Study of the Diffusion of Aerosols Re
leased from Aerial Line Sources Upwind of an Urban Complex. Vol. 1, Final Report; Vol. 2, Data Supplement. Contract DA-42-007-AMC-37(R). Prepared by the Travellers Research Center, 250 Constitution Plaza, Hartford, Conn., for the U.S. Army Dugway Proving Ground, Dugway, Utah.
NRC (National Research Council). 1988. Emergency and Continuous Exposure Guidance Levels for Selected Airborne Contaminants, Volume 8, Litium Chromate and Trichloroethylene. Washington, D.C.: National Academy Press.
Smith, T.B., and M.A. Wolf. 1963. Vertical Diffusion from an Elevated Line Source over a Variety of Terrains. Part A. Final Report. Contract DA-42007-CML-545. Prepared by Meteorology Research, 2420 North Lake Ave., Altadena, Calif., for the U.S. Army Dugway Proving Ground, Dugway, Utah.