Review of the US Navy's Human Health Risk Assessment of the Naval Air Facility at Atsugi, Japan (2001)

The risk-assessment approach used by NEHC stems from EPA's Superfund methodology. Although such a risk assessment is appropriate under some circumstances, NEHC's interpretation of the results is not appropriate, because it does not take into account the limitations of such a risk assessment. In the Superfund approach, estimated RMEs are compared with RfDs. RfDs are conservative estimates of dose rates that are considered safe for a given chemical. Uncertainty and modifying factors are incorporated into them to ensure the protection of public health. They are calculated on the basis of chronic exposure (much longer than the standard tour of duty at NAF Atsugi), and they often focus on chronic, irreversible effects. Therefore, if the RME of a chemical at NAF Atsugi is below the RfD, it is appropriate to conclude that the exposure is not likely to cause adverse health effects. However, if the RME or even an average estimate of exposure to a chemical at NAF Atsugi is above the RfD, it cannot necessarily be concluded with confidence that the exposure is causing adverse health effects. Therefore, comparison of RMEs with the RfD can be used to screen out chemicals that do not pose health risks or to show that simultaneous exposure to a large number of chemicals is safe, but cannot be used to determine the actual health effects in US Navy personnel, their families, and other persons living or working at NAF Atsugi.

Careful consideration should be given to the characteristics of the exposures that are being estimated and to the derivation of the toxicity values with which the exposure estimates are compared for both the risk assessment itself and the communication of those risks. Estimates of exposure should be presented with ranges or confidence intervals. Appendix A of the NEHC draft summary report contains a comparison of the risk-assessment results with the levels at which acute health effects were seen, according to the results of a literature search “to determine potential acute health effects for the specific 24-hour concentrations measured at NAF, Atsugi” (p. 65). The comparison might seem to satisfy the need for an evaluation of acute, possibly reversible effects, but the concentrations measured at NAF Atsugi are compared with toxicity values that were compiled for various purposes and with various protocols. The minimal risk levels (MRLs) and reference exposure levels (RELs), for example, are similar to EPA's RfCs and RfDs, whereas others appear to be more similar to lowest-observed-effect levels. A discussion of how the assumptions and adjustment factors used in the derivation of toxicity values affect the risk assessment should be included in this type of comparison.

Some of the California RELs presented in Appendix A are different from the May 2000 values posted at http://www.oehha.org/air/acute_rels/allAcRELs.html. Table 1-1 presents two examples of acute RELs.

Furthermore, those RELs are based on 1 or 6 h of exposure and are not appropriate for direct comparison with the “specific 24-hour concentrations measured at NAF Atsugi.” Similarly, the comparison with intermediate MRLs is not valid, because a 1-day exposure at or near the intermediate MRL is not necessarily

a cause of concern. In addition, some of the sampling at NAF Atsugi was not for 24 h, and it is not clear whether the specific entries in Appendix A were limited to sampling for 24 h. Because of those concerns with the comparisons in Appendix A, the subcommittee recommends that Appendix A be substantially revised or removed from the NEHC document.

The exposure conditions at NAF Atsugi should be kept in mind when one is considering the toxicity of the chemicals present. For example, NEHC discusses the potential health effects of cadmium, but cadmium is a slowly accumulating toxicant that causes adverse effects only after chronic exposure (Goyer 1996) and is unlikely to cause medically significant health effects after 3-6 yr of exposure at the concentration measured at NAF Atsugi.

A more appropriate comparison would be between the concentrations of pollutants at NAF Atsugi and the point of departure (the level at which effects are seen, before the addition of uncertainty and modifying factors, in a critical study on which a toxicity value is based). Such a comparison would be similar to the margin-of-exposure analyses recommended by the Presidential/Congressional Commission (1997b). Concentrations of pollutants measured at NAF Atsugi higher than the point of departure would be a cause of concern. The health effects are not clear for concentrations at Atsugi above the RfD but below that point of departure; what action to take in response to such an exposure is the subject of a policy decision, and this should be clearly stated.

As discussed by the Presidential/Congressional Commission (1997b), using risk estimate values as bright lines can be useful for setting goals for decision-making. However, because of uncertainty inherent in risk estimates, they should not be used as rigid, exact numbers. NEHC's interpretation and discussion of risk estimates as though they are bright lines are therefore not appropriate. For example, NEHC translates noncancer and cancer risk estimates into threshold numbers of months of exposure after which a 10⁻⁴ risk is reached, such as 32 months for children younger than 6 years old and 98 months for adults (NEHC 2000; p.79). In light of the uncertainty in the risk values, such absolute statements are not justified, and the interpretation fails to recognize and convey the continuous and stochastic nature of risk.

NEHC should clarify the nature and history of the regulatory use of 10⁻⁶ and 10⁻⁴ cancer risk estimates. The practice of setting 10⁻⁶ as a point of departure for cancer risk is a regulatory decision. Similarly, 10⁻⁴ is not a “cancer risk benchmark”, but a value rooted in regulatory decision-making as opposed to medical practice. Travis et al. (1987) demonstrated, on the basis of an analysis of published risk-based decisions by US government agencies, that regulatory levels could be explained to some extent by the numbers of individuals in the potentially exposed population. Specifically, as the theoretically exposed population increased the regulatory threshold for taking some action became increasingly more stringent regardless of the chemical of concern. The calculated estimates, if applied literally at NAF Atsugi, would mean that 10,000 children younger than 6 would have to be exposed to concentrations at the upper-bound of the modeled estimates to produce between 0 and 1.1 additional cancers in those children's lifetimes. NEHC should consult various National Research Council publications (NRC 1983, 1994, 1996) and the Presidential/ Congressional Commission's report (1997a,b) for the regulatory history of the cancer risk values and guidance on their use in decision-making.

The subcommittee notes that many decisions on setting benchmarks, determining acceptable risks, and weighing costs and benefits in risk management are important policy decisions. Such policy decisions should be made with stakeholder participation. Although it is beyond the scope of this subcommittee's task to recommend policy decisions, it does recommend that NEHC clearly differentiate in its report which decisions are based on science and which on policy.

A major problem in the interpretation and presentation of the risk assessment is the inadequacy of the discussion of uncertainties. The importance of an adequate characterization and discussion of uncertainties in a risk assessment has been noted in many publications (NRC 1994, 1996; Presidential Commission 1997a,b). The NEHC draft summary report mentions uncertainty only in saying that the risk assessment was developed with the “minimum degree of uncertainty”; it fails to disclose the types or magnitudes of any source of uncertainty and fails to discuss the impact of uncertainty in the context of the risk-assessment results. Readers are left with the impression that the risk-assessment results are certain. Given the limitations of risk analysis in general (for example, uncertainties in estimating exposure and extrapolation from animal toxicity data at high doses to human health effects at lower doses), that interpretation is inconsistent with the state of the science.

The report should indicate the amount of uncertainty in the results, at least qualitatively, and should clarify what is meant by the “minimum degree of uncertainty.” An adequate discussion of uncertainty should include the sources of uncertainty mentioned in Appendix C of this draft summary report.

The previous COT report (NRC 1995) made a similar comment on the need to convey the uncertainty in risk assessment:

The Navy should be urged to be far more careful in its presentation of risk results. Statements such as that on page 13 of the NEHC (1995) report (“The total cancer risk for carcinogens of 6.62 x 10⁻³ means that there is an increased risk of 7 cancer cases per 1,000 people over the normal lifetime cancer risk rate”) can be misleading. Both cancer and noncancer risks estimated using the methods that NEHC employed are not as certain as such statements imply. To be credible, all risk estimates should be accompanied by descriptions of the assumptions and uncertainties that are associated with them.

The present subcommittee reiterates that thought and finds statements like the following (p. 2 of Public Health Summary, NEHC 2000) regarding the risk-assessment results potentially misleading:

• The calculated cancer risk for the child resident, less than 6 years of age, indicates that the air quality at NAF Atsugi could result in as much as 1.1 additional cancer cases in a population of 10,000 after an exposure period of approximately 3 years.

• The cancer risk for the adult resident is calculated at 3.7 additional cancer cases in a population of 100,000 after an exposure period of 3 years.

Those statements imply that after 3 years of exposure cancer could occur (the correct statement requires pointing out that this is the lifetime risk of cancer) and does not mention the substantial uncertainty in the estimates (the correct statement requires pointing out that this is an upper-bound risk estimate, with a lower bound of zero). They also fail to discuss appropriately that the increase is small relative to lifetime cancer risk.

The NEHC draft summary report does not discuss some of the sources of uncertainty identified in the Draft Human Health Risk Assessment by Pioneer Technologies Corporation (Pioneer 2000). Furthermore, some uncertainties mentioned elsewhere in the text do not appear in the section on uncertainties.

NEHC should consider conducting sensitivity analyses to characterize the uncertainty in the risk assessment as recommended by the National Research Council (1994, 1996) and the Presidential/-Congressional Commission (1997b). Appendix C presents specific details of where uncertainty is inadequately addressed.