Review of a Screening Level Risk Assessment for the Navy Air Facility at Atsugi, Japan (1998)

COT concurs with the overall recommendations made in the 1998 NEHCreport, particularly the recommendation that a full-year air-monitoringstudy be conducted. COT recommends that source emissions testingof the incinerator stacks be conducted, if possible, in additionto air-monitoring. Source emissions testing would provide more accuratedata on the types and concentrations of pollutants being emittedby the incinerator. However, if an air-monitoring study is the onlyviable testing method available to the Navy, careful design of thestudy is essential, and COT recommends the following:

1. The air-monitoring study should be designed not only to estimateexposure to the target population, but also to distinguish the impactof incinerator emissions from background air quality.

2. Nearby sources of air pollutants, other than the incinerator complex,should be characterized and their contributions to total risk evaluated.

3. Ground-level dispersion of pollutants from solid waste piles andliquid waste sources should be evaluated.

4. Adequate QA-QC measures should be taken during air sampling andanalysis to provide accurate measurements of pollutant concentrations.

5. More extensive meteorological data related to local atmosphericstability should be collected and used in dispersion modeling.

6. Continuous or semicontinuous monitoring data should be collectedto enhance ability to correlate meteorological data and emissions-dispersionestimates with ambient concentrations of pollutants.

7. Fine particulate matter less than 2.5 micrometers in aerodynamicdiameter (i.e., PM_2.5) should be monitored, as numerous studies in the last several yearshave implicated fine particles as a threat to human health. Ozone,nitrogen dioxide, carbon monoxide, and sulfur dioxide should alsobe monitored.

The remainder of this report details COT's review of the 1998 NEHC report and is organized into six majorsections: (1) adequacy of the methods used by NEHC to assess healthrisks, (2) scientific validity of the NEHC conclusions, (3) risksto susceptible subpopulations, (4) uncertainties in the NEHC riskassessment, (5) evaluation of NEHC recommendations, and (6) editorialcomments for improving the NEHC report.

In 1997, the Navy collected air samples for 23 days (between July27 and September 11) from nine locations at NAF Atsugi, includingliving, working, and recreational areas, an upwind (background) location,and a fence-line site. The samples were analyzed for PM₁₀, metals, semivolatile organic compounds, volatile organic compounds,formaldehyde, hydrochloric acid, and dioxins and furans. Descriptionsof the methods used to collect and analyze the air samples were notprovided in the 1998 NEHC report. However, at the request of COT,the Navy provided descriptions of those methods. Air sampling wasperformed using established methods, but QA-QC measures were inadequateto provide accurate measurements of pollutant concentrations. Therefore,COT agrees with NEHC that the sampling data can only be used forscreening purposes, because they are not sufficient to estimate actualrisk. COT suggests that the 1998 NEHC report be revised to includedescriptions of the air sampling methods and the QA-QC measures.

NEHC reported that appropriate background data were not availablebecause the selected background location (location 1) was affectedby point-source and fugitive emissions. However, no explanation ordata were provided to indicate why this location was believed tobe substantially affected by incinerator emissions in comparisonwith background conditions. The meteorological data presented inthe 1998 NEHC report (section 5.5, p. 26) show that the wind wasnot blowing from the incinerators toward location 1 when sampleswere collected. Therefore, location 1 would seem to provide a reasonableestimate of background air concentrations. The data collected atthis site showed that pollutant concentrations were high, indicatingthat the incinerator complex may not have been the main source ofpollution during the sampling period.

Lack of background data will make it difficult to attribute contaminantconcentrations measured at NAF Atsugi solely or predominantly tothe incinerator complex. For example, concentrations of PM₁₀ were found to be high at all sampling locations, and the maximumconcentrations were close to average concentrations. That observationraises questions about the general background concentrations of particulatesat NAF Atsugi.

The 1995 NEHC report described the incinerator operator's practice of pouring of liquid waste over solid waste piles to beincinerated. Evaporation could constitute a significant source ofexposure, especially if volatile wastes are involved. Considerationof this potentially significant air-pollution source is absent inthe 1998 NEHC report. NEHC also failed to test air samples for ozone,nitrogen dioxide, carbon monoxide, and sulfur dioxide, as recommendedby COT in 1995 (NRC 1995). COT recommends that those pollutants beincluded in future air-monitoring studies.

In presenting the air-sampling data, the number of air samples andthe specific times they were taken should be included. That typeof information is provided only in the discussion on Risk AssessmentLimitations (section 7.1, pp. 37-38); it should have been discussedearlier in more detail in Data Evaluation, Reduction, and Screening(section 2.0, pp. 2, 8). For example,

the sampling times for the maximum detected concentrations shouldbe presented in Table 2-1 (pp. 9-10) and the averaging times forthe exposure point concentrations should be presented in Table 2-2(p.12). An explanation should be given for sampling some locationsmore often than others. For example, location 1, which is upwindof the plant, was measured on 27 days, and location 4, which is downwindand closest to the incinerator complex, was measured only on 8 days,and yet had the highest hazard indices. Clarification is also neededto explain why location 1 was sampled more often, on 27 days (section7.1, p. 37), than the stated 23 days of sampling.

NEHC used U.S. EPA's Region III risk-based screening concentrations (RBSCs) for ambientair to identify which quantified chemicals in the air samples wereof potential concern. If the maximum detected concentration of apollutant was greater than EPA's Region III RBSC, the pollutant wasconsidered in the NEHC risk assessment, but pollutants that did nothave RBSCs also were evaluated. A total of 24 pollutants were considered,17 of which exceeded EPA's Region III RBSCs and seven of which hadno available RBSC. Because RBSCs were used as thresholds to determinewhether a pollutant was considered in the NEHC risk assessment, COTsuggests that NEHC provide a complete description of how the RBSCswere determined by EPA.

In the NEHC report's explanation of how the analytical data for pollutants were convertedto a usable format for risk assessment (section 2.1, p.8), it states“If a constituent was detected in a sample (e.g., 3 µg/m³) then one-half the sample quantitation limit (e.g., 1.5 µg/m³) was substituted as the concentration for all of the non-detectedvalues in the calculations of exposure point concentrations.” Forconcentration data that are lognormally distributed, as these dataappear to be, the square root of the limit of quantitation is theappropriate value to use. This approach should be used only whenthe limit of detection and the limit of quantitation have been determined.Using a single detection measurement as a limit of quantitation isnot appropriate.

During the times of air sampling, the incinerator operations werereported to be modified from normal conditions because the operatorsof the incinerator became aware of the air monitoring. These modificationsof normal operating procedures were unfortunate because they likelyinvalidate the measurement data and might make them unusable forquantifying the incinerator complex as a source of the measured pollutants.For example, the meteorological data show no correlation betweenwind direction and level of particulate matter (see later discussion).

Compared with the sampling of two locations in 1995, for screeningpurposes, these 1997 sampling data provide a more representativecharacterization of the pollutants to which naval personnel and theirfamilies are exposed. However, the sampling data are inadequate forrisk assessment purposes.

For the purposes of the screening-level risk assessment, NEHC (1998)only considered the inhalation exposure pathway and assumed thatan individual would be present at each location 100% of the time.Exposure-point concentrations for each pollutant were calculatedfor the average exposed individual and the reasonable-maximum-exposed(RME) individual at each sampling location, and three residentialscenarios were considered: (1) adult or child exposed for

3 years, (2) adult or child exposed for 6 years, and (3) adult orchild exposed for a combined total of 30 years. A second scenariowas evaluated at one site (location 5 - ground electronic maintenancearea) for workers who were present for 3, 6, or 30 years. NEHC (1998)estimated the daily intake for each pollutant and normalized thevalues for time and body weight.

For screening purposes, consideration of only the inhalation pathwayis acceptable. However, the NEHC report does not provide an explanationfor why it was necessary to calculate an oral daily intake for eachpollutant in terms of mg/kg/d from inhalation values. As will bediscussed below in the Noncarcinogenic Risk Assessment section, suchestimates are not appropriate for evaluating inhalation exposures.

In general, the procedure used by NEHC to estimate pollutant-intakevalues overestimates exposure because it assumes that particles aredeposited equally in the respiratory tract, when in fact differentparticle sizes are retained to different degrees. Additionally, theNEHC characterization of exposure is probably not representativeof annual exposure because the air-sampling data were collected onlyduring 6 weeks of the year when the wind blows from the incinerationcomplex toward NAF Atsugi. COT believes it will be important to includeindirect exposure pathways, such as resuspension of particles andingestion (particularly important for small children that might ingestcontaminated soil particles), in future comprehensive risk assessments.

To assess the risks from the carcinogens identified in the air samples,NEHC (1998) searched two EPA databases (IRIS and HEAST) for availableinhalation unit risks. Inhalation unit risks were available for 12of the 24 contaminants considered, and those values were convertedinto carcinogenic slope factors (upper-limit carcinogenic risks).Carcinogenic risks for the individual carcinogens were calculatedby multiplying the average daily intake over a lifetime by the carcinogenicslope factor and summing the individual risks to yield the totalcarcinogenic risk. NEHC (1998) considered unacceptable any carcinogenicrisks greater than the EPA benchmarks of 1 × 10⁻⁶ for residential scenarios and 1 × 10⁻⁴ for industrial scenarios.

The method of summing individual upper-limit risks to determine anoverall carcinogenic risk of a mixture overestimates the carcinogenicrisk because not all components of a mixture are likely to be presentat their upper limits. A simple procedure for estimating an upper-limitrisk for a mixture involves taking the square root of the sum ofthe squares of all the individual risks (Gaylor and Chen 1995). Forexample, by using the values in Table 5-3 (p. 25) of the NEHC report,the upper bound on the carcinogenic risk based on average exposuresfor location 5 would be 58% ((6.3² + 14.6² +10.5² + …)^½) of the risk based on the addition of upper-bound risks that weregiven in Table 5-1A (p. 19). Similarly, the risks at locations 6through 9 in Table 5-1A should be multiplied by 0.70, 0.73, 0.67,and 0.91, respectively, to obtain appropriate upper bounds for mixtures.Hence, the carcinogenic risks for 3-year exposures at average concentrationsfor children are estimated to be less than 2.9 × 10⁻⁵ for location 6 (0.70 × 4.1 × 10⁻⁵), less than 3.1 × 10⁻⁵ for location 7 (0.73 × 4.2 × 10⁻⁵), less 2.2 × 10⁻⁵ for location 8 (0.67 × 3.3 × 10⁻⁵), and less than 1.1 × 10⁻⁴ for location 9 (0.91 × 1.2 × 10⁻⁴).

The carcinogenic risk estimates (ranging from 2.7 × 10⁻⁵ to 1.2 × 10⁻⁴) were calculated for the nine sampling locations under the assumptionthat an individual would be present at a site 100% of the time. Forexample, concentrations of hazardous pollutants were highest at thechild development center (location 9). However, a child might spend20% of a tour at the child development center (location 9), 10% inthe playground and picnic area (location 5), 65% in a residentialapartment (location 7), and 5% away from NAF Atsugi. Using that scenario,the average carcinogenic risk for a 3-year exposure calculated fromTable 5-1A would be less than 4.4 × 10⁻⁵ ((0.2 × 1.1 × 10⁻⁴) + (0.1 × 2.2 × 10⁻⁵) + (0.65 × 3.1 × 10⁻⁵)). Likewise, in the highest-risk category (adult), if 30% of theadult exposure was at the ground maintenance building (location 5),65% in a residence apartment (location 7), and 5% away from the facility,the carcinogenic risk is estimated to be less than 1.7 × 10⁻⁴.

Carcinogenic risk estimates generally are based on daily dose averagedover a lifetime. Because no person is exposed to the maximum airconcentrations throughout his or her lifetime, carcinogenic riskbased on the average concentrations provides a better estimate. Airsampling was conducted only during 6 weeks of the year when winddirections were predominantly from the incinerator toward NAF Atsugi.The average exposure over a year might be quite different. The carcinogenicrisk estimates based on the concentrations during this 6-week periodmight not be accurate because of seasonal variation.

COT suggests that calculations of upper limits of carcinogenic riskbe based on upper confidence limits of average exposures. Such estimatesof carcinogenic risk should be calculated for each pollutant at eachlocation on the basis of the duration of exposure for typical work,school, and residential scenarios. Then, an approximate overall upperlimit of carcinogenic risk is calculated from the square root ofthe sum of squares of all the upper limits of individual risks (Gaylorand Chen 1995). COT also suggests that the NEHC report indicate thatthe carcinogenic potential of dioxin has been reclassified by theInternational Agency for Research on Cancer (IARC) as a Group 1 carcinogen(IARC 1997) and that reducing the slope factor for dioxin in underconsideration at EPA. Because dioxin was one of the greatest contributorsto carcinogenic risk at NAF Atsugi, a reduction of its slope factorwill increase the overall carcinogenic risk.

To assess the noncarcinogenic risk from exposure to pollutants identifiedin the air samples, NEHC searched two EPA databases (IRIS and HEAST)for RfCs. RfCs were available for three pollutants, and those valueswere converted by NEHC into inhalation RfDs by multiplying the RfCsby 20 m³/d (assumed daily air intake) and diving the product by 70kg (assumed body weight). A fourth RfD was calculated for PM₁₀ on the basis of EPA's NAAQSs. Hazard indices (the ratio of the average lifetime doseof a particular pollutant to the RfD) were calculated for the individualpollutants and then summed to yield a total hazard index. FollowingEPA guidelines, NEHC considered a noncarcinogenic hazard index exceeding1 unacceptable.

COT does not find the method used by NEHC to evaluate noncarcinogenicrisks to be a valid approach. Specifically, COT disagrees with NEHC's decision to express RfCs in terms of “inhalation RfDs.” There is no such thing as an inhalation RfD. RfDs are derived fororal

exposures, whereas RfCs are derived for inhalation exposures. TheNEHC report does not provide an explanation for why RfCs were convertedinto mg/kg/d. If NEHC's intent was to convert the RfCs into equivalentoral doses, the necessary conversion would be much more complex thanthat used. If NEHC's intent was merely to express the RfC in termsof mg/kg/d, the conversion selected is also inappropriate as it impliesthat the RfC was established on a day-by-day basis rather than alifetime exposure. COT concludes that NEHC's converted values aremisleading and cannot be used to calculate hazard indices.

COT recommends that NEHC use the RfC values themselves as the basisfor evaluating noncarcinogenic risks. EPA defines an RfC as “an estimate(with uncertainty spanning perhaps an order of magnitude) of a continuousinhalation exposure to the human population (including sensitivesubgroups) that is likely to be without appreciable risk of deleteriousnoncancer health effects during a lifetime” (EPA 1994). COT alsorecommends that NEHC replace its description of the RfD (section4.3, pp. 15-16) with a discussion of the RfC.

Although COT recommends that the discussion of the RfD be removedfrom the NEHC report, COT alerts NEHC to several inaccuracies inthe definition and description of the RfD presented in the NEHC report(section 4.3, pp. 15-16). The NEHC report states that “RfDs are developedfor short-term exposures (i.e., subchronic) …”, when, in fact, RfDsare developed for lifetime exposure (i.e., 70 years). The correctEPA definition of an RfD is “an estimate (with uncertainty spanningperhaps an order of magnitude) of a daily exposure to the human population(including sensitive subgroups) that is likely to be without an appreciablerisk of deleterious effects during a lifetime” (Barnes and Dourson1988). The RfD is calculated by determining the most sensitive andsignificant no-observed-adverse-effect-level (NOAEL) or lowest-observed-adverse-effect-level(LOAEL) for noncarcinogenic effects and dividing by a series of uncertaintyand modifying factors. Five uncertainty factors and one modifyingfactor are used in the calculation of the RfD (Barnes and Dourson1988; Dourson 1994). However, NEHC describes only three uncertaintyfactors and one modifying factor. Missing from the NEHC's descriptionare the uncertainty factor for predicting chronic effects from subchronicstudies and the uncertainty factor that represents the expected ratioof the LOAEL to the NOAEL when a LOAEL is used instead of an NOAEL.

Neither carcinogenic nor noncarcinogenic toxicity values were availablefor nine pollutants in the two EPA databases searched by NEHC, soqualitative assessments of toxicity were conducted by using EPA RegionIII toxicity values to calculate their carcinogenic and noncarcinogenicrisks. None of the data, except those for chromium, PM₁₀, and lead, was provided in the NEHC report. Therefore, COT was unableto verify NEHC's conclusion that “… risks for [these] chemicals, except for chromiumdo not add a significant contribution to the risk.”

Two species of chromium—trivalent (Cr III) and hexavalent (Cr VI)—were sampled and analyzed. Cr VI is an EPA class A carcinogen, andCr III is not. Concentrations of Cr VI exceeded the detection limitin only one sample, so risks for Cr VI were not calculated. The 1998NEHC report states, “In the presence of oxidizable organic matter,Cr VI is quickly reduced to Cr

III” (section 5.4, p. 26). COT does not agree with that statement, becauseCr VI will likely be contained in aerosols, probably in the solidphase. The contact at the molecular level between oxidizable organiccompounds in the air and Cr VI inside solid particles seems likelyto be insignificant in the travel time between emission from theincinerator plant and exposure at NAF Atsugi. If relevant studiessupport the NEHC statement, a reference to those studies should beprovided.

NEHC used EPA's current annual NAAQS (50 µg/m³) to calculate the noncarcinogenic risk for PM₁₀ (section 5.4, p. 26). The NAAQS value was expressed in terms ofmg/kg/d by using standard exposure assumptions (i. e., an averagebody weight of 70 kg and a daily air intake of 20 m³). Using thisvalue, PM₁₀ was included in the NEHC's noncarcinogenic risk evaluation (see earlier discussion).

COT disagrees with NEHC's approach for evaluating PM₁₀. There are two NAAQS values for PM₁₀—an annual average (50 µg/m³) and a 24-hour average (150 µg/m³) (EPA 1998a). NEHC's conversion of the annual average to mg/kg/d is not valid. The NAAQSannual average for PM₁₀ already accounts for body weight and expressing the value in termsof “per day” compounds the problem, especially since the NAAQS risk assessmentprocess already differentiates between daily and annual exposure.The conversion used by NEHC is incorrect and obviates the scientificbasis of the NAAQS. COT suggests that a better approach for qualitativelyevaluating PM₁₀ is to identify how many days during the sampling period the NAAQS24-hour average (150 µg/m³) was exceeded. The NAAQS annual mean for PM₁₀ can be used in future assessments when year-round monitoring dataare available. COT also suggests that the statement “Particles largerthan 10 microns are not respirable (i.e., they are too large to reachthe lung)” (Table 4-1, p. 17 and section 5.4, p.26) be removed fromthe NEHC report because this is not a true statement.

COT notes that the only pollutants considered qualitatively in theNEHC report were those that had neither carcinogenic nor noncarcinogenictoxicity values. However, there was only one pollutant (acetaldehyde)for which both types of toxicity values were available. COT believesthat qualitative assessments should be conducted for all of the otherpollutants considered in the risk assessment. For example, dioxinwas considered only in the carcinogenic risk assessment even thoughnoncarcinogenic effects have been associated with this pollutant.The NEHC's Qualitative Assessment of Chemicals Without Toxicity Values(section 5.4, p. 26) should identify those noncarcinogenic effects,including recent findings of effects on the endocrine and immunesystems of laboratory animals. In addition, the carcinogenic classificationby IARC (1997) could be identified for pollutants that were not consideredin the carcinogenic risk estimates. The classification would indicatewhether carcinogenic risks are underestimated by not consideringthose pollutants in the risk evaluation.

NEHC evaluated the meteorological data collected on each day thatair sampling was performed to determine if the measured ambient pollutantconcentrations correlated with wind direction. The hourly wind speedand direction results were evaluated by using EPA's Wind Rose

Program (EPA 1998b). PM₁₀ was monitored because it was believed to be representative of overallparticulate emissions responsible for the majority of the hazardindices.

COT agrees with NEHC's conclusion that the meteorological data show “… very little correlation between wind direction and concentrationsof COCs [constituents of concern] in ambient air” (section 5.5.2,p. 33). However, COT disagrees with the NEHC statement that “Locations7 and 8 had fairly high positive correlations between wind directionand PM₁₀ concentrations.” Figure 5-10 (p. 34) does not justify that statement. What is neededis a measurement of particle sizes and particle-size-related composition;such a measurement would give a better idea about the origin andsources of the airborne pollutants. As it stands now, Figure 5-10(p. 34) clearly shows a high background concentration of PM₁₀ at location 1 (upwind site) independent of the wind direction. Location4 does not show a correlation with wind direction; yet, it was thesite with the highest pollutant concentrations. Likewise, location5, which, according to air-dispersion models, is the site of “highestimpact,” does not show good correlation between wind direction andPM₁₀ concentrations, although it was the site sampled for the secondhighest number of days (18 days). Thus, it could be argued that thesedata show that the incinerator did not significantly contribute toconcentrations of PM₁₀ at NAF Atsugi and that other sources were responsible for the observedpollution.

The argument presented above is based on the assumption that validcorrelations can be drawn from the meteorological data presented.However, the committee believes that the correlation data presentedare not valid for several reasons:

1. The full impact of meteorological variation on receptor exposureto the plume is not represented solely by wind direction.

2. The sampling results are not representative of normal exposuresto the incinerator emissions because incinerator operations werealtered during the sampling period.

3. The 24-hour samples probably represent a mixture of wind directions.

Although some of those reasons were identified in the report as OtherFactors that May Affect Ambient Concentrations (section 5.5.3, p.33-34), the data presented in Figure 5-10 (p. 34) and Table 5-5 (p.35) give the impression that valid correlations can be drawn fromthe data. In addition, explanations for uncertainty about the dataare unclear. For example, Table 6-1 (p. 36-37) states “Insufficientmeteorological data resulted in lack of correlation between winddirection toward sampling site and concentration of chemicals.” Thatstatement implies that poor correlation resulted from lack of data,when, in fact, the correlation analysis is not valid because of thereasons described above. The lack of correlations could be due, inpart, to multiple sources of air pollutants at different directionsfrom NAF Atsgui. The table also states in the “Actions and Results” column, “The impact of wind direction on ambient air COC concentration couldnot be evaluated.” Such statements present a confusing picture andare not in agreement with the conclusion regarding correlation.

Another approach that might be useful for getting a rough estimateof the contribution of incinerator emissions to ambient air, relativeto background, would be to compare results from

location 1 (upwind site) with those from downwind locations on dayswhen the wind direction is out of the south-southwest and relativelyconstant. Additionally, a description of other potential air-pollutionsources in the vicinity of NAF Atsugi should be identified.

The 1998 NEHC report concluded that the estimated carcinogenic andnoncarcinogenic risks to naval personnel and their families at NAFAtsugi were similar to risk levels calculated in 1995 (NEHC 1995).COT agrees that for screening purposes the conclusion regarding carcinogenicrisks is valid. The estimated carcinogenic risk for nearly everyexposure scenario were greater than the EPA acceptable range of 1× 10⁻⁴ to 1 × 10⁻⁶. Dioxins and furans were responsible for the majority of the estimatedcarcinogenic risk; and arsenic, cadmium, and benzene were also importantcontributors.

The air sampling data indicate that PM₁₀ concentrations at NAF Atsugi are above levels known to cause adversehealth effects. However, NEHC's noncarcinogenic risk estimates couldnot be verified by COT, because the values used to calculate thehazard indices were inappropriately converted from RfCs to RfDs.

The risks attributable to the chemicals considered in the 1998 NEHCreport are different from those found in the 1995 NEHC preliminaryrisk evaluation. However, given the potential carcinogenic and noncarcinogenicrisks to naval personnel and their families at NAF Atsugi, COT agreesthat the recommendations made in the 1995 report to reduce potentialexposures and to conduct a comprehensive human health risk assessmentare still valid.

An implicit conclusion of the 1998 NEHC report is that the Jinkanpoincineration complex is the major source of the measured pollutantsat NAF Atsugi. However, as discussed earlier, the background concentrationof pollutants appears to be high, which suggests that sources otherthan the incineration complex are also contributing to the air pollutionat Atsugi.

The only susceptible subpopulation considered in the NEHC (1998)report was children. For screening purposes, COT believes that approachwas appropriate. Overall, the highest risks calculated were at theChild Development Center (location 9) for every exposure scenario.The carcinogenic risk (1.2 × 10⁻⁴ for average exposure and 5.2 × 10⁻⁴ for RME) was above the EPA range of 1 × 10⁻⁴ to 1 × 10⁻⁶ for children living at NAF Atsugi for 3 years.

The uncertainties identified in the 1998 NEHC report (pp. 36-37)were the following:

• Data Evaluation: (1) Air samples were not analyzed for pollutants such as polycyclicaromatic hydrocarbons or hexachlorobenzene; (2) data-quality objectiveswere established for compliance rather than risk-assessment purposes;(3) insufficient meteorological data resulted in a lack of correlationbetween wind direction and concentration of pollutants at NAF Atsugi;and (4) precipitation data were not available for each day that airsampling was performed.

• Exposure Assessment: Major assumptions about exposure scenarios, inhalation rates, andbody weight were used.

• Toxicity Assessment: (1) Not all of the pollutants were included in the risk evaluationbecause of the lack of EPA-approved toxicity values and (2) the availabletoxicity values were derived by using safety factors and upper-boundestimates.

• Calculation of Risks: The report assumed that the adverse health effects of pollutantswere additive in the risk calculations. Antagonistic and synergisticeffects of pollutant mixtures were not evaluated.

• Site-Specific Uncertainties: The report did not evaluate exposures other than the inhalationroute.

COT does not believe that the uncertainty analysis presented in theNEHC report adequately addresses the full range of uncertaintiesinvolved in the risk assessment. COT suggests that NEHC reiteratein section 6.0 (p. 36) that the NEHC report is a screening levelassessment to determine the need for further study, and, therefore,more conservative assumptions were used and a larger range of uncertaintieswere assumed. COT recommends that the following uncertainties alsobe considered:

1. The validity of the hazard indices used to evaluate noncarcinogenicrisks is questionable. The conversion of RfC and annual average NAAQSvalues into RfDs is not appropriate for calculating hazard indicesfor inhaled pollutants. Furthermore, NEHC used RfC values based onchronic exposure to evaluate short-term exposure scenarios.

2. The uncertainty of using some risk values that did not pass throughan adequate review or verification process. The NEHC report statesthat risk values listed in EPA's IRIS and HEAST databases have beenpeer reviewed (section 4.1, p. 14); however, subsequent descriptionsof the databases indicate that some of the values listed in HEASTare “provisional” because they have not been verified by an agency work group.Statements that the risk values used in the risk assessment are EPAapproved values (Table 6-1, Toxicity Assessment, p. 37) should beremoved from the NEHC report.

3. The uncertainty involved with adding together risks based on acuteand chronic toxicity end points and for different target populations,including the difficulty of

adding together the risk of acute mortality in a person with cardiopulmonarydisease with the risk of chronic liver toxicity in a healthy adult.

COT also found serious limitations in the methods used by NEHC tocollect an analyze the air samples. These include:

1. QA-QC measures were inadequate to provide accurate measurementsof pollutant concentrations.

2. NEHC made an unsupported assumption that its selected backgroundsite (location 1) was substantially affected by point-source andfugitive emissions from the incinerator complex. No explanation ordata were provided in the NEHC report to indicate why that locationwas believed to be substantially affected by incinerator emissions.The meteorological data seem to indicate that location 1 might givea reasonable estimate of background concentrations of pollutants,at least as measured by PM₁₀; these data indicate that the background concentration of PM₁₀ at Atsugi is high.

3. Incinerator operations were reportedly modified at times duringthe sampling periods as the operators of the incinerator became awareof the air monitoring. Such modifications of normal operating procedureslargely invalidate the measurement data and make them unusable forcharacterizing the incinerator as a source of measured pollutants.

A point of clarification should also be made with regard to the insufficientmeteorological data. Table 6-1 of the report states that insufficientmeteorological data resulted in a lack of correlation between winddirection and concentration of pollutants in the air. The statementimplies that poor correlation resulted from lack of data, when, infact, the correlation analysis is not valid for a variety of reasons(identified earlier under Impact of Wind Direction on Risk) thatcould also have affected ambient air concentrations.

On the basis of its evaluation of the 1998 NEHC report, COT makesthe following recommendations:

• The NEHC noncarcinogenic hazard indices and risk assessment shouldbe revised to allow comparison of air pollutant concentrations toinhalation RfCs. It is inappropriate to convert RfCs and NAAQS valuesinto RfDs for inhaled pollutants.

• The following uncertainties should also be considered in the NEHCrisk assessment:

  • The validity of the hazard indices used by NEHC to evaluate noncarcinogenicrisks is questionable. NEHC's conversion of RfC and

annual average NAAQS values into RfDs is not appropriate for calculatinghazard indices for inhaled pollutants. Furthermore, NEHC inappropriatelyused RfC values based on chronic exposure to evaluate short-termexposure scenarios.

• Some risk values were not passed through an adequate review or verificationprocess. The NEHC report states that risk values listed in EPA'sIRIS and HEAST databases have been peer reviewed (section 4.1, p.14); however, subsequent descriptions of the databases indicate thatsome of the values listed in HEAST are “provisional” because they have not beenverified by an agency work group.

• The NEHC report inappropriately added risks based on acute and chronictoxicity end points and for different target populations, includingaddition of the risk of acute mortality in a person with cardiopulmonarydisease with the risk of chronic liver toxicity in a healthy adult.

COT also evaluated the recommendations presented in the 1998 NEHCreport. Below, those recommendations are presented and are followedby COT's evaluation and further recommendations.

1. Continue to update the Government of Japan of the potential threatto the health of on-and off-base NAF Atsugi residents, includingthe local Japanese community to enhance the Japanese Government awarenessof the health risks caused by poor air quality at NAF Atsugi.

2. Perform year-round air monitoring at NAF Atsugi outdoor and indoorreceptor sites to reduce uncertainties in the human health risk assessment.

3. Conduct soil trend analysis of dioxin deposition, analyze indoorand outdoor dust samples, as well as investigate the food pathwayto characterize the risk from dioxin exposure.

4. Conduct soil sampling at receptor sites to characterize from soilpathway exposure.

5. Continue to educate day care workers about prevention of exposureand recognition of pica behavior (consumption of non-food items,such as soil) in children to prevent exposures that may result fromcontact with surface soil.

6. Continue to collect medical data from Branch Clinic Atsugi and othercomparable medical treatment facilities concerning respiratory healtheffects that may be attributed to the poor air quality, to determinerates of illness incidence.

7. Continue to educate healthcare providers assigned to Branch ClinicAtsugi in the prompt recognition and treatment of medical conditions,which might be caused/aggravated, by the poor air quality at NAFAtsugi.

8. Continue implementation of a risk communication program to informNAF Atsugi residents of the efforts being conducted to characterizethe exposure to the air quality at NAF Atsugi, in an understandableand effective manner.

9. Continue to advise residents about curtailing outside activities,particularly strenuous physical exertion, during periods when particulatelevels are elevated This recommendation is important for residentswith known cardiovascular or respiratory conditions.

On the basis of the data presented in the 1998 NEHC report, COT agreesthat the recommendations listed above are justified. COT also agreesthat some degree of risk of adverse health effects might exist atNAF Atsugi. The following are COT's specific comments on NEHC recommendationsand additional recommendations from COT:

• NEHC Recommendation 2: A full-year air-monitoring study is warranted, based on the screeningresults. COT recommends that source emissions testing of the incineratorstacks be conducted, if possible, in addition to air monitoring.Source emissions testing would provide more accurate data on thetypes and concentrations of pollutants being emitted by the incinerator.However, if an air-monitoring study is the only viable testing methodavailable to the Navy, careful design of the study is essential.COT recommends that the following:

  1. The air-monitoring study should be designed not only to estimateexposure to the target population, but also to distinguish the impactof incinerator emissions from background air quality.

  2. Nearby sources of air pollutants, other than the incinerator complex,should be characterized, and their contributions to total risk evaluated.

  3. Ground-level dispersion of pollutants from solid waste piles andliquid waste sources should be evaluated.

  4. Adequate QA-QC measures should be taken during air sampling andanalysis to provide accurate measurements of pollutant concentrations.

  5. More extensive meteorological data related to local atmosphericstability should be collected and used in dispersion modeling.

6. Continuous or semicontinuous monitoring data should be collectedto enhance ability to correlate meteorological data and emissions-dispersionestimates with ambient concentrations of pollutants.

7. PM_2.5 should be monitored, as numerous studies in the last several yearshave implicated fine particles as a threat to human health. Ozone,nitrogen dioxide, carbon monoxide, and sulfur dioxide should alsobe monitored.

• NEHC Recommendation 3: Because dioxin analyses are expensive, deposition of dioxin in soiland oral exposure to contaminated soil or food products should bedemonstrated before indepth analyses of the food pathway are conducted.If the soil trend analyses do not show significant dioxin concentrationsor if local produce is not consumed in significant quantities, thefood pathway is unlikely to be a significant source of exposure.

• NEHC Recommendation 4: In addition to soil sampling, NEHC should also consider samplingground water and drinking water to determine potential exposure viathose pathways. As noted in COT's 1995 evaluation, “if the ground water is contaminatedby VOCs [volatile organic compounds] and travels under occupied Navybuildings, VOCs could reach high levels in indoor air by diffusionthrough soil into basements, …” (NRC 1995).”

• NEHC Recommendation 5: Most young children put their hands in their mouths frequently,even when pica behavior is not obvious, thereby ingesting quantitiesof soil that can significantly increase exposure to soil-borne contaminants.This source of exposure should be considered in the assessment ofchildren's risk. If deposition of dioxin or other pollutants in soilis found to be significant, contaminated areas should be avoidedor efforts should be made to clean-up the soil, in addition to educatingdaycare workers about preventive measures to reduce children's exposureto contaminated soil.

• NEHC Recommendation 9: No explanation was provided on how residents at NAF Atsugi wouldknow when particulate concentrations were increased so that theycan curtail outdoor activities. Current real-time monitoring at thebase would provide such information.

Executive Summary, paragraph 1: Noncarcinogenic risk levels appear to have been confused with hazardindex. The statement that the 1995 risk assessment indicated thatair quality at NAF Atsugi could raise the “non-cancer risk to levelshigher than 1”, should be changed to state “increase non-cancer hazard indices to greater than1.”

Section 1.3, paragraph 1 (p. 2): Clarification of the following statement is needed “…emissions from the incinerator complex are carried parallel to thestack height downwind towards the base.” This seems to imply downwashdue to eddies on the downwind side of the stack. Also, base couldmean NAF Atsugi or the base of the stack.

Section 2.4 (p. 9): There is an error in the description of how exposure point concentrationwas calculated for the average exposed individual: “The maximum detectedconcentration was used if the arithmetic mean concentration exceededthe maximum detected concentration.” It is not possible for the meanto exceed the maximum; this explanation should be corrected.

Section 3.0, paragraph 2 (p. 11): It is stated that this section describes the “current and future land uses.” Aside from the site locations which are actually provided in Section2.1, there are no descriptions of these.

Section 3.2 (p. 13): It is not clear what is meant by “contact rate”, which is listed as one of six factors used to estimate pollutantintake.

Table 3-2 (p. 13): In the equation for daily intake, the subscripts “a” and “c” (presumably for adult and child) have not been defined. The terms“At_c and “At_nc,” were defined but not used in the equations. Also, “At_c” could be confused with the “AT_c” in the equation for integrated inhalation dose.

Section 4.1, paragraph 4 (p. 14): This paragraph is unnecessary.

Section 5.0, paragraph 2 (p. 16): The term “hazard index” is used, but is not defined until section 5.1. Also, it is statedthat EPA has used 1 chance in 10,000 as a benchmark for carcinogenicrisk in industrial scenarios. Was OSHA intended here instead of EPA?

Section 5.1, paragraph 1 (p. 16): Reference dose is defined here, but was already defined in section4.3.

Section 5.1, last sentence of paragraph 1 (p.16): This sentence does not make sense. It might be rewritten as follows:“Only the hazard quotients for chronic disease were evaluated, asthe hazard quotients for subchronic effects within a given exposurescenario are typically less than or equal to the hazard quotientsfor chronic effects for the same scenario.”

Table 5-1A (p. 19): For comparative purposes, it would be useful to provide footnotesexplaining what “hazard index” and “cancer risk” mean and what levels are unacceptable.

Figures 5-1, 5-2, 5-3, and 5-4 (pp. 21-24): Replace the black boxes below the x-axis with the location numberswhere the measurements were performed.

Section 5.4, paragraph 4 (p. 26): Replace 91 mg/m³ with 91 µg/m³; on the next line, replace “hazardous” with “adverse”.

Section 5.5.2, paragraph 2 (p. 33): The “lack of wind” should be stated differently if what is meant is that the wind seldomblew from the incinerator toward location 2.

Section 5.5.3, “Other Meteorological Factors”, (p.36): The collision of cold and warm air masses (producing advectiveinversions or occluded fronts) is only one of several reasons forthe formation of atmospheric temperature inversions. The other twodistinct reasons for inversions

are radiational cooling of the earth's surface at night, which cools the lower layers of the atmosphere,and the subsidence associated with high pressure systems.

Table 6-1 (pp. 36-37): Insert “estimates” after “risk” in the footnotes, since it the risk estimate that is or is not conservative.

Section 7.1, p. 37: It is stated that air samples were collected on 23 different daysand that location 1 was sampled on 27 days. If these numbers arecorrect, an explanation should be provided for why samples were takenmore often at location 1. Also, sampling was described as being collectedon different days from 7/27/97 through 9/1/97, which is differentfrom that mentioned in section 1.1 (p. 1) of July 27 to September11, 1997.